Antwerpen | More than two weeks ago
While the success of AI in the last decade, was originally made possible by advances in hardware and compute power, the cost and complexity of the popular AI algorithms is constantly growing in a non sustainable manner. With more complex neural network architectures and applications arising each month, the computational footprint of novel state of the art neural networks is growing with a factor 100 every two years, far surpassing. Moore’s law. As such, the pace of traditional hardware development cannot match and will become a bottleneck for the field of AI to further advance. There is thus a need for the adoption novel compute paradigms, AI accelerators and hardware software co-design efforts.
For instance, many real-world problems contain for (potentially a large part) uncertainty and incomplete information, while also need to cope with limited resource, and real-time demands. Towards uncertainty, probabilistic computing is a field of interest as the technique aims to incorporate these uncertainties in its models and make probabilistic predictions or decisions [1, 2]. Techniques of interest, are among others, Bayesian networks and Markov models. To cope with resource constraints, we can also explore other paradigms like neuromorphic computing and low power sensor processing. The exact topic can depend on the interests of the candidate.
In this research topic we want to investigate the impact of novel compute paradigms and novel hardware implementations on the algorithmic level. A first step will be to research different emulation strategies of those hardware paradigms in software as a starting point to benchmark different architectures and algorithms. Later, we can study the impact at the different layers of different algorithms or AI models. The learnings can potentially be transferred to other hardware efficient and edge AI algorithms or paradigms in a later stage.
We offer you a challenging, stimulating and pleasant research environment, where you can contribute to international research on artificial intelligence with a close link to the underlaying hardware. While you will work in the AI & Data department, you will also be working together with imec hardware and university teams to jointly come up with a full stack solution.
Our ideal candidate for this position has the following skills:
References:
[1] Kaiser, J., & Datta, S. (2021). Probabilistic computing with p-bits. Applied Physics Letters, 119(15).
[2] Misra, S., Bland, L. C., Cardwell, S. G., Incorvia, J. A. C., James, C. D., Kent, A. D., ... & Aimone, J. B. (2022). Probabilistic neural computing with stochastic devices. Advanced Materials, 2204569.
Required background: Master’s degree in Computer Science, Informatics, Physics, Engineering or Electronics, with knowledge about artificial intelligence and machine learning
Type of work: Modelling, algorithmic and system design, experimentation, literature study
Supervisor: Steven Latré
Co-supervisor: Tom De Schepper
Daily advisor: Julie Moeyersoms
The reference code for this position is 2024-092. Mention this reference code on your application form.