For the last decade, the developments in dedicated hardware for Machine Learning (ML) has been crucial for recent fundamental advancements in Artificial Intelligence (AI). The central paradigm of the current hardware architectures is to minimise inefficiencies of von-Neumann systems - costly data movement between physically separated memory and compute units. This is typically achieved through aggressive parallelisation and reuse of data. However, current hardware solutions, although optimised explicitly for parallel computing, use digital CMOS technology and conventional von- Neumann architecture. Digital components are inherently unsuitable for the realisation of analogue synapses/weights in artificial neural networks, and the sequential nature of von-Neumann architecture is intrinsically inefficient for vector-matrix multiplications that dominate most ML algorithms. The approaching demise of Moore's Law makes the need for approaches beyond CMOS technologies all the more needed.
This symposium will cover recent developments in smart adaptive materials and devices able to implement specific brain-inspired functionalities in a compact and power-efficient manner. Examples of these include the functionality of synapses, neurons and their assemblies (e.g. plasticity, adaptation, spiking, integration, operation synchronization and auto-organization). The focus will be given to memristive materials systems such as oxides, ferroelectrics, ferroics, organic and inorganic, self-assembly materials, among the others. Apart from the electrical operations, the symposium will investigate the interaction of memristive devices with external non-electric stimuli, such as light, magnetic fields, exploring possibilities of developing compact integrated sensor/processing/memory units. Furthermore, the symposium will address emerging fields of bio-hybrid systems covering materials for brain-computer interfaces.
Finally, the symposium will provide a perfect platform for the exchange of ideas between many different scientific communities, from solid-state physicists, material scientists, chemists, electrical engineers, scientists working in fields of neuromorphic computing, computational neuroscientists, and computer scientists.
University College London
Giuliana DI MARTINO
University of Cambridge
Institut de Ciència de Materials de Barcelona (ICMAB-CSIC)
Helmholtz-Zentrum Berlin fur Materialien und Energie
Ensemble3 sp. z o.o.
133 Wólczyńska St.
Functional Materials Technology Group
Optical Nanocharacterization Group
Inverse Materials Design Group
Next-Generation Energy Systems Group
Biophotonic Applications Group
Solar Energy Conversion Group
Oxide Single Crystals Group