Minimal Hebbian Rule

Tags

adaptive-intelligence

Is the [see page 12, quintessential] hebbian rule based around Hebbs postulum.

$$\begin{array}{r}\mathrm{\Delta }{w}_{ij}=\alpha v_i^{\text{post}}{v}_j^{\text{pre}}\end{array}$$

Which states that the weight of a synaptic connection between a pre-synaptic neuron $j$ and a post-synaptic neuron $i$ should be proportional to the product of both neurons output-potential. $\alpha$ in the above equation is some non-negative constant and is used as a learning-rate.

Note By this minimalist definition we only change the weights for a neuron when both its input neuron and itself is firing. If either the output or input potential is 0 then the change in weights will also be 0.