From schneide at cs.cmu.edu Mon Apr 2 17:19:39 2007 From: schneide at cs.cmu.edu (Jeff Schneider) Date: Mon, 02 Apr 2007 17:19:39 -0400 Subject: [Research] Reminder - Thesis Proposal - Kaustav Das 4/3/07 Message-ID: <4611736B.9070003@cs.cmu.edu> Hi Everyone, Please come to Kaustav's thesis proposal Tuesday morning at 10. Jeff. -------- Original Message -------- Date: 4/3/07 Time: 10:00AM Place: 1507 Newell-Simon Hall Title: Detecting Anomalous Records in Large Categorical Datasets Speaker: Kaustav Das, PhD candidate Advisor: Jeff Schneider Abstract: We consider the problem of detecting anomalies in high dimensional categorical datasets. In most applications, anomalies are defined as data points that are 'abnormal'. Quite often we have access to data which consists mostly of normal records, along with a small percentage of unlabelled anomalous records. We are interested in the problem of unsupervised anomaly detection, where we use the unlabelled data for training, and detect records that do not follow the definition of normality. A standard approach is to create a model of normal data, and compare test records against it. A probabilistic approach builds a likelihood model from the training data. Records are tested for anomalies based on the complete record likelihood given the probability model. For categorical attributes, bayes nets give a standard representation of the likelihood. While this approach is good at finding outliers in the dataset, it often tends to detect records with attribute values that are rare. Sometimes, just detecting rare values of an attribute is not desired and such outliers are not considered as anomalies in that context. In this thesis we present an alternative definition of anomalies, and propose an approach of comparing against marginal distribution of attribute subsets. We show that this is a more meaningful way of detecting anomalies, and has a better performance over semi-synthetic as well as real world datasets. We propose to extend this method to detecting anomalous groups of records. We also propose to incorporate user feedback in a semi-supervised learning framework. Committee: Jeff Schneider (Chair) Gregory Cooper (University of Pittsburgh) Geoffrey Gordon Christos Faloutsos -- ******************************************************************* Diane Stidle Business & Graduate Programs Manager Machine Learning Department School of Computer Science 4612 Wean Hall Carnegie Mellon University 5000 Forbes Avenue Pittsburgh, PA 15213-3891 Phone: 412-268-1299 Fax: 412-268-3431 Email: diane at cs.cmu.edu URL:http://www.ml.cmu.edu From mjbaysek at cs.cmu.edu Thu Apr 5 11:40:25 2007 From: mjbaysek at cs.cmu.edu (Michael J. Baysek) Date: Thu, 05 Apr 2007 11:40:25 -0400 Subject: [Research] IMPORTANT: Run Windows Update Right Away In-Reply-To: <000101c7778b$774aaa90$c6bf0280@fac.cs.cmu.edu> References: <000101c7778b$774aaa90$c6bf0280@fac.cs.cmu.edu> Message-ID: <46151869.7000806@cs.cmu.edu> Hi Lab & friends, You may have received notice from facilities (or friend) to run Windows Update in response to MS07-017. This vulnerability is particularly bad. There are already a number of exploits in use by spammers and malware for this problem. I need to stress to you, Run Windows Update Right Away. [insert typical anti-MS rant here] Keeping in line with their historical promptness for fixing root exploits, Microsoft knew about the problem since December. It seems it takes four months and billions of dollars to patch a vulnerability in code that processes an animated mouse cursor. Michael J. Baysek, Systems Analyst Carnegie Mellon University - Auton Lab www.cmu.edu - www.autonlab.org 412-268-8939 Help Desk wrote, On 04/05/07 10:05: > April 5, 2007 > > Microsoft released a critical Windows security update on April 3, 2007. > The security update addresses a vulnerability with animated cursor files, > which is being actively exploited and is considered by Microsoft and the > security community to be critical enough to warrant this out of cycle patch. > > * All Windows PCs must be patched. * > > To patch your PC: > > Click on the update icon on the taskbar (it is a globe or yellow shield with > an exclamation point, depending on which version of Windows you're running). > > If you do not have such an icon, run Windows Update by going to the site: > > http://windowsupdate.microsoft.com > > After patching, you must reboot your PC in order for the patch to take > effect. > > Additional information about the vulnerabilities addressed in this patch can > be found at: > > http://www.microsoft.com/technet/security/Bulletin/MS07-017.mspx > > Microsoft and others are reporting some problems caused by the patch, though > Facilities has seen no problems with supported applications. > More information on known problems with this patch can be found at: > > http://support.microsoft.com/kb/935448 > > If you have any questions or problems applying these patches, please contact > the SCS Help Desk at x8-4231 or send mail to help at cs.cmu.edu. > > Thank you for your attention, > > SCS Help Desk > > > > > -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/x-pkcs7-signature Size: 3245 bytes Desc: S/MIME Cryptographic Signature URL: From kristens at cs.cmu.edu Thu Apr 12 16:17:14 2007 From: kristens at cs.cmu.edu (Kristen Schrauder) Date: Thu, 12 Apr 2007 16:17:14 -0400 Subject: [Research] CMU Car Insurance Message-ID: <008f01c77d3f$9216fc20$fadc0280@adm.ri.cmu.edu> For anyone who rents a car when traveling for the lab, attached is the current insurance card for this year. CMU has been refusing to reimburse people who elect to take the insurance from the rental company so be sure to carry this with you when traveling. I will be keeping hard copies of this in my office so feel free to pick one up from me anytime before you travel. Thanks! Kristen Schrauder, Administrative Coordinator Carnegie Mellon University - Robotics Institute Newell-Simon Hall, Room 3128 5000 Forbes Avenue Pittsburgh, PA 15213 Phone: 412.268.7551 Fax: 412.268.7350 -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: 06 - 07 PA Auto Financial Responsibility Card.pdf Type: application/pdf Size: 29879 bytes Desc: not available URL: From awm at google.com Fri Apr 27 10:31:12 2007 From: awm at google.com (Andrew W. Moore) Date: Fri, 27 Apr 2007 10:31:12 -0400 Subject: [Research] Fwd: Internship / job opportunities in Pittsburgh In-Reply-To: References: Message-ID: <982f89350704270731g406ec449gf2c812003f3ec7c9@mail.gmail.com> Apologies if redundant information... ---------- Forwarded message ---------- From: Henry Schneiderman Date: Apr 26, 2007 4:23 PM Subject: Internship / job opportunities in Pittsburgh To: vasc-seminar at cs.cmu.edu Available Positions at Pittsburgh Pattern Recognition ------------------------------------------------------------------------ Summer Internship Pittsburgh Pattern Recognition has an opening for CS / ECE intern for the summer of 2007. The intern will work closely with staff members on various projects involving automatic video processing software based on our face detection and recognition technology. Specific work may include visualization, multi-platform graphical user interfaces development, video I/O, and external device I/O. Applicant must have good software engineering skills. Background in GUI design, computer hardware, computer vision is also desirable. Intern will be paid a weekly salary of $600. Software Engineer Pittsburgh Pattern Recognition seeks a software engineer to develop products for both commercial and government applications. This position requires strong software engineering skills, familiarity with computer hardware, and familiarity with computer vision / image processing methods. Ideal candidate must demonstrate a strong academic record, a pragmatic approach to problem solving, attention to detail, proven ability to work successfully on group projects, and a strong desire to learn new skills. Applicants should hold a B.S. or M.S. in Electrical Engineering, Computer Engineering, Computer Science, or a related field. Salary ranges from $60,000 - $75,000 commensurate with background and experience, plus full health and retirement benefits. Computer Vision Scientist Pittsburgh Pattern Recognition seeks a researcher to perform focussed algorithm development on problems in object and face recognition. This position requires a thorough understanding of the fundamental concepts in applied mathematics, probability theory, signal processing, and machine learning. Ideal candidate must demonstrate a strong academic record, a pragmatic approach to problem solving, attention to detail, proven ability to work successfully on group projects, and a strong desire to learn new skills. Applicants should hold an advanced degree in Electrical Engineering, Computer Engineering, Computer Science, or a related field. Salary commensurate with background and experience, plus full health and retirement benefits. Pittsburgh Pattern Recognition is an EOE employer. All job applications are maintained on file for two years from the date received. To apply, please e-mail a cover letter and resume to careers at pittpatt.com. About Pittsburgh Pattern Recognition: Pittsburgh Pattern Recognition (PittPatt) develops software for interpreting digital photographs and video streams. Its Face Detection and Tracking Software Development Kit (FT-SDK) can reliably locate human faces in photographs and track the motion of human faces in video. Face detection and tracking has applications in security, video compression, automated photograph annotation, video navigation, people counting, photofinishing and video-based security. PittPatt's technology is widely considered to be the world leader in object detection and tracking, placing first in both the U.S. Intelligence Community's 2005 ARDA evaluation and a similar international evaluation, CLEAR, in 2006. This technology has also been exhibited at a number of U.S. science museums and appeared Wired Magazines's NextFest 2004. PittPatt was founded in 2004 as a spin-off from Carnegie Mellon University's Robotics Institute. See http://www.pittpatt.com for more information. -------------- next part -------------- An HTML attachment was scrubbed... URL: From awd at cs.cmu.edu Mon Apr 30 12:13:16 2007 From: awd at cs.cmu.edu (Artur Dubrawski) Date: Mon, 30 Apr 2007 12:13:16 -0400 Subject: [Research] After a long break, Auton Lab meetings are back! Message-ID: <1177949596.1722.42.camel@localhost> Here is an announcement of the next, very exciting one: Time & Place: Tuesday, May 15th, at 11:30am, NSH 1507 Food: Yes Speaker: Brent Bryan Title & Abstract: Efficiently Computing Minimax Expected-Size Confidence Regions Given observed data and a collection of parameterized candidate models, a 1-alpha confidence region in parameter space provides useful insight as to those models which are a good fit to the data, all while keeping the probability of incorrect exclusion below alpha. With complex models, optimally precise procedures (those with small expected size) are, in practice, difficult to derive; one solution is the Minimax Expected-Size (MES) confidence procedure. The key computational problem of MES is computing a minimax equilibria to a certain zero-sum game. We show that this game is convex with bilinear payoffs, allowing us to apply any convex game solver, including linear programming. Exploiting the sparsity of the matrix, along with using fast linear programming software, allows us to compute approximate minimax expected-size confidence regions orders of magnitude faster than previously published methods. We test these approaches by estimating parameters for a cosmological model. A copy of the paper can be found on the autonlab.org website.