Last application date: Aug 18, 2019 12:00
Department: TW07 - Department of Telecommunications and information processing
Employment category: Doctoral fellow
Contract: Limited duration
Degree: Master of Science degree (at the start of the PhD) and a background in Electrical Engineering or Computer Science
Occupancy rate: 100%
Vacancy Type: Research staff
For the project “Realtime and Power-efficient AI”, we are looking for a motivated early stage researcher in the fields of embedded AI, computer vision and deep learning. The researcher fellow will be hosted at the Image Processing and Interpretation (IPI) research group of Ghent University's Faculty of Engineering and Architecture for an initial period of 12 months, to be renewed up to 4 years with the aim of obtaining a PhD.
In several industrial applications (e.g., artificial intelligence in the edge, simultaneous localization and mapping…), local early sensor fusion and distributed cooperative fusion are of big importance. Early fusion techniques often involve custom processing steps, such as point cloud to depth map conversion steps, projections of 3D data onto 2D images (and back, during backpropagation) which are not covered by standard neural network libraries and which often use radically different paradigms (e.g., differentiable programming) or involve different resource requirements (e.g., dynamically varying in time). For this purpose, software mapping tools have to be designed to integrate such algorithms into a heterogeneous platform.
The PhD researcher will work on 1) software mapping tools to program and compile traditional deep learning and novel sensor fusion algorithms onto a heterogeneous platform with multi-core CPU, GPU and AI accelerator (which will be developed simultaneously by the project partners) and 2) a performance estimation framework for the platform. The performance estimation framework allows rapid assessment of the cost of running different machine learning workloads and sensor fusion algorithms on AI accelerators. This allows algorithm designers to assess the resource cost of various algorithmic alternatives they have, without requiring their complete development and mapping.
IPI has a long standing expertise in multi-camera, AI and real-time computer vision and has several projects in this area. The new researcher will therefore work as a team with existing researchers. They will also cooperate with the other researchers in the “Realtime and Power-efficient AI” project.
You will receive a PhD scholarship according to the general conditions at Ghent University. The tax free scholarship includes full social security coverage (net monthly amount starting at ± 1.900 EUR/month, the exact amount depending on family status and experience). The initial contract will be for a period of 1 year; this contract can be extended up to a total of 4 years, subject to good performance. You will work at the Image Processing and Interpretation research group, which is also a core research group of imec. Information about IPI can be found on the web: http://ipi.ugent.be
Please submit your application by email to Prof. Bart Goossens (Bart.Goossens@UGent.be) and Prof. Wilfried Philips (Wilfried.Philips@UGent.be). In your email, please include the following:
These documents need not be on the topic of the positions.Läs mer
|Titel||PhD Student - Department of Telecommunications and information processing|
|Job location||Sint-Pietersnieuwstraat 33, 9000 Gent|
|Publicerad||juli 15, 2019|
|Sista ansökningsdatum||augusti 18, 2019|
|Ämnen||Kommunikationsteknik,   Programmeringsspråk,   Telekommunikationsteknik,   Tillämpad matematik,   Sannolikhetsteori,   Elektroteknik,   Beräkningsmatematik,   Maskininlärning,   Datorseende,    and 2 more. Signalbehandling,   Bildbehandling  |