Linear regression:Parameters :: Neural networks:?

Linear regression uses a set of numeric parameters – the slope and intercept – that are adjusted to fit the data. In a neural network the analogous adjustable quantities are the weights and biases that sit on each connection and neuron. These weights and biases are tuned during training so the network learns to map inputs to outputs. For example, a simple two‑layer network has a weight matrix between the input and hidden layer and another between hidden and output, plus bias vectors for each layer, which together form the model’s parameters. Thus, just as linear regression’s parameters define its predictions, a neural network’s weights and biases define its predictive behavior.