Brain-inspired computational primitives for reliable neuromorphic intelligence
summary
Concrete results and numbers drive the claims: analog CMOS neuron circuits with memory encoded in network structure, chips of 200–1,000 neurons, and an encoder that cuts dimensionality so a reservoir computing readout detects seizures from scalp EEG. Field demos report 10× lower power for a smart-irrigation prototype, on‑device beehive audio monitoring, and low‑latency neuromorphic control for a robot playing air hockey; limitations and hardware tradeoffs (time constants, capacitor size, and noisy encodings) are discussed and partly solved.
AI-written summary · about
chapters (13)
- 0:00Neuromorphic technologies intro
- 2:35Neuromorphic computation goals
- 5:26Brain inspires reliable computing
- 10:25Brain vs standard computing
- 13:36Biologically inspired hardware
- 17:09Challenges driving neuromorphic design
- 22:49Edge intelligence needs
- 24:15Health monitoring in neuromorphic
- 26:01Precision farming irrigation
- 27:46Beehive audio monitoring
- 28:28Robotic control latency
- 30:13Future of neuromorphic intelligence
- 32:21Q&A
related talks
01h 57mIntroduction to Neuromorphic Computing
43mKeeping Moore's law alive: Neuromorphic computing
24mMemristor-based neuroelectronics for artificial and hybrid intelligence
20mRecap and Closing Remarks: Neuropunk revolution and energy consumption
01h 14mTissue vs silicone musings on the future of deep learning hardware and software
01h 16mRethinking Computing with Neuro-Inspired Learning: Algorithms to Hardware
42mOn the Link Between Energy & Information for the Design of Neuromorphic Systems
30mCracking the code: neuronal networks in the brain