MLP (multi-layer perceptron)
A multi-layer perceptron (MLP) is a feedforward neural network made up of an input layer, one or more hidden layers, and an output layer. Each layer contains units that combine incoming values using learned weights and biases, then typically apply an activation function before passing results to the next layer. Training adjusts those weights and biases so the network's outputs better match examples in the training data. With hidden layers and nonlinear activations, an MLP can learn complex relationships and is used for tasks such as classification and regression on structured data. Unlike recurrent networks, a standard MLP does not itself maintain memory of earlier inputs, so it is not inherently designed to model sequence order.