Despite tremendous progress in recent deep learning algorithms and hardware, there still exists three to four orders of energy efficiency gap between artificial and natural intelligence. Towards bringing this gap, up-to-date results of neuroscience needs to be abstracted to the algorithmic level, so that they can be implemented in engineering solutions. Based on understanding how the brain works, engineers from various disciplines need to collaborate on learning algorithms, knowledge representation, and architectures. The objective of this workshop is to bring together researchers from across the JUMP (Joint University Microelectronics Program, funded by SRC and DARPA) community to discuss the most promising approaches towards the overarching goal of bridging the energy efficiency gap between artificial and natural intelligence. The workshop will explicitly focus on approaches that are well beyond today’s deep neural networks and neural network accelerators. Topics of interest will include bio-inspired network models and learning algorithms, neuro-mimetic devices and circuits, and hardware that embodies the information processing principles of the brain. Exemplary research efforts from both academia and industry will be presented. The workshop aims to establish a forum to discuss the current practices, as well as future research needs in the corresponding fields.
For workshop details and agenda