Memristive Devices In Neuromorphic Systems
summary
Organic polyaniline volatile memristors are built with a solid Li-doped polymer electrolyte and show hysteresis, rectification and tunable conductivity; volatile devices form the reservoir and nonvolatile memristors form the readout. Perceptrons and XOR are implemented plus a reservoir computing classifies digits demonstration on experimental hardware. A hardware STDP loop reproduces Pavlovian conditioning as a STDP Pavlov circuit, and memristors bridge cortical slices so paired neurons reach ~2 ms spike timing with live neurons. Devices now cost ~10 pJ per event, target ~100 fJ; software is billions of times less efficient.
AI-written summary · about
chapters (13)
- 0:00Memristive devices overview
- 1:10Neuromorphic systems defined
- 4:44Memristor characteristics
- 5:50Memristor oscillations
- 6:11Noise benefits in computing
- 8:22Cross-talk advantages
- 10:38Spike timing dependent plasticity and learning algorithms
- 13:09Pavlovian learning
- 14:58Reservoir computing
- 17:33Living neuron interfaces
- 20:06Energy efficiency
- 21:11Closing remarks
- 21:33Question and answer session
related talks
01h 57mIntroduction to Neuromorphic Computing
37mHodgkin-Huxley, Memristors, and Edge of Chaos
41mBrain-inspired computational primitives for reliable neuromorphic intelligence
43mKeeping Moore's law alive: Neuromorphic computing
01h 05mReal In-Memory Processing
01h 13mComputation in Networks of Neurons in the Brain II
12mComputation with sequences of neuronal assemblies
42mOn the Link Between Energy & Information for the Design of Neuromorphic Systems