ABI Salon

Upcoming events

May 30th, 2024

Title: Energy-efficient machine intelligence on neuromorphic accelerators

Invited talk in the ABI theme "Interplay between complexity and artificial & biological intelligence".

3.30pm - 5pm, Room 237B, Advanced Research Centre (ARC)

Alessandro Pierro, Researcher @ Neuromorphic Computing Lab, Intel Corporation & Ph.D. Student @ Institute for Computer Science, Ludwig Maximilian University of Munich

Abstract:

Neuromorphic computing takes inspiration from the brain to design novel algorithms and hardware platforms. Leveraging sparse, event-driven, and massively parallel processing enables lower latency and better energy efficiency on critical machine intelligence workloads. This seminar will cover the motivation and principles behind modern neuromorphic architectures, focusing on the Intel Loihi 2 research chip as a representative example. Moreover, through two case studies, we will discuss the hardware-aware design methodology and how it can be applied to leverage the specific neuromorphic feature set. Our results on combinatorial optimization and sequence modeling show great promise for our approach to accelerate compute-intensive tasks, especially in edge and energy-constrained domains, and advocate for more hardware-informed design in computational sciences.

ADD EVENT TO GOOGLE CALENDAR

Past events

January 31st, 2024: 

ABI salon launch workshop

1:30 pm - 5:00, Room 237C, Advanced Research Centre (ARC)

Speakers:

Dr Benjamin Peters, Marie Curie Fellow, School of Psychology & Neuroscience, University of Glasgow

Dr Giulia Marcucci, Research Associate at the School of Physics and Astronomy, University of Glasgow

Dr Paul Henderson, Lecturer, School of Computing Science, University of Glasgow

March 7th, 2024

Title: Physics & Complexity in AI hardware

A lecture starting the ABI theme "Interplay between complexity and artificial & biological intelligence". 

3.30pm - 5pm, Room 237B, Advanced Research Centre (ARC)

Dr Giulia Marcucci, Research Associate at the School of Physics and Astronomy, University of Glasgow

Abstract:

The interplay between complexity and artificial and biological intelligence is a fascinating frontier combining insights from nonlinear physics, computational science, and neuroscience. This seminar explores the profound connection between these diverse fields, focusing on the pivotal role of complex wave propagation in shaping our understanding and development of intelligent systems. We delve into how complex waves, ubiquitous in domains such as optics and hydrodynamics, serve as a bridge between fundamental principles and practical applications in computing and neural processing. 

ADD EVENT TO GOOGLE CALENDAR

April 11th, 2024

Title: Harnessing multistability: Expanding the capabilities of artificial neural networks via multifunctionality

Invited talk in the ABI theme "Interplay between complexity and artificial & biological intelligence".

3.30pm - 5pm, Room 237B, Advanced Research Centre (ARC)

Dr Andrew Flynn, Postdoctoral Researcher at the INFANT Research Centre and the School of Mathematical Sciences, University College Cork, Ireland

Abstract:

Multifunctionality describes a biological neural network's ability to harness multistability in order to perform various tasks without altering its network properties. The advantages of extending multifunctionality to the domain of artificial neural networks (ANNs) will be made clear during this talk. Multifunctionality unlocks several new machine learning application areas for ANNs such as: data-driven modelling of multistability, generating chaotic itinerancy, novel memory recall techniques, and reconstructing transitions present in the epileptic brain. This talk will outline how multifunctionality has so far been realised in an artificial setting with a reservoir computer (RC), a dynamical system in the form of an ANN. The formalism of generalised synchronisation is used to describe how to train a RC to achieve multifunctionality and also explore some of the challenges involved in realising multifunctionality. The results presented in this talk not only illuminate the exotic dynamics and exciting applications of multifunctional RCs but also highlight the importance of a dynamics-driven approach when training ANNs to display a broader level of intelligence through performing multiple tasks using the same set of trained weights while also confronting the ‘explainability’ elephant in the room with us and artificial intelligence. 

ADD EVENT TO GOOGLE CALENDAR

April 25th, 2024

Title: Gain-controlled nonlinearities and interactions for photonic reservoir computing

Invited talk in the ABI theme "Interplay between complexity and artificial & biological intelligence".

3.30pm - 5pm, Studio 2, Advanced Research Centre (ARC)

Dr Juan Sebastian Totero Gongora, EPSRC Quantum Technology Fellow, Senior Lecturer and a member of the Emergent Photonics Research Centre at Loughborough University

Abstract:

Nonlinear optical waves are emerging as an attractive substrate for photonic reservoir computing. By combining high bandwidth and low power consumption, optical devices bear the promise to augment, or for some specific scenarios replace, CMOS-based electronic computing devices. A key challenge is the limited scalability of photonic technologies based on one-to-one implementations of artificial neurons or logic gates. In this presentation, I will provide an overview of current research on complexity-based neuromorphic photonic systems, where information processing is driven by complex nonlinear interactions among thousands of optical waves in single optical components. In particular, I will discuss how the nonlinear interactions between the optical waves (an essential ingredient for efficient reservoir computing) can be tuned and controlled in gain systems (e.g., integrated lasers or doped fibres) to fine-tune the generalisation performance of the system.

ADD EVENT TO GOOGLE CALENDAR