Connectionists: Vacancy: PhD or Postdoc position - Designing general-purpose analog computing based on machine learning

[Joni Dambre]

Imec-IDLab-UGent

IDLab is a research group of UGent, as well as a core research group of imec.  IDLab performs fundamental and applied research on data science and internet technology, and counts over 300 researchers.  Our major research areas are machine learning and data mining; semantic intelligence; multimedia processing; distributed intelligence for IoT; cloud and big data infrastructures; wireless and fixed networking; electromagnetics, RF and high-speed circuits and systems. In particular, the group has been studying various kinds of neural networks for more than 20 years. It has been at the forefront of deep learning research ever since it became popular a decade ago. Illustrative of this success are an excellent track record at Kaggle competitions and the fact that four of our former PhD students are now working at Google’s Deepmind and one at Google Brain.

The job

In the context of the European H2020 project PHRESCO, we offer a PhD or postdoc position at the intersection of machine learning, photonics and dynamical systems theory.

The PHRESCO  research project involves the design of integrated photonic analog computers based on a machine learning technique called Reservoir Computing.  For this technique, a physical nonlinear dynamical system (the reservoir) is considered to be computing a number of functions of its past and present inputs in its state variables. To perform useful computation with such a system, these functions can be linearly combined to approximate a desired input/output behavior. The overall nature of the state functions is generally affected by some global parameters which can be tuned to best accommodate the desired functionality.

The resulting modules offer tunable general-purpose analog computing, but they can not be scaled up to solve complex tasks. In order to go further with this, the next step is to develop a systematic design methodology that allows to automatically extract/design a multi-reservoir system from a task description that is specified by examples, like in a typical machine learning flow. Although the project addresses photonic implementations, the design methodology should be suitable for any analog implementation medium. This research is situated at the intersection of machine learning, analog hardware design, nonlinear dynamical systems and should also build upon the vast experience that exists related to system design (e.g., in the digital community).

 
Profile:

We are looking for an excellent PhD student or postdoc with a keen interest in exploring alternatives to traditional digital computation. Candidates should ideally have experience with machine learning. A background in hardware design or the development of a hardware design methodology is an asset. A background in photonics is not necessary, but some knowledge is helpful. Also, candidates should be very creative in combining ideas from different domains.

We offer an exciting job in a stimulating environment, with a nice amount of flexibility and academic liberty. Your research will be supported by our own research group, with a thorough expertise in machine learning and (photonic) reservoir computing, as well as the photonics research group in the INTEC department (group of prof. P. Bienstman, also at Ghent University - imec), which mainly addresses the technological implementation of such systems.

Requirements: 

Applicants should have a Masters degree (for a PhD position) or PhD (for a postdoc position), preferably in computer science or electronics engineering or computational/cognitive neuroscience. Note: to be admissible to the PhD-program, your degree must be equivalent to 5 years of engineering studies (bachelor + master) in the European Union.
You have an excellent academic track record (graduation cum laude or grades in the top 25% percentile).
You have a background in machine learning.
You can illustrate  you interdisciplinary mindset by previous interdisciplinary projects or by combining a background in as many as possible of the following fields: neural networks, analog or digital hardware design methodology, optimization theory, computational or cognitive neuroscience, signal processing, nonlinear control theory, compressed sensing, stochastic processes, applied physics, photonic/optical computing.
You are interested in and motivated by the research topic, as well as in obtaining a PhD degree.
You have excellent analytical skills.
You are a native Dutch speaker or speak and write English fluently (C1 CEFR level).
You have good communication skills.
You are eligible for a Flemish PhD scholarship (i.e., you have not enjoyed such a scholarship before).
You are preferably available from January 1st or shortly after.
 
Our offer:

We offer a fully funded PhD scholarship for a maximal period of 4 years (upon positive progress evaluation). The PhD research is fundamental and innovative, but with clear practical applications. You will join a young and enthusiastic team of researchers, post-docs and professors. This PhD position is available from January 1st, 2018.

Interested?

Apply with extensive motivation letter, scientific resume, diplomas and detailed academic results (courses and grades), English proficiency scores (for non Dutch speakers), relevant publications, and two reference contacts.

For any questions, contact prof. dr. ir. Joni Dambre <mailto:Joni.Dambre at UGent.be?subject=Vacancy:%20Designing%20general-purpose%20analog%20computing%20%20based%20on%20machine%20learning>.

After the first screening, suitable candidates will be invited for an interview (also possible via Skype) and may get an assignment for skills assessment.





prof. dr. ir. Joni Dambre,

IDLab, Ghent University - imec
Electronics and Information Systems Department, Engineering Faculty
iGent building,
Technologiepark 15, 9052 Zwijnaarde, Belgium
work phone: +32-9-264.34.09
email: Joni.Dambre at UGent.be



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20171127/1964ecfd/attachment.html>