From alexwade at gmail.com Thu Aug 2 15:41:22 2007 From: alexwade at gmail.com (Alex Wade) Date: Thu, 2 Aug 2007 12:41:22 -0700 Subject: Connectionists: Cosyne 2008: Call for workshop proposals Message-ID: <76eaaa9a0708021241h88eaba8p8b6349dcba7f5a86@mail.gmail.com> ----------------------------------------------------------------------------------------- Cosyne08 - CALL FOR WORKSHOP PROPOSALS March 3-4, 2008 Snow Bird, Utah http://cosyne.org/wiki/Cosyne_08_workshop_submissions ----------------------------------------------------------------------------------------- PROPOSAL DEADLINE: Sept. 15th, 2007 A series of workshops will be held after the main Cosyne meeting (http://cosyne.org/). The goal is to provide an informal forum for the discussion of important research questions and challenges. Controversial issues, open problems, comparisons of competing approaches, and alternative viewpoints are encouraged. Workshop topics include, but are not limited to, the following: neurobiology and computational models of visual and auditory processing of complex stimuli, time perception, olfactory computation, multisensory integration, memory, Bayesian inference, decision making, active sensation, motor control; principles of unsupervised learning, reinforcement learning; neural coding; computation with spikes, with network dynamics/adaptation mechanisms on multiple time scales; dendritic processing; reward and neuromodulation; microcircuitry of the cortical column; computational anatomy; multi-scale brain modeling; foraging; neurogenesis; ________________________________________________________________ WORKSHOP DETAILS: -- There will be 4-8 workshops/day, running in parallel. -- Each workshop is expected to draw between 15 and 80 people. -- The workshops will be split into morning (7:30-10:30 AM) and afternoon (4:30-7:30 PM) sessions. -- Snow Bird is a ski resort, located 30 miles (typically less than an hour) from the Salt Lake City airport. -- Buses from the main conference will be provided. -- Up-to-date information and descriptions of previous workshops may be found at http://cosyne.org/wiki/Cosyne_08_workshop_submissions ________________________________________________________________ SUBMISSION INSTRUCTIONS: Deadline: September 15th, 2007 Format: plain text only -- please no attachments email to: fsommer at berkeley.edu Proposals should include: -- Name(s) and email address(es) of the the organizers. A primary contact should be designated. -- A title. -- A description of: what the workshop is to address and accomplish, why the topic is of interest, who the targeted group of participants is. -- Names of potential invitees. Preference will be given to workshops with the most confirmed speakers. -- Proposed workshop length (1 or 2 days). Most workshops will be limited to a single day. If you think your workshop needs 2 days, please explain why. -- A *brief* resume of the workshop organizer along with a *brief* list of publications (about half a page total). ________________________________________________________________ WORKSHOP ORGANIZERS RESPONSIBILITIES: -- Coordinate workshop participation and content. -- Moderate the discussion. ________________________________________________________________ SUGGESTIONS: Experience has shown that the best discussions during a workshop are those that arise spontaneously. A good way to foster these is to have short talks and long question periods (e.g. 30 + 15 minutes), and have plenty of breaks. Also, when it comes to the number of talks, in the words of Jerry Brown, less is more. We recommend fewer than 10 talks. ________________________________________________________________ WORKSHOP COSTS: Detailed registration costs, etc, will be available at http://cosyne.org/ Please note: Cosyne does NOT provide travel funding for workshop speakers. All workshop participants are expected to pay for workshop registration fees. Participants are encouraged to register early, in order to qualify for discounted registration rates. Cosyne does provide free workshop registration for workshop organizers. ________________________________________________________________ COSYNE 2008 WORKSHOP CHAIR: Fritz Sommer, Jascha Sohl-Dickstein (UC Berkeley) COSYNE 2008 EXECUTIVE COMMITEE: Tony Zador (CSHL) Alex Pouget (U Rochester) Zach Mainen (CSHL) Eero Simconelli (NYU) Matteo Carandini (Smith Kettlewell) ________________________________________________________________ QUESTIONS: email fsommer at berkeley.edu From announce at ccnconference.org Thu Aug 2 16:26:45 2007 From: announce at ccnconference.org (announce@ccnconference.org) Date: Thu, 2 Aug 2007 14:26:45 -0600 Subject: Connectionists: CCNC 2007/Dynamical Neuroscience XV --- ONLINE REGISTRATION NOW AVAILABLE! Message-ID: <200708021426.45774.announce@ccnconference.org> IMPORTANT UPDATE: * Online registration is now available via the conference website: www.ccnconference.org/page5.html PLEASE TRY TO REGISTER AS EARLY AS POSSIBLE AS THIS HELPS THE ORGANIZING COMMITTEE IN OUR PLANNING! OTHER UPDATES: * The new deadline for abstract submission is Friday August 17, 2007 * Our meeting has been officially assigned to the San Diego Convention Center by SfN * The symposium "Computational Models in Biological Psychiatry", moderated by Michael Frank, replaces the previously publicized one on sequence learning ------------------------------------------------------------------------------- ~ Call-for-Abstracts ~ 3RD ANNUAL CONFERENCE ON COMPUTATIONAL COGNITIVE NEUROSCIENCE www.ccnconference.org To be held in conjunction with Dynamical Neuroscience XV immediately prior to the 2007 SOCIETY FOR NEUROSCIENCE (SfN) meeting, November 3-7, 2007 at the San Diego Convention Center, San Diego, CA. * CONFERENCE DATES: Thu-Fri November 1 & 2, 2007 * REGISTER AT: www.ccnconference.org/page5.html The inaugural CCNC 2005 meeting held prior to Society for Neuroscience (SfN) in Washington, DC (also in conjunction with the Dynamical Neuroscience satellite) was a great success, with approximately 250 attendees, 60 presented posters, and strongly positive reviews. For 2006, we went to Houston for the much smaller Psychonomics meeting and still had over 100 attendees and almost 50 posters. In future years, we will continue to rotate among different neuroscience and psychology meetings. ____________________________________________________________________________ * DEADLINE FOR SUBMISSION OF ABSTRACTS: Friday, August 17, 2007 (NEW DATE!) Abstracts are to be submitted online via the website: www.ccnconference.org/page6.html As in past years, there will be two categories of submissions: -Poster only -Poster, plus short talk (15 min) to highlight the poster Abstracts should be limited to 250 words. Women and underrepresented minorities are especially encouraged to apply. Reviewing of posters will be inclusive and only to ensure appropriateness to the meeting. Short talks will be selected on the basis of research quality, relevance to conference theme, and expected accessibility in a talk format. Abstracts not selected for short talks will still be accepted as posters as long as they meet appropriateness criteria. * NOTIFICATION OF POSTER ACCEPTANCE: Approx. September 5, 2007 * CONTRIBUTED SHORT TALK SELECTION: Approx. September 15, 2007 __________________________________________________________________________ Program: * 2007 Keynote Speakers: Alex Pouget, University of Rochester Read Montague, Baylor College of Medicine * 3 Symposia, each including a mixture of modelers and non-modelers and focused on a common theme or issue: ** Use of computational and cognitive models in functional brain-imaging Moderator: Todd Braver, Washington University - St. Louis ** Computational models in biological psychiatry Moderator: Michael Frank, University of Arizona ** Hippocampal neurogenesis in learning and memory Moderator: Janet Wiles, University of Queensland * Approximately 12 short talks will be chosen featuring selected posters * Poster sessions ____________________________________________________________________________ 2007 Planning Committee: Suzanna Becker, McMaster University Jonathan Cohen, Princeton University Nathaniel Daw, New York University David Noelle, University of California, Merced Maximilian Riesenhuber, Georgetown University Medical Center Randall O'Reilly, University of Colorado, Boulder (ex officio) Executive Staff: Thomas Hazy, University of Colorado, Boulder For more information and to sign up for the mailing list visit: www.ccnconference.org _______________________________________________ From ralf.steinberger at jrc.it Mon Aug 6 10:17:53 2007 From: ralf.steinberger at jrc.it (Ralf Steinberger) Date: Mon, 06 Aug 2007 16:17:53 +0200 Subject: Connectionists: JRC-Acquis: bilingual alignments for 231 language pairs now available Message-ID: <00fd01c7d834$942043f0$d547bf8b@IPSC.TLD> Bilingual alignments for all 231 language pairs of the JRC-Acquis parallel corpus are now freely available online. We are pleased to announce that the bilingual alignments for all 231 language pairs of the JRC-Acquis corpus are now available online for download. The JRC-Acquis is a freely downloadable multilingual parallel corpus in 22 languages comprising of a total of over 1 Billion words. SIZE AND FORMAT - 22 languages (all official EU languages except Irish) - Average corpus size per language: 28.9 million words + 19 Million words in annexes, etc. - 23,000 texts per language (less in Bulgarian, Maltese and Romanian) - XML Format according to TEI P4, UTF-8-encoded - Aligned bilingually at paragraph level (often equivalent to sentences or sentence parts), using Vanilla. - Modular: download the languages you need. LANGUAGES Bulgarian, Czech, Danish, Dutch, English, Estonian, German, Greek, Finnish, French, Hungarian, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Slovak, Slovene, Spanish, Swedish. TEXT TYPES - Documents on contents, principles and political objectives of the EU Treaties; - EU legislation; - Declarations; - Resolutions; - Acts; - International agreements. PARAGRAPH ALIGNMENT Paragraph alignment for all 231 language pairs was carried out with the Vanilla aligner and is available for download. Paragraphs in the JRC-Acquis are frequently equivalent to sentences or even sentence parts. Version 2.2 of the JRC-Acquis corpus (210 language pairs, still available on the same website) was additionally aligned with HunAlign. - Paragraph-aligned for all 231 language pairs; - Paragraphs are sentence parts, sentences, or groups of sentences; - Using the Vanilla aligner; - Over 1 Million alignments per language pair (on average for all language pairs); - 85.43% one-to-one alignments (on average for all language pairs). MANUAL SUBJECT DOMAIN CLASSIFICATION - Manually classified according to EUROVOC subject domains; - Selected from 6000 hierarchically organised classes, wide-coverage; - suitable to experiment with multilingual multi-label categorisation. USE / DOWNLOAD - Download from http://langtech.jrc.it/JRC-Acquis.html; - Usage free for research purposes. FOR MORE DETAILS You will find a detailed description of version 2.2 of the corpus in the following paper. Please use the following reference when you mention the JRC-Acquis in any publications. We would be pleased to hear how you use the corpus. Steinberger Ralf, Bruno Pouliquen, Anna Widiger, Camelia Ignat, Toma? Erjavec, Dan Tufi?, D?niel Varga (2006). 'The JRC-Acquis: A multilingual aligned parallel corpus with 20+ languages'. Proceedings of the 5th International Conference on Language Resources and Evaluation (LREC'2006). Genoa, Italy, 24-26 May 2006. Available at http://langtech.jrc.it/#Publications. The JRC's Language Technology group specialises in the development of highly multilingual text analysis tools and in cross-lingual applications. An example is our multilingual (19 languages) news analysis application NewsExplorer, publicly accessible at http://press.jrc.it/NewsExplorer. Related JRC developments (both covering 22+ languages): - NewsBrief (http://press.jrc.it): breaking news detection and display of the very latest thematic news from around the world; - Medical Information System MedISys (http://medusa.jrc.it): displays the latest health-related news from around the world according to themes and diseases. Ralf Steinberger European Commission - Joint Research Centre (JRC) IPSC - SeS - EMM - Language Technology http://langtech.jrc.it, http://press.jrc.it/NewsExplorer From altun at tti-c.org Wed Aug 8 12:27:52 2007 From: altun at tti-c.org (Yasemin Altun) Date: Wed, 8 Aug 2007 11:27:52 -0500 Subject: Connectionists: PhD position at MPI for Biological Cybernetics: Learning from structured data Message-ID: <2334F75B-D792-4AEB-81F8-EE3FEFCE7580@tti-c.org> PhD. Studentship at the MPI for Biological Cybernetics: Learning from structured data A position for a PhD-studentship is available in the Empirical Inference department (Bernhard Schoelkopf) at the Max Planck Institute in Tuebingen, Germany (see http://www.kyb.tuebingen.mpg.de/bs), in the area of machine learning for structured data (Yasemin Altun). We invite applications of candidates with an outstanding academic record including a strong mathematical or analytical background. Max Planck Institutes are publicly funded research labs with an emphasis on excellence in basic research. Tuebingen is a university town in southern Germany, see http://www.tuebingen.de/kultur/ english/index.html for some pictures. The Max Planck Society is an equal opportunity employer: Handicapped individuals are strongly encouraged to apply, and so are women in areas in which they are underrepresented. Inquiries and applications, including a CV (with complete lists of marks, copies of transcripts etc.) and a short statement of research interests (matching some of the above areas) should be sent to altun at tti-c.org or Max Planck Institute for Biological Cybernetics Sabrina Nielebock Spemannstr. 38; 72076 Tuebingen Germany Tel. +49 7071 601 551 Fax +49 7071 601 552 In addition, please arrange for two letters of reference to be sent directly to the address above. Applications will be considered immediately and until the positions are filled. From tobias at nld.ds.mpg.de Thu Aug 9 05:56:29 2007 From: tobias at nld.ds.mpg.de (Tobias Niemann) Date: Thu, 09 Aug 2007 11:56:29 +0200 Subject: Connectionists: Postdoctoral Researcher and PhD-Student in Theoretical Neuroscience In-Reply-To: <468A63AD.2010302@nld.ds.mpg.de> References: <468A63AD.2010302@nld.ds.mpg.de> Message-ID: <46BAE4CD.4080605@nld.ds.mpg.de> The Max-Planck-Institute for Dynamics and Self-Organization (G?ttingen, Germany) and Bernstein Center for Computational Neuroscience (BCCN) G?ttingen invites applications for Postdoctoral Researcher and PhD-Student positions in Theoretical Neuroscience Recent progress in neuroscience enables experimental neuroscientists to simultaneously record the activity of up to hundreds of neurons in the brains of animals engaged in a cognitive task. The development of adequate models and mathematical tools for the analysis of such large scale neuronal activity patterns is thus an important challenge in theoretical neuroscience. The successful candidate will use approaches from statistical physics and dynamical systems theory to develop mathematical methods and models for the analysis of the coordinated dynamics of large ensembles of neurons and apply these methods to analyse in vivo and in vitro multi neuronal recordings. We are looking for applicants with a degree in physics or applied mathematics, preferably with prior experience in statistical physics or nonlinear dynamics and probabilistic data analysis, and for interdisciplinary research at the border of theoretical physics and neuroscience. Prior biological or neuroscience training is welcome but not required. The candidate's research will be supported by the recently established Bernstein Center for Computational Neuroscience (BCCN) in G?ttingen. G?ttingen is a center of neuroscience in Europe hosting numerous internationally recognized neuroscience research institutions, including three Max Planck Institutes, the European Neuroscience Institute, the German Primate Research Center, and G?ttingen University's Centers for Systems Neuroscience (ZNV) and for the Molecular Physiology of the Brain (CMPB). The BCCN integrates theoretical and experimental research groups from these institutions to foster interdisciplinary research in computational neuroscience specifically supporting close collaboration between theorists and experimental researchers. Please submit your application preferably in one single PDF-document, including cover letter, CV, statement of interests, list of publications, names of possible referees, relevant certificates until September 15, 2007, to: jobs at bccn-goettingen.de (Subject: ThN PostDoc/PhD) While e-mail is preferred, applications may also be submitted in hardcopy to the following address: Prof. Dr. Theo Geisel Subject: ThN PostDoc/PhD Bernstein Center for Computational Neuroscience (BCCN) G?ttingen Max-Planck-Institute for Dynamics and Self-Organization Bunsenstrasse 10 D - 37073 G?ttingen, Germany http://www.ds.mpg.de The MPIDS is an equal opportunity employer. From lmate at gatsby.ucl.ac.uk Sat Aug 11 18:51:45 2007 From: lmate at gatsby.ucl.ac.uk (Mate Lengyel) Date: Sun, 12 Aug 2007 00:51:45 +0200 Subject: Connectionists: postdoc in computational neuroscience Message-ID: <46BE3D81.9090406@gatsby.ucl.ac.uk> UNIVERSITY OF CAMBRIDGE The University is committed to equality of opportunity DEPARTMENT OF ENGINEERING Senior Research Associate in Computational Neuroscience A position exists for a Senior Research Associate (equivalent of a senior postdoctoral fellow) to work on theories of spike timing-based memory in the hippocampus. The project is funded by the Wellcome Trust and will involve work at the recently established Computational and Biological Learning Lab (learning.eng.cam.ac.uk) in close collaboration with Prof. Peter Dayan (Gatsby Computational Neuroscience Unit, UCL, www.gatsby.ucl.ac.uk). The project also involves collaboration with the groups of Dr. Ole Paulsen (Department of Physiology, Anatomy and Genetics, Oxford University, noggin.physiol.ox.ac.uk) and Dr. Francesco Battaglia (SILS Center for Neuroscience, University of Amsterdam) providing direct access to relevant in vitro and in vivo electrophysiological data. The aim of the project is to develop normative theories of spike-timing based interactions (neural dynamics and synaptic plasticity rules) between hippocampal neurons for efficient memory processing that make testable prediction at the electrophysiological level (starting from Lengyel et al, Nat Neurosci 2005; Lengyel & Dayan, Advances in NIPS 2007). Candidates must have a strong analytical background and demonstrable interest in theoretical neuroscience. They should have a PhD or equivalent in computational neuroscience, physics, mathematics, computer science, machine learning or a related field. Preference will be given to candidates with sufficient programming skills to run numerical simulations (eg. in C or MatLab) and expertise with neural network models, analysis of dynamical systems, and Bayesian techniques. Familiarity with the neurobiology of the hippocampus is an advantage. The appointment will be for 2 years initially (renewable for a 3rd year) starting 1 January 2008 or as soon as possible thereafter. Salary is highly competitive and is in the range ?33,779 to ?42,791 p.a. Further details may be obtained from Dr M?t? Lengyel, Department of Engineering, Trumpington Street, Cambridge, CB2 1PZ, (Tel +36 1 224 8327, Fax +36 1 224 8310, email lmate at gatsby.ucl.ac.uk), to whom a letter of application, a statement of research interests, and a CV (in pdf or plain text formats if possible) with the names and full contact details (including e-mail addresses) of three referees should be sent so as to reach him not later than 5 September 2007. Shortlisted applicants will be interviewed between 24 September-5 October, 2007. From hava at cs.umass.edu Sun Aug 12 12:03:11 2007 From: hava at cs.umass.edu (Hava Siegelmann) Date: Sun, 12 Aug 2007 13:03:11 -0300 Subject: Connectionists: please post: POST DOC position available now Message-ID: <46BF2F3F.7010802@cs.umass.edu> Immediate Post Doc Position is Available The BINDS lab at the University of Massachusetts at Amherst is offering an immediate postdoc position in the area of "MEMORY" with focus on changes in memories (reconsolidation), computational modeling, and the use of these to advance current machine learning paradigms and applications. The position is available to start now. Strong background is required in at least some of the related areas, and the particular research project will be fit to the strength of the candidate and the original project. Contact with CV and letters of recommendations directly to hava at cs.umass.edu thanks -- Hava T. Siegelmann http://www.cs.umass.edu/~hava/ From Randy.OReilly at colorado.edu Mon Aug 13 13:21:23 2007 From: Randy.OReilly at colorado.edu (Randall C. O'Reilly) Date: Mon, 13 Aug 2007 11:21:23 -0600 Subject: Connectionists: Faculty Position at University of Colorado Boulder Message-ID: <200708131121.24047.Randy.OReilly@colorado.edu> BEHAVIORAL NEUROSCIENCE FACULTY POSITION AT THE UNIVERSITY OF COLORADO AT BOULDER The Behavioral Neuroscience program in the Psychology Department at the University of Colorado invites applications for a tenure-track Assistant Professor position. We seek a broad range of applicants whose interest could be in the neural basis of memory, addiction, stress adaptations, pain modulation, recovery from brain damage, emotion and sensory processing. Researchers with a developmental/aging approach to these problems also will be considered. The selected candidate will have training in Neuroscience, postdoctoral experience, and demonstrated research independence, and will be expected to develop and maintain a strong extramurally-funded research program, as well as contribute to the department?s undergraduate and graduate teaching mission. A candidate?s qualifications are more important than the specific area of expertise. This individual will join a strong Neuroscience group on the Boulder Campus (see Univ. of Colorado Center for Neuroscience Web-site for more information: http://www.colorado.edu/neuroscienceprogram/index.html), and have the opportunity to interact with the Neuroscience group at the University of Colorado Health Sciences Center in Denver (http://www.uchsc.edu/neuroscience/). Applicants should submit a curriculum vitae, relevant reprints, a concise statement of research and teaching plans, and arrange to have three letters of recommendation sent to: Chair, Behavioral Neuroscience Search Committee, Department of Psychology, Muenzinger Building Room D244, UCB 345, University of Colorado at Boulder, Boulder, CO 80309. A review of completed applications will begin on Nov. 15, 2007 and will continue until the position is filled. The University of Colorado at Boulder is committed to diversity and equality in education and employment. Contact Person: Steven F. Maier (steven.maier at colorado.edu) From nips2007publicity at msn.com Fri Aug 17 15:49:44 2007 From: nips2007publicity at msn.com (NIPS 2007 Publicity) Date: Fri, 17 Aug 2007 12:49:44 -0700 Subject: Connectionists: [NIPS2007] Call for Demos Message-ID: CALL FOR DEMONSTRATIONS - NIPS 2007 Neural Information Processing Systems -- Natural and Synthetic NIPS 2006 Conference -- December 3 - 6, 2007 Hyatt Regency Vancouver, BC, CANADA www.nips.cc Demonstration Proposal Deadline: September 21, 2007 Would you like to interactively demonstrate your novel hardware, software, or wetware technology, your robot, or your chip to people at the NIPS 2007 Conference? The Neural Information Processing Systems Conference has a Demonstration Track that will run in parallel with the popular evening Poster Sessions. Demonstrators will have a chance to show their live interactive demos in the areas of hardware technology, neuromorphic and biologically-inspired systems, robotics, and software systems. The only hard rules are that the demo must show novel technology and must be LIVE and INTERACTIVE! (It is not a back-door Poster Session.) The full call for demonstrations is at the following URL: http://nips.cc/Conferences/2007/Calls/CallForDemos Giacomo Indiveri and Xubo Song From terry at salk.edu Fri Aug 17 17:23:56 2007 From: terry at salk.edu (Terry Sejnowski) Date: Fri, 17 Aug 2007 14:23:56 -0700 Subject: Connectionists: NEURAL COMPUTATION - SEPTEMBER 2007 In-Reply-To: Message-ID: Neural Computation - Contents - Volume 19, Number 9 - September 1, 2007 Letters A Multichip Neuromorphic System for Spike-Based Visual Information Processing Jacob Vogelstein, Udayan Mallik, Eugenio Culurciello, Gert Cauwenberghs and Ralph Etienne-Cummings Visual Recognition and Inference Using Dynamic Overcomplete Sparse Learning Joseph Murray and Ken Kreutz-Delgado Saccade Updating by Vector Subtraction: Analysis of Multiplicative Gainfield Models Carlos Cassanello and Vincent Ferrera Representations of Space and Time in the Maximization of Information Flow in the Perception-Action Loop Alexander Klyubin, Daniel Polani, and Chrystopher Nehaniv Computational Properties of Networks of Synchronous Groups of Spiking Neurons Judith Dayhoff Variable Time Scales of Repeated Spike Patterns in Synfire Chain with Mexican-Hat Connectivity Kosuke Hamaguchi, Masato Okada, and Kazuyki Aihara Asymptotic Behaviour and Synchronizability Characteristics of a Class of Recurrent Neural Networks Christof Cebulla Self-Organizing Maps with Asymmetric Neighborhood Function Takaaki Aoki and Toshio Aoyagi Parametric Embedding for Class Visualization Tomoharu Iwata, Kazumi Saito, Naonori Ueda, Sean Stromsten, Thomas Griffiths, and Joshua B. Tenenbaum MISEP Method for Post-Nonlinear Blind Source Separation Chun-Hou Zheng, De-Shuang Huang, Kang Li, George Irwin and Zhan-Li Sun ON-LINE - http://neco.mitpress.org/ SUBSCRIPTIONS - 2007 - VOLUME 19 - 12 ISSUES Electronic only USA Canada* Others USA Canada* Student/Retired $60 $63.60 $114 $54 $57.24 Individual $100 $106.00 $154 $90 $95.40 Institution $782 $828.92 $836 $704 $746.24 * includes 6% GST MIT Press Journals, 238 Main Street, Suite 500, Cambridge, MA 02142-9902. Tel: (617) 253-2889 FAX: (617) 577-1545 journals-orders at mit.edu http://mitpressjournals.org/neuralcomp From denver.h.dash at intel.com Wed Aug 15 12:51:55 2007 From: denver.h.dash at intel.com (Dash, Denver H) Date: Wed, 15 Aug 2007 09:51:55 -0700 Subject: Connectionists: CFP: Machine Learning Algorithms for Event Detection Message-ID: -------------------------------- Call for Papers -------------------------------- Machine Learning Algorithms for Event Detection A Special Issue of the Machine Learning Journal Submission deadline: November 28, 2007 http://www.pittsburgh.intel-research.net/~dhdash/mlj_eventdetection.html Denver Dash, Dragos Margineantu, and Weng-Keen Wong, guest editors We would like to invite submissions for a special issue of the Machine Learning Journal on "Machine Learning Algorithms for Event Detection". Event Detection is the task of monitoring a data source and detecting the occurrence of an event that is captured within that source. There are several sources of complexity for recent applications of event detection problems: * The variety of data sources is exploding, encompassing multivariate records, images, video, audio, spatio-temporal data, text documents, unstructured data and relational data; * The volume of data can be enormous, often measured in Terabytes; * Applications often involve monitoring of human life or critical assets and thus require extreme timeliness, high true-positive rates or low false-positive rates; * The event may be localized or distributed in time and/or space; * The event may be a never-before-seen "day-zero" event, which does not exist in training data; * The data source can be a single sensor, an array of identical sensors or an inhomogeneous mix of various sensors; * The problem is often exacerbated by the presence of an active adversary. These complexities pose an array of challenges for machine learning. Often the standard paradigms of supervised learning, unsupervised learning or even semi- supervised and active learning do not fit the event detection problems well. Addressing these issues would thus fill some important gaps in machine learning research and would impact many of the most pressing real-world applications being studied today, such as security, public health, biology, environmental sciences, manufacturing, astrophysics, finance, and business. The topics of interest include, but are not limited to: * Event detection in complex data such as video, audio, spatio-temporal data, text documents, functional neuro-imaging data, and relational data; * Anomaly detection; * Monitoring and surveillance based on sensor data and on multiple data sources; * The integration of learning and domain knowledge for event detection; * Analysis of the capabilities of learning algorithms for event detection; * Automated event detection in safety-critical applications; * Algorithms and tools for online event detection; * Online limiting of false alarm rates, analysis of error tradeoffs, risk models; * Scaling up detection algorithms to large populations; * Distributed algorithms for monitoring and surveillance; * Validation and testing of event detection and surveillance systems, and metrics for their performance; * Dealing with adversaries in surveillance tasks; * Machine learning research for related novel application domains. We encourage prospective authors to contact us (e-mail to d.margin at comcast.net) with a brief summary of their paper concept for feedback, especially for survey papers or for papers focused on applications. Submissions are expected to represent high-quality, significant contributions in the area of machine learning algorithms and/or applications of machine learning. Application papers are expected to describe the application in detail and to present novel solutions that have some general applicability (beyond the specific application). The authors should follow standard formatting guidelines for Machine Learning Journal manuscripts. Administrative notes: --------------------- * Authors retain the copyrights to their papers. (See publication agreement on the MLJ website: http://pages.stern.nyu.edu/~fprovost/MLJ). * Submissions and reviewing will be handled electronically using standard procedures for Machine Learning (http://mach.edmgr.com). * Authors must register with the system before they can submit their manuscripts. * Authors must select the appropriate Article Type, "Machine Learning for Event Detection", when submitting their manuscripts. * Accepted papers will be published electronically and citable immediately (before the print version appears). Schedule: --------- * Submission Deadline: November 28, 2007. * Acceptance Decisions: March 10, 2008. * Camera-Ready Papers Due: May 5, 2008. From Randy.OReilly at colorado.edu Tue Aug 21 05:14:45 2007 From: Randy.OReilly at colorado.edu (Randall C. O'Reilly) Date: Tue, 21 Aug 2007 03:14:45 -0600 Subject: Connectionists: Announcing: Emergent Neural Network Simulation Software (formerly PDP++) Message-ID: <200708210314.45574.Randy.OReilly@colorado.edu> Announcing: The Emergent Neural Network Simulation System http://grey.colorado.edu/emergent/index.php/Main_Page Emergent is a major rewrite of the widely used PDP++ system. Emergent is a comprehensive simulation environment for creating complex, sophisticated models of the brain and cognitive processes using neural network models. These networks can also be used for all kinds of other more pragmatic tasks, like predicting the stock market or analyzing data. Emergent includes a full GUI environment for constructing networks and the input/output patterns for the networks to process, and many different analysis tools for understanding what the networks are doing. It has a new tabbed-browser style interface (in Qt 4), with full 3D graphics (via Open Inventor/Coin3D), and powerful new GUI programming tools and data processing and analysis capabilities. It supports the same algorithms as PDP++: Backpropagation (feedforward and recurrent), Self-Organizing (e.g., Hebbian, Kohonen, Competitive Learning), Constraint Satisfaction (e.g., Boltzmann, Hopfield), and the Leabra algorithm that integrates elements of all of the above in one coherent, biologically-plausible framework. Relative to PDP++, the main advances are: * Much easier to modify and extend the "scripting" of network training through a new GUI-based programming system -- everything is transparent and user-modifiable. Considerable support is included for implementing complex psychological tasks via this programming environment. * The tabbed browser allows everything to be contained within a single window, with full search functions, cut/copy/paste, drag-and-drop, etc, for a modern, highly efficient working environment. * Everything has been boiled down to the most basic, general-purpose elements, which can now be combined in more powerful, "emergent" ways. Environments and monitor data and all other forms of data have been consolidated in a single powerful DataTable object that supports many different kinds of operations (e.g., database-style Joins and Sorts, vector and matrix math, 3d graphing, statistics, etc). With convenient interfaces for DataTables in the GUI programming environment, flexible and efficient data processing and analysis functions can be readily performed. * Has a greater variety of network visualization tools, and a built-in virtual environment simulator (based on the popular ODE toolkit) allows networks to interact with a realistic simulated environment, to explore more embodied and robotic functionality. * Standard GPL license, ./configure build process, native look-and-feel on all 3 major platforms (Linux, Mac, Windows), easily-installable binary packages (including apt & yum on linux), and dynamically-loadable plugin modules. Relative to the prevalent use of MATLAB and other general-purpose tools for neural neural network simulation, Emergent offers several important advantages: * completely open source, free software. * highly optimized execution speed, including distributed memory computation, while also supporting complex biologically-based neural architectures. * designed specifically to make research simulations easily accessible to other users with minimal additional effort: built-in documentation system, pervasive comment fields, accessible, transparent interface. In brief, if you're doing large scale, complex neural network models, Emergent offers many advantages. From mail at jan-peters.net Wed Aug 22 11:21:50 2007 From: mail at jan-peters.net (Jan Peters) Date: Wed, 22 Aug 2007 17:21:50 +0200 Subject: Connectionists: NIPS 2007 WORKSHOP: Robotics Challenges for Machine Learning Message-ID: <4C49CD7F-1DB4-4A65-B57B-D4CCAC5C3EB6@jan-peters.net> *** Apologies for Multiple Postings *** ======== ==== CALL FOR POSTERS ==== =========== NIPS 2007 WORKSHOP: Robotics Challenges for Machine Learning Dates: 7-8 December, 2007 Organizers: Jan Peters (Max Planck Institute for Biological Cybernetics & USC), Marc Toussaint (Technical University of Berlin) WWW: http://www.robot-learning.de email: nips07 at robot-learning.de Abstract Submission Deadline: October 21, 2007 Acceptance Notification: October 26, 2007 ======== ==== CALL FOR POSTERS ==== =========== Abstract: Creating autonomous robots that can assist humans in situations of daily life is a great challenge for machine learning. While this aim has been a long standing vision of robotics, artificial intelligence, and the cognitive sciences, we have yet to achieve the first step of creating robots that can accomplish a multitude of different tasks, triggered by environmental context or higher level instruction. Despite the wide range of machine learning problems encountered in robotics, the main bottleneck towards this goal has been a lack of interaction between the core robotics and the machine learning communities. To date, many roboticists still discard machine learning approaches as generally inapplicable or inferior to classical, hand-crafted solutions. Similarly, machine learning researchers do not yet acknowledge that robotics can play the same role for machine learning which for instance physics had for mathematics: as a major application as well as a driving force for new ideas, algorithms and approaches. Some fundamental problems we encounter in robotics that equally inspire current research directions in Machine Learning are: -- learning and handling models, (e.g., of robots, task or environments) -- learning deep hierarchies or levels of representations (e.g., from sensor & motor representations to task abstractions) -- regression in very high-dimensional spaces for model and policy learning -- finding low-dimensional embeddings of movement as an implicit generative model -- methods for probabilistic inference of task parameters from vision, e.g., 3D geometry of manipulated objects -- the integration of multi-modal information (e.g., proprioceptive, tactile, vision) for state estimation and causal inference -- probabilistic inference in non-linear, non-Gaussian stochastic systems (e.g., for planning as well as optimal or adaptive control) Robotics challenges can inspire and motivate new Machine Learning research as well as being an interesting field of application of standard ML techniques. Inversely, with the current rise of real, physical humanoid robots in robotics research labs around the globe, the need for machine learning in robotics has grown significantly. Only if machine learning can succeed at making robots fully adaptive, it is likely that we will be able to take real robots out of the research labs into real, human inhabited environments. To do so, future robots will need to be able to make proper use of perceptual stimuli such as vision, proprioceptive & tactile feedback and translate these into motor commands. To close this complex loop, machine learning will be needed on various stages ranging from sensory-based action determination over high-level plan generation to motor control on torque level. Among the important problems hidden in these steps are problems which can be understood from the robotics and the machine learning point of view including perceptuo-action coupling, imitation learning, movement decomposition, probabilistic planning problems, motor primitive learning, reinforcement learning, model learning and motor control. Format: The goal of this one-day workshop is to bring together people that are interested in robotics as a source and inspiration for new Machine Learning challenges, or which work on Machine Learning methods as a new approach to robotics challenges. In the robotics context, among the questions which we intend to tackle are Reinforcement Learning, Imitation, and Active Learning: * What methods from reinforcement learning scale into the domain of robotics? * How can we improve our policies acquired through imitation by trial and error? * Can we turn many simple learned demonstrations into proper policies? * Does the knowledge of the cost function of the teacher help the student? * Can statistical methods help for generating actions which actively influencing our perception? E.g., Can these be used to plan visuo-motor sequences that will minimize our uncertainty about the scene? * How can image understanding methods be extended to provide probabilistic scene descriptions suitable for motor planning? Motor Representations and Control: * Can we decompose human demonstrations into elemental movements, e.g., motor primitives, and learn these efficiently? * Is it possible to build libraries of basic movements from demonstration? How to create higher-level structured representations and abstractions based on elemental movements? * Can structured (e.g., hierarchical) temporal stochastic models be used to plan the sequencing and superposition of movement primitives? * Is probabilistic inference the road towards composing complex action sequences from simple demonstrations? Are superpositions of motor primitives and the coupling in timing between these learnable? * How to generate compliant controls for executing complex movement plans which include both superposition and hierarchies of elemental movements? Can we find learned versions of prioritized hierarchical control? * Can we learn how to control in task-space of redundant robots in the presence of under-actuation and complex constraints? Can we learn force or hybrid control in task-space? * Is real-time model learning the way to cope with executing tasks on robots with unmodeled nonlinearities and manipulating uncertain objects in unpredictable environmental interactions? * What new regression techniques can help real-time model learning to improve the execution of tasks on robots with unmodeled nonlinearities and manipulating uncertain objects in unpredictable environmental interactions? Learning structured models and representations: * What kind of probabilistic models provide a compact and suitable description of real-world environments composed of manipulable objects? * How can abstractions or compact representations be learnt from sensori-motor data? * How can we extract features of the sensori-motor data that are relevant for motor control or decision making? E.g., can we extract visual features of objects directly related to their manipulability or ``affordance''? Posters: We are open for any posters posing problems for machine learning and for presenting machine learning algorithms with applications in robotics. The deadline for abstract submissions is October 21, 2007 and the notification will be October 26, 2007 Abstract Submission Deadline: October 21, 2007 Acceptance Notification: October 26, 2007 From giulio.sandini at iit.it Mon Aug 20 10:05:27 2007 From: giulio.sandini at iit.it (Giulio Sandini) Date: Mon, 20 Aug 2007 16:05:27 +0200 Subject: Connectionists: PhD fellowships: Brain Machine Interface project at IIT Message-ID: <004301c7e333$299ada40$7cd08ec0$@sandini@iit.it> The Robotics, Brain and Cognitive Sciences (RBCS) Department of the Italian Institute of Technology (IIT) is offering fellowships for the in-vivo studies of BRAIN MACHINE INTERFACE. These fellowships are part of a multidisciplinary project aiming at 'reading' the brain to understand and extract motor signals which may be used to control an artificial limb. The project will be developed jointly at RBCS department of IIT by a group of scientists coordinated by Luciano Fadiga and including: Stefano Panzeri, Alessandro Vato, Gytis Baranauskas, Davide Ricci and Franco Bertora. The BMI project addresses topics such as the design of microelectrode and microelectronics devices for chronic in-vivo recording, electrophysiological and brain signals recording, investigation of the coding/decoding issue, functional identification of brain motor/premotor areas, and direct connection to artificial actuators. More specifically the seven research themes proposed are (short abstract and scientist in charge are included at the end of the message): . Theme 5.8: The Neural Interface Problem: Enhanced in-vivo electrodes by nanomaterial coatings . Theme 5.9: The Signal Treatment Problem . Theme 5.10: The Brain Signal Decoding Problem . Theme 5.11: The Movements vs. Actions Problem . Theme 5.12: The Neurophysiology of the Human Brain . Theme 5.13: The Role of Sensory Feedback in Brain Machine Interface . Theme 5.14: Machinery for Functional Brain Analysis Interested applicants should refer to one of the following website to download instructions on how to apply and/or contact directly the scientists in charge (below) for more information regarding the individual research plans. http://www.liralab.it/IIT_school/CICLOXXIII/Concorso.htm http://www.iit.it/phd_positions --- Prof. Giulio Sandini Italian Institute of Technology Robotics, Brain and Cognitive Sciences Department Phone: +39 010 7178101 - Fax +39 010 720321 and LIRA-Lab, University of Genova Phone: +39 0103532779 - Fax: +39 010353.2948 http://www.liralab.it http://sandini.liralab.it -------------------------------------------- RESEARCH TOPICS PROPOSED Theme 5.8 - The Neural Interface Problem: Enhanced in-vivo electrodes by nanomaterial coatings Tutor: Dr. Davide Ricci. N. of available positions: 1 Within the Brain Machine Interface research project of IIT, that has the ultimate goal of extracting and decoding brain signals to drive artificial actuators, a key issue is the investigation on how such brain signals can be extracted from electrical recordings with the necessary temporal and spatial resolutions. Nanomaterial coatings, such as carbon nanotubes, both unmodified or bio-functionalized, offer the possibility to improve the recording properties of traditional metal electrodes. Through direct integration of nanomaterials in the electrode fabrication process, this Ph.D. research project will deal with the following tasks: (1) designing efficient, long-term recording microelectrodes; (2) investigating the possibility to record signals from the surface of the cortex; (3) investigating the problem of input impedance and making attempts to reduce it without loss in signal-to-noise ratio; (4) studying how to minimize tissue reactions, such as glyosis. The ideal candidate would have a background in one or more of the following fields: material science, electrochemistry, micromechanics, nanotechnology, physics. For further details concerning the research project, please contact: davide.ricci at iit.it) Theme 5.9 - The Signal Treatment Problem Tutor: Dr. Gytis Baranauskas N. of available positions: 1 Any brain signal has to be amplified and processed before it can be used to control a prosthetic device or a robotic manipulator. Moreover, the device that amplifies and elaborates brain signals should be as small as possible. We already have an integrated circuitry that is smaller than a finger nail and that amplifies 64 independent neuronal signals. The goal of this largely electronic engineering project is to build a single chip powered by radio-waves that amplifies and processes signals from hundreds of neurons in such a way that the chip output can be directly fed into the artificial system driving a robotic arm. We expect to test this chip in animals as well as in human patients. Thus, we are looking for a PhD student interested in analog and digital microelectronics for biomedical applications and preferentially with background in physics. The selected student will be working in close collaboration with the project 5.8 team (see above). For further details concerning the research project, please contact: baranauskas at elet.polimi.it Theme 5.10: The Brain Signal Decoding Problem Tutor: Prof. Stefano Panzeri N. of available positions: 1 A fundamental question in the development of brain-machine-interfaces is how to extract information about sensory stimulus or motor commands from a single-trial observation of neuronal activity. This mathematical-analysis PhD project will aim at addressing this question by investigating systematically which features of different types of recordings of neural activity (such as spike trains of well isolated neurons, field potentials, multiple-unit activity or other) convey the most information about sensory stimuli or motor actions. We will develop data analysis techniques based on the principles of information theory and then apply them to recordings of brain activity provided by our experimental collaborators, with the goal of determining how best to decode these brain signals. The ideal candidate for this PhD studentship will a have a strong degree in a numerate discipline such as physics, statistics, mathematics or computer science. No previous knowledge of neuroscience is needed, although a strong motivation to contribute to brain research is essential. For more details concerning the research project, please contact: stefano.panzeri at manchester.ac.uk Theme 5.11: The Movements vs. Actions Problem Tutor: Prof. Luciano Fadiga N. of available positions: 1 Apart from very few exceptions, the research groups currently working at BMI are doing their attempts by recording from the primary motor cortex. Their goal is to decode directional tuning and individual muscles control signals. We consider this approach quite risky. First of all because several researchers are now disputing the idea that the motor cortex codes the direction of reaching in absolute terms, second because recent neurophysiological evidence shows that actions and not movements are mainly coded by the brain. Within this field of research, one PhD student will be involved in cortical electrophysiology to record single neurons' signals. The aim is twofold: to study and understand the motor commands generated by the brain during goal-directed acts and to set up long-term chronical recording techniques, firstly in monkeys and then in humans. Backgrounds in computer science, electronics and basic neuroscience are required. For further details concerning the research project, please contact: luciano.fadiga at iit.it Theme 5.12: The Neurophysiology of the Human Brain Tutor: Dr. Elisa Molinari N. of available positions: 1 This work will concern brain imaging (functional magnetic resonance) to investigate the cortical and subcortical activity of the motor system during goal-directed actions. Through this project we will better understand the functional correlates of motor planning/execution by analyzing data and developing new single-subject analysis techniques. This will be done by taking into account both the statistical significance and the intensity (signal-to-noise ratio) of the activations. We are looking forward for one PhD student which should be competent in physics, computer science and basic neuroscience. For further details concerning the research project, please contact: elisa.molinari at iit.it Theme 5.13: The Role of Sensory Feedback in Brain Machine Interface. Tutor: Dr. Alessandro Vato. N. of available positions: 1 Within this field of research we will study in animal models (and then in human patients) the relevance of sensory afferents for controlling an artificial effector. Somatosensory real-time feedback is fundamental for motor planning and for executing "on-line" errors correction during movements. In people with sensory motor disabilities, the sensory information that cannot reach the brain, can be "substituted" through an intact sensory channel (i.e. eyes or ears) different from the damaged one. Alternatively, the damaged sensory pathway can be "replaced" trying to achieve the same sensation in an artificial way. The goal of this project is to design an encoder interface to stimulate the sensory cortex of behaving rats conveying sensory information related to the state of an external device. The encoder will be part of a Bidirectional Brain Machine Interface System in which neural signals recorded directly from the rat's motor cortex will control an external device and real-time feedback will be provided via electrical stimulation of the sensory cortex. The candidate for this PhD position will be required to have a background in computer science, electronics and basic neuroscience. For further details concerning the research project, please contact: alessandro.vato at iit.it Theme 5.14: Machinery for Functional Brain Analysis Tutor: Dr. Franco Bertora N. of available positions: 1 In addition, and in parallel with the preceding themes, there is at IIT an ongoing program to investigate the frontiers of functional MRI. Any fMRI of the motor cortex has so far been performed on subjects confined in a supine/prone position in the limited volume of a traditional scanner. There are reasons to think that the analysis of subjects performing motor tasks in a more "natural" environment could produce different and more meaningful results. A study is currently in progress to determine the feasibility of a scanner allowing functional brain analysis of a human adult in a standing or sitting position. We are looking for one PhD student with background in physics, electronics, signal processing and MRI to explore the possibly novel imaging techniques (MRI sequences, data acquisition modalities and image reconstruction) to be included in the development of the scanner. For further details concerning the research project, please contact: franco.bertora at iit.it From D.Hardoon at cs.ucl.ac.uk Mon Aug 20 10:12:40 2007 From: D.Hardoon at cs.ucl.ac.uk (David R. Hardoon) Date: Mon, 20 Aug 2007 15:12:40 +0100 Subject: Connectionists: NIPS 07: Music, Brain & Cognition Workshop Message-ID: <89258F6F-A8BF-4444-A540-88AFC967BBBA@cs.ucl.ac.uk> Apologies for cross posting and please forward to whom ever this may be of interest to. NIPS'07 Workshop - Whistler, BC, December 7-8, 2007 Music, Brain & Cognition Day 1: "Learning the Structure of Music and its Effects on the Brain" Day 2: "Models of Sound and Cognition" ====================================================== http://homepage.mac.com/davidrh/MCBworkshop07/ Call for contributions ------------ We call for paper contribution of up to 8 pages to the workshop using NIPS style. The accepted papers will be available for downloading from this site. Selected papers will be considered for publication in a special issue of ?Journal of New Music Research?. Day 1 - Machine learning based models for learning the structure of music - Models for predicting style of performers - Analysis and models of fMRI/EEG/MEG scans from musical stimuli (as opposed to simplistic auditory stimuli) - Predicting music generated patterns in fMRI/EEG/MEG - Strategies for embedding representations of musical experience into generative music / performance systems - Methods for generative musical performance and composition - Generative music and/or performance systems based on models of brain functioning - Similar and further models for learning and analysing the structure of music Day 2 - Computational models of cognitively inspired sound processing - Top down control of musical processing of pitch, onset, timbre - Models of musical memory, saliency, attention - Models of music development and learning - Computer aided sound design - Models as above, applied to other domains (e.g. speech and vision) with potential application in music Accepted papers will be either presented as a talk or poster (with poster spotlight) Papers should be submitted to the organisers D.Hardoon at cs.ucl.ac.uk, hpurwins at iua.upf.edu and please indicate if you wish to present on day 1 or day 2 and whether you only wish to present a poster. Important Dates ------------ Deadline for submissions: October 10, 2007 Notification of acceptance: October 31, 2007 Workshop taking place: December 7-8, 2007 Description ------------ Music is one of the most widespread of human cultural activities, existing in some form in all cultures throughout the world. The definition of music as organised sound is widely accepted today but a na?ve interpretation of this definition may suggest the notion that music exists widely in the animal kingdom, from the rasping of crickets' legs to the songs of the nightingale. However, only in the case of humans does music appear to be surplus to any obvious biological purpose, while at the same time being a strongly learned phenomenon and involving significant higher order cognitive processing rather than eliciting simple hardwired responses. A two day workshop will take place at NIPS 07 (Vancouver, Canada) and will span topics from signal processing and musical structure to the cognition of music and sound. In the first day the workshop will provide a forum for cutting edge research addressing the fundamental challenges of modeling the structure of music and analysing its effect on the brain. It will also provide a venue for interaction between the machine learning and the neuroscience/brain imaging communities to discuss the broader questions related to modeling the dynamics of brain activity. During the second day the workshop will focus on the modeling of sound, music perception and cognition. These have provide, with the crucial role of machine learning, a break-through in various areas of music technology, in particular: Music Information Retrieval (MIR), expressive music synthesis, interactive music making, and sound design. Understanding of music cognition in its implied top-down processes can help to decide which of the many descriptors in MIR are crucial for the musical experience and which are irrelevant. The organisers of the workshop are investigators for three main European projects; Learning the Structure of Music (Le StruM), Closing the loop of Evaluation and Design (CLOSED), Emergent Cognition through Active Perception (EmCAP) Mention. The target group is of researchers within the fields of (Music) Cognition, Music Technology, Machine Learning, Psychology, Sound Design, Signal Processing and Brain Imaging. For more details, please go to http://homepage.mac.com/davidrh/ MBCworkshop07/ Day 1 - http://homepage.mac.com/davidrh/MBCworkshop07/Day_1.html Day 2 - http://homepage.mac.com/davidrh/MBCworkshop07/Day_2.html Organisers ------------ Day 1 * David R. Hardoon (University College London) * Eduardo Reck-Miranda (University of Plymouth) * John Shawe-Taylor (University College London) Day 2 * Hendrik Purwins (Universitat Pompeu Fabra) * Xavier Serra (Universitat Pompeu Fabra) * Klaus Obermayer (Berlin University of Technology) ---------------------------------------------------------------------- "Who dares... wins" Dr. David R. Hardoon The Centre for Computational Statistics & Machine Learning Intelligent Systems Research Group Dept. of Computer Science University College, London Gower Street London, UK WC1E 6BT Tel: +44 (0) 20 7679 0425 Fax: +44 (0) 20 7387 1397 Email: D.Hardoon at cs.ucl.ac.uk www: http://homepage.mac.com/davidrh/ From dnoelle at ucmerced.edu Tue Aug 21 00:40:28 2007 From: dnoelle at ucmerced.edu (David C. Noelle) Date: Mon, 20 Aug 2007 21:40:28 -0700 Subject: Connectionists: Faculty Positions in Cognitive Science at the University of California, Merced Message-ID: <1187671228.3052.16.camel@74-60-145-129.mrc.clearwire-dns.net> University of California, Merced Tenured Faculty Position in Cognitive Science The University of California at Merced invites applications for a Full or Associate Professor of Cognitive Science, to join a growing group of researchers building a world-class program in cognitive science. We seek individuals with an outstanding publication record and a well-established interdisciplinary research program, especially those with broad research interests. Applicants should have a proven track record in obtaining external funding and attracting graduate students. To apply or for more information, please visit our website: http://jobs.ucmerced.edu/n/academic/position.jsf?positionId=1081 . Application deadline: 10/1/07. AA/EOE. Also see: http://cogsci.ucmerced.edu . ---------- University of California, Merced Assistant Professor of Cognitive Science The University of California at Merced invites applications for an Assistant Professor of Cognitive Science, to join a growing group of researchers building a world-class program in cognitive science. Preference will be given to those with broad research and teaching interests related to perception. Applicants should have a background in cognitive science or related fields, a strong publication record, an interdisciplinary research program, and a commitment to obtaining extramural funding. To apply or for more information, please visit our website: http://jobs.ucmerced.edu/n/academic/position.jsf?positionId=1082. Application deadline: 12/1/07. AA/EOE. Also see: http://cogsci.ucmerced.edu . From Jakob.Verbeek at inria.fr Mon Aug 20 04:03:55 2007 From: Jakob.Verbeek at inria.fr (Jakob Verbeek) Date: Mon, 20 Aug 2007 10:03:55 +0200 Subject: Connectionists: Postdoc Position Computer Vision & Machine Learning at INRIA Rhone-Alpes Message-ID: <46C94AEB.9050206@inria.fr> The LEAR research team at INRIA Rhone-Alpes in Grenoble (France) is seeking outstanding postdoctoral researchers to join our lab for one or two years, starting fall 2007 to early 2008. Candidates should have a strong background and interest in Computer Vision or Machine Learning, and ideally both. The postdoc will work primarily within the EU project CLASS, with focus on learning from loosely annotated images, such as found eg. on web-based image archives such as flickr. The goal is to improve text-based search results by analyzing co-occurrence of tags and visual content, a second goal is the detection of missing tags and erroneous tags. The LEAR (Learning and Recognition in Vision) team studies methods for recognition and detection of visual categories, as well as large-scale image databases. Many of LEAR projects interface between machine learning and computer vision, investigating generative and discriminative probabilistic models, as well as kernel and boosting methods. Currently our group has 4 full-time staff members (Cordelia Schmid, Bill Triggs, Frederic Jurie, Herve Jegou), 3 postdocs, 2 engineers, 9 PhD students, and several MSc students. Grenoble is a lively city which hosts many foreign students and researchers. Located in the heart of the French Alps its direct surroundings offer great outdoor recreation including skiing, cycling, and hiking. Paris can be reached from Grenoble in 3h by train. Candidates should send a CV, and a list of three references electronically to Cordelia.Schmid at inrialpes.fr and Jakob.Verbeek at inrialpes.fr References: - Website LEAR http://lear.inrialpes.fr - Website CLASS http://class.inrialpes.fr From randi.cohen at taylorandfrancis.com Mon Aug 20 11:29:25 2007 From: randi.cohen at taylorandfrancis.com (Cohen, Randi) Date: Mon, 20 Aug 2007 11:29:25 -0400 Subject: Connectionists: New Machine Learning and Pattern Recognition Book Series Message-ID: <4E08CBAE345C2942A61C4A65874BE9E5027D7C8D@US-EXBE01.NA.CorpLAN.net> ------------------- A Call for Authors ------------------- Introducing a New Series from Chapman & Hall/CRC Press Machine Learning and Pattern Recognition Series AIMS AND SCOPE: The field of machine learning has experienced significant growth in the past two decades as new algorithms and techniques have been developed and new research and applications have emerged. This series reflects the latest advances and applications in machine learning and pattern recognition through the publication of a broad range of reference works, textbooks, and handbooks. We are looking for single authored works and edited collections that will: * Present the latest research and applications in the field, including new mathematical, statistical, and computational methods and techniques * Provide both introductory and advanced material for students and professionals * Cover a broad range of topics around learning and inference The inclusion of concrete examples, applications, and methods is highly encouraged. The scope of the series includes, but is not limited to, titles in the areas of machine learning, pattern recognition, computational intelligence, robotics, computational/statistical learning theory, natural language processing, computer vision, game AI, game theory, neural networks, and computational neuroscience. We are also willing to consider other relevant topics, such as machine learning applied to bioinformatics or cognitive science, which might be proposed by potential contributors. Series Editors Ralf Herbrich Microsoft Research Ltd. Cambridge, UK Ralf.Herbrich at taylorandfrancis.com Thore Graepel Microsoft Research Ltd. Cambridge, UK Thore.Graepel at taylorandfrancis.com Proposals for this series may be submitted to the series editors or directly to: Randi Cohen Computer Science Acquisitions Editor Chapman & Hall/CRC Press Randi.Cohen at taylorandfrancis.com www.crcpress.com From odobez at idiap.ch Tue Aug 21 04:26:11 2007 From: odobez at idiap.ch (Jean-Marc Odobez) Date: Tue, 21 Aug 2007 10:26:11 +0200 Subject: Connectionists: PhD in human activity modeling at IDIAP (Switzerland) Message-ID: <46CAA1A3.6030808@idiap.ch> PhD student position in computer vision: human and scene activity modeling The IDIAP Research Institute (www.idiap.ch), a laboratory associated with EPFL (Swiss Federal Insitute of Technology, Lausanne), is looking for a researcher interested in human activity modeling and recognition from video sequences in indoor or outdoor scenes. The goal will be to investigate principled methods based on generative and discriminative modeling to address various issues in multi-person tracking, event and activity recognition, and visual data mining. One focus of the research will be the use of long term observations to model human and scene activities. The research will initially be conducted in the context of the CARETAKER project (http://www.ist-caretaker.org/), a multi-site project funded by the European Community. The project aims at recognizing events from long term data recorded over a network of cameras in real public transportation sites. We are seeking for candidates with the follwing profile: - solid knowledge of and programming experience with C++ - initial background in computer vision/image processing or machine learning - fluent in English, both written and spoken - willing to travel Start date: As soon as possible. We will hire as soon as a strong candidate is found. Duration: 3 to 4 years. The successful Ph.D student will be enrolled in the EPFL doctoral program. Appointment for the PhD student position is for a maximum of 4 years, provided successful progress are made. Contact: Dr Jean-Marc Odobez odobez at idiap.ch Tel : +41 (0)27 721 77 26 About IDIAP: ------------ IDIAP is a research institute associated with EPFL (Swiss Federal Institute of Technology, Lausanne) active in the fields of information retrieval, speech, machine learning and vision (www.idiap.ch). IDIAP is located in the town of Martigny in Valais, a scenic region in the south of Switzerland, surrounded by the highest mountains of Europe, and offering exciting recreational activities, including hiking, climbing and skiing, as well as varied cultural activities. IDIAP is an equal opportunity employer and offers competitive salaries and conditions at all levels in a young, dynamic, and multicultural environment. Although IDIAP is located in the French part of Switzerland, English is the main working language at IDIAP. Free French courses are also provided. Application: ------------- Interested candidates should submit their application (detailed CV, name of three references) by email to odobez at idiap.ch with cc to jobs at idiap.ch. From mark.plumbley at elec.qmul.ac.uk Tue Aug 21 09:16:37 2007 From: mark.plumbley at elec.qmul.ac.uk (Mark Plumbley) Date: Tue, 21 Aug 2007 14:16:37 +0100 Subject: Connectionists: ICA 2007: Registration deadline approaching Message-ID: <3399496864F99445B051FD9556FF3B6F6ED469@staff-mail1.vpn.elec.qmul.ac.uk> [Apologies for multiple receipts of this message.] *** REGISTRATION DEADLINE APPROACHING *** ICA 2007 7th International Conference on Independent Component Analysis and Signal Separation London, UK 9-12 September 2007 www.ica2007.org www.elec.qmul.ac.uk/ica2007 There are only a few days left until the advanced registration deadline for ICA 2007, to be held at Queen Mary, University of London, from Sunday 9 September to Wednesday 12 September 2007. For programme information, including information on the keynote talks, tutorials, oral and poster presentations, see www.elec.qmul.ac.uk/ica2007/programme.html Dates & Deadlines On-campus Accommodation: 23 August 2007 Advanced Registration: 24 August 2007 Tutorials: 9 September 2007 Conference: 10-12 September 2007 ------- -- Dr Mark D Plumbley Centre for Digital Music Department of Electronic Engineering Queen Mary University of London Mile End Road, London E1 4NS, UK Tel: +44 (0)20 7882 7518 Fax: +44 (0)20 7882 7997 Email: mark.plumbley at elec.qmul.ac.uk http://www.elec.qmul.ac.uk/people/markp/ From marcus at idsia.ch Thu Aug 23 23:06:32 2007 From: marcus at idsia.ch (Marcus Hutter) Date: Fri, 24 Aug 2007 13:06:32 +1000 Subject: Connectionists: PhD Student Scholarships Message-ID: <09e001c7e5fb$c9f61c50$3cd6cb96@crl174ml1s> The Computer Sciences Lab at the Australian National University is actively looking to recruit creative PhD candidates with excellent mathematical background, interested in - machine learning and/or - artificial intelligence. A non-exhaustive list of areas of interest is - Bayesian statistics, - universal forecasting, - information theory, - agent/control theory, - complexity theory, - decision theory, - image processing, - philosophy of science, - and/or related areas. Australian students: Please go to http://www.hutter1.net/official/jobs.htm for further information and how to apply. International students: You need a Scholarship from your country or university, or apply till 31 August for a Scholarship at the ANU. Warning: These international scholarships are quite difficult to get, and you would typically need a couple of relevant and high quality publications in the area already, or excellent results from a top university Masters degree. Please go to http://www.hutter1.net/official/jobs.htm and send me a pre-application well before the End of August. ______________________ Marcus Hutter, Assoc. Prof. RSISE, Room B259, Building 115 Australian National University Corner of North and Daley Road Canberra ACT 0200, Australia Phone: +61(0)2 612 51605 Fax: +61(0)2 612 58651 Email: marcus.hutter at anu.edu.au http://rsise.anu.edu.au/~marcus From d.polani at herts.ac.uk Fri Aug 24 15:31:36 2007 From: d.polani at herts.ac.uk (Daniel Polani) Date: Fri, 24 Aug 2007 21:31:36 +0200 Subject: Connectionists: PhD Studentship: Steps Toward Enactive Artificial Intelligence. Message-ID: <18127.12824.528283.738539@perm.feis.herts.ac.uk> UNIVERSITY OF HERTFORDSHIRE, UK A collaborative Interfaculty Research Studentship of the School of Computer Science (Adaptive Systems Research Group)/School of Humanities (Philosophy) on "Steps towards Enactive Artificial Intelligence" is available, to start in October 2007, or as soon as possible thereafter. Recent significant advances in the application of information-theoretic formalisms to embodied agents have contributed to clarify the role of an agent's embodiment and indicates possible mechanisms for the self-organized formation of favoured concepts and behaviours in both biological and artificial agents through interaction with the environment over time. The PhD Studentship will aim at further deepening our theoretical and practical understanding of the interaction between external environment and the development of the agent's internal models, with possible ramifications to addressing issues in biological information processing and fundamental questions of psychology and philosophy. Applications are invited from candidates with excellent programming skills and a very strong mathematical background from a suitable quantitative discipline (e.g. computer science, mathematics, physics). While previous knowledge is not required, the candidate should have a strong interest in artificial intelligence, artificial life and/or philosophy of cognition and a keen ambition to contribute to these fields. Experience in autonomous robots would be useful, but is not required. The PhD will be supervised by Daniel Polani and Daniel Hutto. Closing date for applications: 7th September 2007 Please note that, due to funding body restrictions, eligibility is limited to UK and EU citizens. For an application form, please email: research.studentships at herts.ac.uk For further information on the project, please email: d.polani at herts.ac.uk See also the homepages: Daniel Polani: http://homepages.feis.herts.ac.uk/~comqdp1/ Daniel Hutto: http://perseus.herts.ac.uk/uhinfo/schools/hum/subject/phil/staff/hutto/hutto_home.cfm //////////////////////////////////////////////////////////////////////// The Adaptive Systems Research Group in the Centre for Computer Science and Informatics Research of STRI is a strongly interdisciplinary team including roboticists, biologists, cognitive scientists, mathematicians and computer scientists. Members of the group are internationally leading in research including social and developmental robotics, emotion modelling, evolution of sensors and the perception-action loop. Associated to the group is the Hertfordshire Interactive Systems and Robotics Laboratory that is particularly suited for experiments involving physical robots. The laboratory has a range of small and medium-sized mobile robotic platforms as well as humanoid robots. Group members are currently involved in the following FP6 projects: Robotcub, Cogniron, Feelix Growing, eCircus, Iromec, Humaine, euCognition, and Euron. The group, consists of 5 core academic leaders, 15 research fellows and 14 PhD students plus a regular influx of shorter-term research visitors from other leading labs in Japan and the EU. More information about the group can be found at: URL: http://adapsys.feis.herts.ac.uk/ From hans.ekkehard.plesser at umb.no Fri Aug 24 19:51:58 2007 From: hans.ekkehard.plesser at umb.no (Dr. Hans Ekkehard Plesser) Date: Sat, 25 Aug 2007 01:51:58 +0200 Subject: Connectionists: Scientific Software Developer Position Message-ID: <200708250151.58637.hans.ekkehard.plesser@umb.no> Scientific Software Developer ============================= A full-time position as overingeni?r (code 1087) or senioringeni?r (code 1181) is available at the Depart?ment of Mathematical Sciences and Technology of the Norwegian University of Life Sciences. The position is financed through a grant from the Research Council of Norway and is limited to four years. The Computational Neuroscience Group at the Department of Mathematical Sciences and Technology of the Norwegian University of Life Sciences has just received a substantial grant under the eScience program of the Research Council of Norway (RCN): eNEURO?-?Multilevel Modeling and Simulations. We are thus expanding our activities and are looking for a scientific software developer to support us in the development of advanced neuronal modeling software as part of the eNEURO project. For information on our group and the eNEURO project, see arken.umb.no/compneuro. Our group presently consists of three permanent faculty members (Einevoll, Plesser, Wyller), two post-docs and four doctoral students, and will grow substantially in the next few years. We have close collaborations with experimental and computational neuroscientists at University of Oslo (Storm, Heggelund), University of California at San Diego (Dale, Devor), RIKEN Brain Science Institute (Diesmann, Gr?n) and the Honda Research Institute Europe (Gewaltig). We enjoy priori?tized access to the Norwegian national scientific high-performance computing resources as part of the eNeuro grant (www.notur.no). As a member of the NEST Initiative, we are closely involved in the development of the NEST simulator for large neuronal networks, one of the leading simulators in the field (www.nest-initiative.org). As a scientific software developer in our group, you will work closely with scientists and other NEST developers on the development of the simulator and other software tools for modeling neurons and neural networks. You will also be involved in organizing data sharing tools to facilitate the collaboration among our project partners around the world. As an applicant, you are strong on advanced C++ programming techniques including parallel programming and experienced with Python. Experience with collaboration tools, e.g. code reposi?tories and wikis, is a plus. Applicants should have at least a M.Sc. in computer science, physics, mathematics or computational biology. The working language in our group is English. We would like to increase the number of women in our group and encourage women to apply. The position is financed through the eNeuro grant of the Research Council of Norway limited to four years, starting at your earliest convenience and preferably no later than 1 January 2008. Salary as overingeni?r will be according to state salary level 49-56, depending upon your qualifications. Experienced candidates may qualify for employment as senioringeni?r (senior engineer) with a salary according to levels 54-58 (NOK 394.500-424.000 p.a., approx Euro 49.000-53.000, USD 66.500-71.500). Please contact assoc. prof. Hans E. Plesser (+47-64965467, hans.ekkehard.plesser at umb.no) or prof. Gaute T. Einevoll (+47-95124536, gaute.einevoll at umb.no) for more information! Deadline for applications: Friday, 21 September 2007. Please submit your application either electronically via www.jobbnorge.no, ID code 41144, or by convential mail to Dept. of Mathematical Sciences and Technology, Norwegian University of Life Sciences, PO Box 5003, 1432 Aas, Norway. Please mark your letter with position code 07/1159. The Norwegian University of Life Sciences (UMB) is one of Norway's leading institutes in the life sciences, including biology, food and environmental science, land use and natural resource management. Through education and research, UMB shall contribute to sustaining the livelihood of present and future generations. UMB is nationally and internationally extensively involved in research, education, information and innovation. UMB offers education at all academic levels. UMB has about 2900 students and 900 staff and is located at Aas, 30 minutes south of Oslo. -- Dr. Hans Ekkehard Plesser Associate Professor Dept. of Mathematical Sciences and Technology Norwegian University of Life Sciences Phone +47 6496 5467 Fax +47 6496 5401 Email hans.ekkehard.plesser at umb.no Home http://arken.umb.no/~plesser From vroth at inf.ethz.ch Mon Aug 27 11:26:20 2007 From: vroth at inf.ethz.ch (Volker Roth) Date: Mon, 27 Aug 2007 17:26:20 +0200 Subject: Connectionists: Open Ph.D. Position at University of Basel Message-ID: <46D2ED1C.6060207@inf.ethz.ch> Apologies for cross posting and please forward to whom ever this may be of interest to. ------ Open Ph.D. Position at University of Basel ------- 1 Ph.D. position is open in the new research group "Biomedical Data Analysis" headed by Volker Roth within the department of Computer Science at University of Basel, Switzerland. The main focus of this new group concerns applications of Machine Learning in the growing area of Computational Life Sciences. Starting date: October 2007 (or later). Non-exhaustive list of possible topics: - kernel methods, (un)supervised learning and clustering - feature selection for biomarker discovery - Machine Learning approaches to survival analysis - biomedical image analysis - network inference - integration of heterogeneous data Applicants should have a master degree (or equivalent) in Computer Science, Bioinformatics, Mathematics or Physics. The ideal researcher will have a strong background on Statistics and Machine Learning, and a solid programming experience. The position is planned for ca. 3 years. The salary and conditions at University of Basel are attractive, and Basel is a very livable city. For more information, or to apply, please contact Volker Roth: vroth at inf.ethz.ch When applying, please explain why you are interested in the topic. Please also provide at least one reference. -- ================================================================= Volker Roth Institute for Computational Science Tel.: +41-44-63 23179 ETH Zurich, CAB G 81 Fax: +41-44-63 21562 Universit?tstrasse 6 email: vroth at inf.ethz.ch CH-8092 Zurich, Switzerland ================================================================= From mr287 at georgetown.edu Mon Aug 27 12:35:31 2007 From: mr287 at georgetown.edu (Maximilian Riesenhuber) Date: Mon, 27 Aug 2007 12:35:31 -0400 Subject: Connectionists: Postdoctoral Position: Exploring and exploiting the limits of fast visual recognition Message-ID: <46D2FD53.7060102@georgetown.edu> Postdoctoral Position: Exploring and exploiting the limits of fast visual recognition Riesenhuber Lab Department of Neuroscience Georgetown University Washington, DC We have an opening for a postdoctoral fellow (pending finalization of budget negotiations), starting in the late fall, to participate in a research project studying the limits of visual object recognition and cognitive processing at high image presentation rates and under dual-task conditions, with the goal of utilizing target-related neural signals recorded via high-density EEG for a real-time neurally-based target detection system combining machine and biological vision. The project thus offers the opportunity to do interesting research on visual object recognition, attentional modulations, cognitive control, and brain-machine-interfaces. A strong quantitative background and experience in neural data analysis are required. Experience with EEG and psychophysics is a strong plus, as is a background in biological and/or machine vision. The position is for an initial period of one year with the possibility of extension for an additional two years depending on progress. Salary is competitive. Due to funding restrictions, candidates should be US citizens or permanent residents. Our lab investigates the computational mechanisms underlying human object recognition as a gateway to understanding information processing and learning in cortex. In our work, we combine computational modeling with psychophysical, fMRI and most recently EEG data from our own lab and collaborators, as well as with single unit data obtained in collaboration with physiology labs. For more information, see http://maxlab.neuro.georgetown.edu. Interested candidates should send a CV, a brief (1 page) statement of research interests, representative reprints, and the names and contact information of three references by email to Maximilian Riesenhuber (mr287 at georgetown.edu). Review of applications will begin immediately, and will continue until the position is filled. Informal inquiries (to mr287 at georgetown.edu) are welcome. ********************************************************************** Maximilian Riesenhuber phone: 202-687-9198 Department of Neuroscience fax: 202-784-3562 Georgetown University Medical Center email: mr287 at georgetown.edu Research Building Room WP-12 3970 Reservoir Rd., NW Washington, DC 20007 http://maxlab.neuro.georgetown.edu ********************************************************************** From stiber at u.washington.edu Mon Aug 27 22:42:22 2007 From: stiber at u.washington.edu (Michael Stiber) Date: Mon, 27 Aug 2007 19:42:22 -0700 Subject: Connectionists: Neural Coding 2007 -- Student/Postdoc Travel Support Message-ID: <61A4D28B-7D23-4329-9246-69145459DB0E@u.washington.edu> Neural Coding 2007 -- Student/Postdoc Travel Support Funding is available for US postdoctoral and predoctoral (graduate and undergraduate) researchers to attend the 2007 Neural Coding meeting and its associated Joint US/Uruguay Workshop on Neural Dynamics, to be held in Montevideo, Uruguay on November 4-12, 2007. Support in the amount of $2000, to be used toward registration fees, hotel expenses, and other travel costs, will include post-meeting travel to research laboratories in the southern cone region of South America; contacts to research laboratories will be provided (participants will be responsible for arranging the specifics of their own travel). This will be an excellent opportunity to establish contacts with international experts and a small but very valuable and growing community of Uruguayan and Latin American scientists. The Neural Coding symposia bring together scientists from different fields with the conviction that multidisciplinary approaches are essential for better understanding neural coding mechanisms as well as their disturbances in clinical cases. Hence, the attendees of the Neural Coding Workshop should be prepared to cross the borders of their own disciplines. Intense discussions of experimental, modeling, and analytical approaches are expected. Major emphasis will be placed on biologically inspired formal and computer models which could elucidate the functionally relevant dynamics of the neural coding mechanisms, including their possible roles in causing and treating neurological and related diseases. For more information on the meeting, please see the meeting web site at . The deadline for applications is September 14. In your application, please include: 1. For postdoctoral researchers and graduate students, an abstract prepared in accordance with the instructions at the Neural Coding web site. Alternatively, you may indicate that you are a co-author of an already submitted paper (please indicate the title and author list of the paper as well). You will be expected to present your research during the Joint Workshop. 2. A personal statement describing your research interests and plans and how this experience will factor into them. Your personal statement should also indicate: your commitment to attend both the Joint Workshop and the main Neural Coding meeting, a plan for your stay in the southern cone area of South America beyond these (see the Neural Coding web site for a list of contacts), and your agreement to submit afterwards a brief summary of your experiences at the Joint Workshop and during your travels in the region, including prospective new collaborations. 3. A curriculum vitae. 4. The name and contact information (including email) for a faculty member or other scientific reference who will be sending a (separate) letter of reference for you. Please arrange for this letter to be sent directly to us. All application materials should be sent via email to . Awardees will be notified of the grant within two weeks after the deadline date. Support is provided by the US National Science Foundation under grant number OISE-0652336. From isabelle at clopinet.com Tue Aug 28 16:06:57 2007 From: isabelle at clopinet.com (Isabelle Guyon) Date: Tue, 28 Aug 2007 13:06:57 -0700 Subject: Connectionists: Results of the ALvsPK challenge Message-ID: <46D48061.4050509@clopinet.com> Results of the Agnostic Learning vs. Prior Knowledge Challenge ----------------------------------------------------------------- For the first few month of the challenge, AL lead over PK, showing that the development of good AL classifiers is considerably faster. As of March 1st 2007, PK was leading over AL on four out of five datasets. We extended the challenge five more month, but the best performances did not significant improve during that time period. On datasets not requiring real expert domain knowledge (ADA , GINA , SYLVA ), the participants entering both track obtained better results in the PK track, using a special-purpose coding of the inputs and/or the outputs, exploiting the knowledge of which features were uninformative, and using "shared weights" for redundantfeatures. For two datasets (HIVA and NOVA ) the raw data was not in a feature representation and required some domain knowledge to preprocess data. The winning data representations consist in low level features ("molecular fingerprints" and "bag of words"). From the analysis of this challenge, we conclude that agnostic learning methods are very powerful. They quickly yield (in 40 to 60 days) to performances, which are near the best achievable performances. General-purpose techniques for exploiting prior knowledge in the encoding of inputs or outputs or the design of the learning machine architecture (e.g. via shared weights) may provide an additional performance boost, but exploiting real domain knowledge is both hard and time consuming. The net result of using domain knowledge rather using than low level features and relying on agnostic learning may actually be to worsen results, as experienced by some entrants. This fact seems to be a recurrent theme in machine learning publications and the results of our challenge confirm it. Future work includes incorporating the best identified methods in our challenge toolkit CLOP . The challenge web site remains open for post-challenge submissions (http://www.agnostic.inf.ethz.ch/) For more details on the analysis, see: http://clopinet.com/isabelle/Projects/agnostic/Results.html. From bengio at idiap.ch Wed Aug 29 13:06:28 2007 From: bengio at idiap.ch (Samy Bengio) Date: Wed, 29 Aug 2007 19:06:28 +0200 (CEST) Subject: Connectionists: call for papers: NIPS*2007 Workshop on Efficient Machine Learning Message-ID: Hello, Please advertise the following CALL FOR PAPERS: NIPS*2007 Workshop on Efficient Machine Learning Overcoming Computational Bottlenecks in Machine Learning Whistler, Canada, December 7-8, 2007 http://bigml.wikispaces.com/cfp Overview -------- The ever increasing size of available data to be processed by machine learning algorithms has yielded several approaches, from online algorithms to parallel and distributed computing on multi-node clusters. Nevertheless, it is not clear how modern machine learning approaches can either cope with such parallel machineries or take into account strong constraints regarding the available time to handle training and/or test examples. This workshop will explore two alternatives: * modern machine learning approaches that can handle real time processing at train and/or at test time, under strict computational constraints (when the flow of incoming data is continuous and needs to be handled) * modern machine learning approaches that can take advantage of new commodity hardware such as multicore, GPUs, and fast networks. This two-day workshop aims to set the agenda for future advancements by fostering a discussion of new ideas and methods and by demonstrating the potential uses of readily-available solutions. It will bring together both researchers and practitioners to offer their views and experience in applying machine learning to large scale learning. Topics of Interest ------------------ * efficient parallelization of machine learning algorithms and algorithms that make use of new hardware architectures * sub-linear training algorithms for virtually infinite datasets * new online boosting, online kernel, and other efficient non-linear online training algorithms * efficient feature extraction for classification and detection * adapted structures for very large number of features per example * evolving under strict time/space constraints * coarse-to-fine and "focusing" algorithms for detection Submission Procedure -------------------- We encourage the submissions of extended abstract. The suggested abstract length is about 2 pages. The invited speakers will be allocated between 40 and 60 minutes, while the authors of the accepted abstracts will be allocated between 30 and 40 minutes to present their work (to be determined according to submissions). In addition, the abstracts will be available to a broader audience on the dedicated web site. The authors should submit their extended abstract to bigml.nips at gmail.com in pdf. An email confirming the reception of the submission will be sent by the organizers. Important Dates --------------- * Aug 28: Workshop announcement / call for abstracts * Oct 12: Abstract submission deadline * Nov 1: Notification of acceptance * Dec 7 and 8: Workshop Invited Speakers ---------------- * Yali Amit, University of Chicago * Yoshua Bengio, University of Montreal * Michael Burl, NASA JPL * Corinna Cortes, Google * Dennis DeCoste, Microsoft * Don Geman, John Hopkins University * Dan Pelleg and Elad Yom-Tov, IBM Research * Yann LeCun, New York University * Srinivasan Parthasarathy, Ohio State University * Nicol N. Schraudolph, National ICT Australia Organizers ---------- * Samy Bengio, Google * Corinna Cortes, Google * Dennis DeCoste, Microsoft Live Labs * Francois Fleuret, IDIAP Research Institute * Ramesh Natarajan, IBM T.J. Watson Research Lab * Edwin Pednault, IBM T.J. Watson Research Lab * Dan Pelleg, IBM Haifa Research Lab * Elad Yom-Tov, IBM Haifa Research Lab Other Members of the Programme Committee ---------------------------------------- * Yali Amit, University of Chicago * Gilles Blanchard, Fraunhofer Institut FIRST, Berlin * Ronan Collobert, NEC * Yves Grandvalet * IDIAP Research Institute * Jiri Matas, Czech Technical University, Prague * Sam Roweis, Google ---- Samy Bengio Research Scientist in Machine Learning. Google, 1600 Amphitheatre Pkwy, Building 47-171D, Mountain View, CA 94043, USA tel:+1 (650) 253-2563, mailto:bengio at google.com, http://bengio.abracadoudou.com From mseeger at gmail.com Fri Aug 31 05:38:47 2007 From: mseeger at gmail.com (Matthias Seeger) Date: Fri, 31 Aug 2007 11:38:47 +0200 Subject: Connectionists: NIPS 07 workshop on approximate inference for continuous/hybrid models: Call for contributions Message-ID: <43c7cd3f0708310238r74299d58i93b3bd57d34c70e9@mail.gmail.com> *** Apologies for Multiple Postings *** ================== CALL FOR CONTRIBUTIONS ================== Neural Information Processing Systems (NIPS) 2007 Workshop: Approximate Bayesian Inference in Continuous/Hybrid Models Dates: 7-8 December, 2007. Whistler, CA Organizers: Matthias Seeger, Max-Planck Biological Cybernetics, Tuebingen David Barber, University College London Neil Lawrence, University of Manchester Onno Zoeter, Microsoft Research Cambridge WWW: http://intranet.cs.man.ac.uk/ai/nips07/ E-mail: abichm at gmail.com Abstract Submission Deadline: October 21, 2007 Notification of Acceptance: November 1, 2007 Sponsored by the Pascal Network of Excellence and by Microsoft Research Cambridge ================== CALL FOR CONTRIBUTIONS ================== Abstract: The workshop will provide a forum to discuss unsolved issues, both practical and theoretical, pertaining to the application of approximate Bayesian inference in continuous variable and hybrid models. The emphasis of the workshop will be in understanding the particular difficulties in this class of models and the differential strengths and weaknesses of available deterministic (variational) approximation techniques, as opposed to the arguably better understood field of approximate inference in discrete variable systems. The target audience are practitioners, providing insight into and analysis of problems with certain methods or comparative studies of several methods, as well as theoreticians interested in characterizing the hardness of continuous distributions or proving relevant properties of an established method. We welcome contributions in the areas of Statistics (e.g. Markov Chain Monte Carlo methods), Information Geometry, Optimal Filtering, or other related fields if an effort is made of bridging the gap towards variational techniques. Format: The workshop will be single-day, comprising of a tutorial introduction, invited talks (20 to 30 mins), and presentations of contributed work, with time for discussions. Depending on quality and compatibility with workshop aims, slots for brief talks and posters will be allocated. We intend to have an interactive workshop, and will give priority to contributions of novel ideas not yet established in Machine Learning, and to critical and careful empirical comparative studies over polished applications of established methods to standard problems. We encourage the applicant to try to address some of the aims listed below or on the workshop website. We encourage contributions from related fields such as * Statistics (e.g. Markov Chain Monte Carlo methods) * Information Geometry * Filtering, Dynamical Systems if they can motivate the potential applicability to analyzing variational inference techniques. Contributions of this sort could be tutorial in nature. Contributions should be communicated to the program committee (the organizers) in form of an extended abstract (up to 8 pages in the NIPS conference paper style), sent to the mail address stated at the beginning of this mail. Submissions sent after the deadline (see beginning of mail) or violating the format constraint will not be reviewed. Motivation: Many of the most important problems in Machine Learning and related application areas are most naturally and succinctly treated using continuous variable models. Several important continuous latent variable models come to mind, each underlying a host of applications: - Gaussian Process Models with non Gaussian likelihoods - Sparse Linear Models (Bayesian ICA, Relevance Vector Machine, Sparse Image Coding, Non-negative Image Coding, Compressed Sensing) - Dynamical Systems (Filters, Smoothers, Tracking, Switching Models, Latent State Space Models) Several variational inference approximations have been applied successfully to continuous models. However, most of these techniques originated in Statistical Physics or Information Theory, where systems of interest typically consist of discrete variables throughout, and theoretical analyzes, convergence, or performance guarantees are predominantly available for the discrete case and focus primarily on discrete attributes such as graph topology. Properties of variational approximations are much less well understood when applied to continuous models. In some applications (such as gene or metabolic network identification), continuous variables are artificially discretized in order to allow the use of better understood discrete techniques. In others, Monte Carlo techniques predominate. In both cases, potential users of variational methods are probably deterred by the general lack of theoretical understanding available. In many applications (for example the models listed above), variational methods are the state of the art today. However, as opposed to the situation with discrete models, we still lack adequate understanding of which characteristics of a given realistic model make accurate variational inference simple or hard, whether such properties are transferable or specific to certain methods, and which empirical signatures allow such difficulties to be detected in a given method. Continuous models give rise to difficulties not present in discrete ones. On the one hand, the Gaussian family allows efficient Bayesian computations with many variables, even if no independence structure is present. However, non-Gaussian continuous families are usually not closed under marginalization, an essential operation for any message passing scheme, so that projections become necessary. Also, intractable integrals often require additional convex bounding or numerical quadrature. Both steps introduce errors typically not present in discrete variable schemes. Target Audience: We welcome participants to share their experiences on practical problems. However, in contrast to the usual `success stories' for an established method, we invite descriptions of practical difficulties in applying approximate inference methods, and how these were analyzed and dealt with in the application in a principled manner. While the goal is to understand deterministic (variational) approximations better, we welcome contributions of researchers working on Monte Carlo approximate inference if an effort is made towards bridging the gap between the fields. In this context, we encourage contributions of tutorial nature or of preliminary ideas. The workshop is therefore intended to appeal both to practitioners with insight into the difficulties in approximate inference in continuous systems, and to theorists with an interest in characterizing the complexity of posterior distributions or in analyzing properties of approximate inference methods. Aims: The aim of the workshop is to study deterministic approximation methods and characterizations of inference complexity in continuous and hybrid systems. Specifically, several important practical open issues are: * Variational mean field Bayes methods are very frequently used, due to their generic derivation, ease of implementation, and numerical stability. However, growing evidence suggests that the methods may in practice be often severely biased, giving rise to adverse effects such as over-pruning of parameters. Why is this the case, and can it be improved upon? * Learning hyperparameters requires inference as a subroutine. Which estimation biases do different methods imply? Can they be corrected? * When and why might Expectation Propagation (EP) be superior to Variational mean field Bayes? * Continuous families are not closed under marginalization, and projections (moment matching, variational KL minimization) are often needed. What properties do different projections have, and how do they affect the final solution? * EP sometimes has severe numerical stability issues. For which models can its convergence be guaranteed? Do numerical problems reveal the hardness of a problem, or do they arise from specific shortcomings of the method? * Which methods are numerically (in)stable, and on which problems? Are the difficulties inherent, or can the methods be stabilized? How do stable methods behave on problems where others are inherently instable? * Is posterior multi-modality the only property that makes a problem hard? Which practically relevant properties of a non-Gaussian distribution render its approximation by a Gaussian difficult? * Markov chain Monte Carlo (MCMC) is provably efficient for log-concave posterior distributions. What is the role of log-concavity in current variational methods (many Gaussian Process and Sparse Linear Models are log-concave)? * Most methods make use of ideas coming from other communities (numerical quadrature, convex duality, scale mixtures). How have approximation errors been quantified there, and can we transfer these ideas? ================== CALL FOR CONTRIBUTIONS ================== WWW: http://intranet.cs.man.ac.uk/ai/nips07/ E-mail: abichm at gmail.com Abstract Submission Deadline: October 21, 2007 Notification of Acceptance: November 1, 2007 From gunnar.raetsch at tuebingen.mpg.de Fri Aug 31 10:04:27 2007 From: gunnar.raetsch at tuebingen.mpg.de (=?ISO-8859-1?Q?Gunnar_R=E4tsch?=) Date: Fri, 31 Aug 2007 16:04:27 +0200 Subject: Connectionists: CFP: New Problems and Methods in Computational Biology Message-ID: Dear colleagues, I would like to invite you to participate in the workshop on New Problems and Methods in Computational Biology http://www.mlcb.org on the 7th or 8th of December at NIPS'07 in Whistler, B.C. (http:// nips.cc). If you would like to contribute then please send an extended abstract by *October 15, 11:59am (Samoa time)* to nips-compbio at tuebingen.mpg.de (details below). I am looking forward to meet you there! Gunnar Raetsch New Problems and Methods in Computational Biology http://www.mlcb.org A workshop at the Twenty-First Annual Conference on Neural Information Processing Systems (NIPS 2007) Whistler, BC, Canada, December 7-8, 2007. Deadline for submission of extended abstracts: October 15, 2007 WORKSHOP DESCRIPTION The field of computational biology has seen dramatic growth over the past few years, in terms of newly available data, new scientific questions and new challenges for learning and inference. In particular, biological data is often relationally structured and highly diverse, and thus requires combining multiple weak evidence from heterogeneous sources. These sources include sequenced genomes of a variety of organisms, gene expression data from multiple technologies, protein sequence and 3D structural data, protein interaction data, gene ontology and pathway databases, genetic variation data (such as SNPs), and an enormous amount of text data in the biological and medical literature. These new types of scientific and clinical problems require novel supervised and unsupervised learning approaches that can use these growing resources. The workshop will host presentations of emerging problems and machine learning techniques in computational biology. We encourage contributions describing either progress on new bioinformatics problems or work on established problems using methods that are substantially different from standard approaches. Kernel methods, graphical models, feature selection and other techniques applied to relevant bioinformatics problems would all be appropriate for the workshop. SUBMISSION INSTRUCTIONS Researchers interested in contributing should send an extended abstract of 1-6 pages in PDF format to nips-compbio at tuebingen.mpg.de by October 15, 2007, 11:59pm (Samoa time). No special style is required. Authors may use the NIPS style file, but are also free to use other styles as long as they use standard font size (11-12 pt) and margins. All submissions will be anonymously peer reviewed and will be evaluated on the basis of their technical content. A strong submission to the workshop typically presents a new learning method that yields new biological insights, or applies an existing learning method to a new biological problem. However, submissions that improve upon existing methods for solving previously studied problems will also be considered. Please note that accepted abstracts will be posted at http:// www.mlcb.org. Authors may submit two versions of their abstract, a longer version for review and a shorter version for posting to the web page. The workshop allows submissions of papers that are under review or have been recently published in a conference or a journal. This is done to encourage presentation of mature research projects that are interesting to the community. The authors should clearly state any overlapping published work at time of submission. Authors of accepted abstracts will be invited to submit full length versions of their contributions for publication in a special issue of BMC Bioinformatics. ORGANIZERS Gal Chechik, Department of Computer Science, Stanford University Christina Leslie, Memorial Sloan-Kettering Cancer Center William Stafford Noble, Department of Genome Sciences, University of Washington Gunnar Raetsch, Friedrich Miescher Laboratory of the Max Planck Society (Tuebingen, Germany) Quaid Morris, Terrence Donnelley Centre for Cellular and Biomolecular Research, University of Toronto Koji Tsuda, Max Planck Institute for biological Cybernetics (Tuebingen, Germany) PROGRAM COMMITTEE Pierre Baldi, UC Irvine Kristin Bennett, Rensselaer Polytechnic Institute Mathieu Blanchette, McGill University Florence d'Alche, Universite d'Evry-Val d'Essonne, Genopole Eleazar Eskin, UCLA Brendan Frey, University of Toronto Nir Friedman, The Hebrew University of Jerusalem Michael I. Jordan, UC Berkeley Alexander Hartemink, Duke University Michal Linial, The Hebrew University of Jerusalem Klaus-Robert Mueller, Fraunhofer FIRST Uwe Ohler, Duke University Alexander Schliep, Max Planck Institute for Molecular Genetics Bernhard Schoelkopf, Max Planck Institute for Biological Cybernetics Eran Segal, The Weizmann Institute Jean-Philippe Vert, Ecole des Mines de Paris +-------------------------------------------------------------------+ Gunnar R?tsch http://www.fml.mpg.de/raetsch Friedrich Miescher Laboratory Gunnar.Raetsch at tuebingen.mpg.de Max Planck Society Tel: (+49) 7071 601 820 Spemannstra?e 39, 72076 T?bingen, Germany Fax: (+49) 7071 601 801 From jpineau at cs.mcgill.ca Fri Aug 31 14:08:25 2007 From: jpineau at cs.mcgill.ca (Joelle Pineau) Date: Fri, 31 Aug 2007 14:08:25 -0400 (EDT) Subject: Connectionists: Post-doc position in RL at McGill Message-ID: <48301.132.206.3.38.1188583705.squirrel@mail.cs.mcgill.ca> ******************************************************* Post-Doctoral Fellowship in Reinforcement learning ******************************************************* Description: The Reasoning and Learning Laboratory (rl.cs.mcgill.ca) at McGill University is inviting applications for a one to two year Post-Doctoral Fellow position in the area of Statistical Machine Learning, with an emphasis on Reinforcement Learning. The candidate should be interested in investigating the development of new representations and algorithms for learning in dynamic systems with hidden state information, as well as applications of these techniques to problems in automated treatment design for chronic diseases. Qualifications: Candidates must hold a recent Ph.D. in computer science, statistics, mathematics, engineering or a related field. Strong mathematical, statistical and computational skills are required. Substantial evidence of expertise in the area of reinforcement learning is required. Multi-disciplinary expertise in cognitive science, neuroscience, or bioinformatics is a plus. Application procedure: Interested candidates should email a copy of their CV, research statement, list of publications, and names of 3 references to Joelle Pineau (jpineau at cs.mcgill.ca). Start date: September 2007 (or later) Location: Montreal, Canada From ps629 at columbia.edu Fri Aug 31 20:10:00 2007 From: ps629 at columbia.edu (Paul Sajda) Date: Fri, 31 Aug 2007 20:10:00 -0400 Subject: Connectionists: Position for Postdoctoral Fellow: Cortically-coupled Computer Vision Message-ID: Position for Postdoctoral Fellow: Cortically-coupled Computer Vision The Laboratory for Intelligent Imaging and Neural Computing (LIINC) at Columbia University has an immediate opening for a Postdoctoral Fellow to participate in our research program in "Cortically-coupled Computer Vision (C3Vision)". The C3Vision program looks to synergistically couple biological and computer vision systems using a combination of brain machine interfaces, machine learning and pattern classification, and image understanding within the context of understanding the advantages and limits of both biological and computer vision. Applicants should have a background in one, and preferably several, of the following: machine vision (especially content based indexing and automated image labeling), machine learning, neural signal processing, neuroimaging (EEG and/or fMRI), real-time systems design and programming. LIINC is in the Department of Biomedical Engineering at Columbia University and interacts closely with other departments at Columbia, Including Electrical Engineering, Biological Sciences, Computer Science and Neuroscience. In addition, the C3Vision project includes collaborators at other academic institutions as well as in industry, and the project involves both basic and applied research which will ultimately lead to testable systems. Interested candidates should send via email their CV, three representative papers, the names of three references, and cover letter to Prof. Paul Sajda (ps629 at columbia.edu). Applications will be considered immediately. The position is for one year, with the option to renew for 2-3 years, given satisfactory performance and available funding. Paul Sajda, Ph.D. Associate Professor Department of Biomedical Engineering Columbia University 351 Engineering Terrace Building, Mail Code 8904 1210 Amsterdam Avenue New York, NY 10027 tel: (212) 854-5279 fax: (212) 854-8725 email: ps629 at columbia.edu http://liinc.bme.columbia.edu