PhD position in Excitable Hopfield Networks
PhD - Leuven | Just now
Human brain has been an inspiration for development of AI algorithms especially for Neuromorphic Computing. In recent years, this field has gained significant attention thanks to its low energy consumption and low latency properties with its application to traditional neural network architectures over the last decade. In one such areas, initial works on Hopfield Networks with neuromorphic event regulation has been done very recently this year.
While traditional static Hopfield Networks with passive nodes and static amplifier edges act as auto associative memories, the advent of dynamic associative memories with excitable Hopfield Networks is promising and impactful. The design of these novel Hopfield Networks with excitable nodes and network interconnects offers inherent hierarchical design and scalable machine. To push the frontier research in this direction, this PhD research aims to develop novel algorithms based on the following perspective:
(a) event based Hopfield Network with hierarchical, heterogeneous and scalable properties.
(b) develop algorithms based on hardware constraints from a co-design perspective.
(c) learning rules beyond Hebbian learning appropriate for the novel network architecture.
During the PhD, you will investigate the novel Hopfield Networks in close collaboration with academic supervision from Host University (KU Leuven, Prof. Rodolphe Sepulchre) and scientific supervision from Dr. Srutarshi Banerjee of Imec’s N-Algorithms for Alternative Computing Systems (NACS) team. Prospective candidates should be passionate, self-driven with strong interest in machine learning, control systems, neuroscience principles, and related domains. The candidate should be able to develop algorithms and motivate their benefits for cutting-edge applications.
Main Responsibilities
- Conduct literature reviews on Hopfield Networks, Neuromorphic Excitability, (Neuromorphic) Control Systems, Hebbian learning, and related areas.
- Design algorithms based on neuromorphic excitability.
- Demonstrate hierarchical, heterogeneous and scalable properties of the model.
- Follow up research work with publications in top-tier Journals and / or Conferences.
- Fulfil PhD requirements of the Host University as necessary.
We are looking for a creative, highly motivated, passionate and disciplined PhD candidate with the following qualifications:
- You have (or will obtain in the coming months) a Master’s degree in Electrical Engineering, Computer Science, Physics, Neuroscience or a related field, with excellent academic results.
- You have acquired expertise in one or several of the following areas: Neuromorphic algorithm design, control design, computational neuroscience, associative memories.
- You have a genuine interest in developing Neuromorphic models.
- You possess solid programming skills (e.g., Python, C, etc.).
- Interest in hardware-software co-design.
- You are willing to learn neuromorphic platforms such as SNNTorch, NEST, etc.
- You have good communication skills in English, both written and oral.
- You must be a team player and ready to change strategy / directions if necessary.
Required background: Master’s degree in Electrical Engineering, Computer Science, Physics, Neuroscience or a related field
Type of work: 40 % theoretical / modeling work, 40 % computer implementation / simulation, 20 % literature
Supervisor: Rodolphe Sepulchre (KU Leuven)
Daily advisor: Srutarshi Banerjee
The reference code for this position is 2027-124. Mention this reference code in your application.