eISSN 2093-744X
pISSN 1598-2645
pISSN 1598-2645
Direct training algorithms
| Algorithm | Neuron model | Architecture | Input encoding | Output decoding | Features |
|---|---|---|---|---|---|
| SpikeProp (2000, [37]) | SRM | Shallow network | Population code | Time-to-first code | Surrogate gradient; multiple delayed synaptic terminals |
| ReSuMe (2005, [38]) | don’t care | (FF, RNN, LSM)+ trainable single layer | Spike train | Spike train | Train the weights for the last layer; STDP & anti-STDP |
| PES (2011, [40]) | IF/LIF model | Two-layered network | Spike train (firing rate) | Spike train (firing rate) | MSE loss for decoded value |
| STBP (2018, [42]) | LIF | Shallow network | Spike train (rate code) | Spike train (firing rate) | BPTT-like over spatial & time domains |
| BP-STDP (2019, [32]) | LIF | Deep network | Spike train (spike count) | Direct output (spike count) | Backpropagation + STDP |
| SBBP (2019, [43]) | IF/LIF | Deep network | Spike train (rate code) | Direct output (membrane potential) | Surrogate gradient |
International Journal of Fuzzy Logic and Intelligent Systems 2021;21:317~337 https://doi.org/10.5391/IJFIS.2021.21.4.317
© IJFIS
© 2020. The Korean Institute of Intelligent Systems. / Powered by INFOrang Co., Ltd