Java Deep Learning Projects

Weights and biases

Besides the state of a neuron, synaptic weight is considered, which influences the connection within the network. Each weight has a numerical value indicated by W_ij, which is the synaptic weight connecting neuron i to neuron j.

Synaptic weight: This concept evolved from biology and refers to the strength or amplitude of a connection between two nodes, corresponding in biology to the amount of influence the firing of one neuron has on another.

For each neuron (also known as, unit) i, an input vector can be defined by x_i= (x₁, x₂,...x_n) and a weight vector can be defined by w_i= (w_i1, w_i2,...w_in). Now, depending on the position of a neuron, the weights and the output function determine the behavior of an individual neuron. ...