PhD candidate in Efficient Deep Learning
Faculty of Science – Informatics Institute
- Publication date
- 24 July 2018
- Level of education
- Salary indication
- €2,266 to €2,897 gross per month
- Closing date
- 21 August 2018
- 38 hours per week
- Vacancy number
The Informatics Institute of the Faculty of Science invites applications for a PhD candidate in Efficient Deep Learning for four years. The position is part of the Dutch STW Efficient Deep Learning program.
The Efficient Deep Learning (EDL) program
The EDL program combines the fields of machine learning and computing: both disciplines are already strong in the Netherlands and now connected by seven Dutch academic institutes and more than 35 other (industrial) partners in- and outside the Netherlands. The EDL program contains seven use case driven EDL research projects: P1) DL as a service, P2) Reconstruction, matching and recognition, P3) Video analyses and surveillance, P4) High tech systems and materials, P5) Human and animal health, P6) Mobile robotics, and P7) DL platforms. The common goal for all seven EDL projects is to significantly improve the applicability of DL among others by creating data efficient training, and tremendously improving computational efficiency, both for training and inference.
Partners in EDL-P4 are the Dutch universities TU/e, UvA and VU, and also Thermo Fischer, Qualcomm (Scyfer), the Netherlands eScience center, ASTRON and SURFsara.
The PhD position
The candidate will work on DL for extreme-scale system health management. In modern radio telescopes, System Health Management (SHM) is crucial for detection and remedying of errors. Scale and complexity of the systems involved are increasing: we must deal with ~100K distributed sensors, huge data volumes and real-time compute requirements (petascale to exascale). The effectiveness and efficiency of current SHM approaches are limited in this context. We will develop semi-supervised and active learning SHM approaches that perform data fusion to combine sensor time series data with system health data (network, HDD, memory, software, logs), deal with the lack of labelled data, and unbalanced and changing labels. This is beneficial for maintenance, operations, and costs, but also crucial for the scientific results, as accurate knowledge of the instruments is essential for calibration.
Although we focus concretely on application of this technology in radio astronomy, similar problems arise in high-energy physics, ecology, life sciences and urban planning (e.g., LHC, Xenon3T, KM3Net). Similar problems occur in large-scale distributed simulations, e.g. in water management, computational chemistry and climate science. Therefore, we will develop generic methods and software.
The candidate will work on the development of energy and resource efficient DL algorithms. She or he will work on implementations to accelerate DL by programmable hardware (e.g. GPUs).
- conduct interdisciplinary research with a special focus on the design and implementation of novel scalable DL algorithms that deal with extremely large, heterogeneous and sparse data in real time;
- investigate novel architectures for real-time efficiency and power efficiency at exascale;
- implement and validate DL-based SHM in radio telescopes;
- publish in top tier conferences and journals in relevant areas;
- provide teaching assistance (10%);
- complete a PhD thesis.
The PhD candidate will work at the Systems and Networking lab of the Informatics Institute at the University of Amsterdam and at ASTRON (Dwingeloo). Regular travelling between the two sites is expected.
We are looking for candidates whose skills match with (a large part of) the following profile:
- an MSc in Computer Science, AI or related disciplines with excellent grades;
- demonstratable experience with High-Performance Computing (parallel programming, networking, GPUs);
- affinity with Signal Processing, Machine Learning (ideally deep learning), and a good mathematical background are important;
- excellent programming skills (in C, C++, Python);
- a team player who enjoys a multicultural and interdisciplinary environment in which academic-industrial collaboration is central;
- good communication and organization skills;
- excellent English language skills (writing and presenting).
For further information about this vacancy you can contact:
The appointment will be full-time (38 hours a week) for a period of four years (initial employment is 18 months). Periodic evaluations will be held after 9 and 14 months. Upon positive evaluation, the appointment will be extended to a total of 48 months. The appointment must lead to a dissertation (PhD thesis). An educational plan that includes attendance of courses, summer and/or winter schools, and national and international meetings will be drafted for the PhD candidate. The PhD candidate is also expected to assist in teaching of undergraduate students.
The salary is in accordance with the university regulations for academic personnel. The salary will range from €2,266 (first year) up to a maximum of €2,897 (last year) gross per month (scale P) based on a full-time appointment. There are also secondary benefits, such as 8% holiday allowance per year and the end of year allowance of 8.3%. The Collective Labour Agreement for Dutch Universities is applicable.
Among other things, we offer:
- competitive pay and excellent benefits;
- extremely friendly working environment;
- high-level of interaction;
- international environment (10+ nationalities in the group);
- access to high-end computing facilities (cluster with 4,000+ cores);
- one of the best AI ecosystems on the planet;
- new building located near the city center (10 minutes by bicycle) of one of Europe’s most beautiful and lively cities;
- access to one of the largest radio telescopes in the world (LOFAR) in a beautiful environment (a national park).
Since Amsterdam is a very international city where almost everybody speaks and understands English, candidates need not be afraid of the language barrier.
The UvA is an equal opportunity employer. We celebrate diversity and are committed to creating an inclusive environment for all employees. We value a spirit of enquiry and endurance, provide the space to keep asking questions and cherish a diverse atmosphere of curiosity and creativity.
You may only submit your application online using the link below. We will accept applications until 21 August 2018. To process your application immediately, please quote the vacancy number 18-453.
The committee does not guarantee that late or incomplete applications will be considered.
In the research statement (see below) you should clearly describe your research vision in relation to the topics that we described, as well as a clear statement which would be your direction of research.
You should also include the following information, in separate PDF files (not zipped), using surname, initials and a self-evident word as file names, e.g., Smith J CV:
- a curriculum vitae (including a URL allowing download of pdf of MSc. thesis - if relevant);
- a letter of motivation (at most 1 page) explaining why you are interested in this position;
- a research statement (at most 2 pages), explaining your research interests and how you think they can be related to the topics mentioned in the Job description above;
- a list of all university courses taken, including a transcript of grades;
- the name and contact details (including email address) of three referees who can provide details about your profile (one of whom should be the main supervisor of your Master thesis). #LI-DNP
No agencies please