Modelling the Minds

We investigate how the human brain adapts to healthy ageing, neurological and psychiatric disorders, and how it re-engineers itself in most efficient ways. This includes generating new brain cells to offset anticipated loss, or redirecting the natural functions of neural stem cells to support protective mechanisms. Our guiding principle is that nature often finds the most effective solutions, and by studying these processes, we provide valuable insights that can inform neuromorphic research carried out by our colleagues in the Institute.– Sandrine Thuret, Professor of Neuroscience

We translate biological insights into computational and hardware systems by designing neuromorphic architectures that mimic the brain’s efficiency and adaptability. Our work spans emerging memory devices, in-memory computing architectures and event-driven spiking algorithms that learn reliably under noise and variability. By bridging nanoscale electronics with large-scale learning models, we aim to create energy-efficient and scalable platforms for artificial intelligence. Being part of the institute helps us to use different perspectives of learning and adaptation in the brain to design reliable and efficient computing hardware. – Bipin Rajendran, Professor of Intelligent Computing Systems

We are currently developing synthetic synapses needed for the brain-inspired, so-called neuromorphic computing systems, which require much less energy than conventional ones. We work on memritors (memory resistors), which are circuit elements that “remembers” its resistance value based on what signals have passed through previously. They work similarly to brain synapses, which adjust their connectivity based on previous signals, making the brain both complex and highly efficient. Our memritors now have both electrical and optical memory capabilities, bringing us a step closer to mimicking the brain. – Anatoly Zayats, Professor of Experimental Physics

I am a Digital Humanist and my group is currently drawing inspiration from Kahneman’s cognitive model of “fast and intuitive” vs “slow and deliberate” thinking. Our research uses this model as a framework to understand how meaning, inference and behavioural shaping unfold in AI systems. We use Fast AI to model real-time inference and prediction and Slow AI for retrospective, memory-rich model training.– Mark Coté, Reader in Data and Society