| 摘要 |
PROBLEM TO BE SOLVED: To provide a technique for unsupervised training of input synapses of primary visual cortex (V1) simple cells and other neural circuits.SOLUTION: The unsupervised training method utilizes simple neuron models for both Retinal Ganglion Cell (RGC) and V1 layers. The model simply adds the weighted inputs of each cell, where the inputs can have positive or negative values. The resulting weighted sums of inputs represent activations that can also be positive or negative. The weights of each V1 cell are adjusted depending on a sign of corresponding RGC output and a sign of activation of that V1 cell in the direction of increasing the absolute value of the activation. The RGC-to-V1 weights are positive and negative for modeling ON and OFF RGCs, respectively.SELECTED DRAWING: Figure 5 |
