IARPA scopes out brain-inspired computing
- By Derek Major
- Jan 15, 2016
The Intelligence Advanced Research Projects Activity is looking for information on neurally inspired computing principles so it can better understand the challenges and potential opportunities for developing next-generation computers.
Addressing the request for information specifically to neuroscientists and computer scientists, IARPA seeks answers to questions relating to four characteristics of brains and how they might apply to computers:
Spike-based representations: Brains operate using spike-based codes that appear infrequently across populations of neurons, suggesting a plausible role for approximate computation.
Asynchronous computation or transient coordination: Brains do not update all computing elements at once, but rather neurons only transiently coordinate their activity.
Learning: Brains employ plasticity mechanisms that operate continuously to support learning.
Co-local memory storage: Brains do not strictly segregate memory and computing elements; synaptic inputs to a neuron can store memories and support computation.
Responses will be used to help organize a workshop on this topic at the 2016 Neuro Inspired Computational Elements conference and to inform future programs in this technical area, IARPA said.
The RFI is available here. Responses are due Jan. 29.
Derek Major is a former reporter for GCN.