Multilayer Perceptron (MLP) is a type of artificial neural network that consists of an input layer, hidden layer and output layer. MLP is suitable for supervised learning tasks such as classification and pattern recognition, and it is often used in deep learning applications.
The input layer of an MLP consists of a certain number of neurons, or a set of designated “input nodes”. Each neuron has a certain weight that corresponds to its numerical input and a bias value. A bias value is added to the calculated sum of inputs for each neuron. The output layer of an MLP is the evaluated result of the input layer using the weights and bias values, and is fed forward to the hidden layer.
The hidden layer is a middle layer between the input and output layer of an MLP. This layer contains a certain number of neurons that each are used to process the input information and extract a certain pattern. This is done by applying an activation function to the output from the input layer. The most commonly used activation function is the Rectified Linear Unit (ReLU).
The MLP is one of the most used architecture in the field of artificial intelligence and machine learning due to its simplicity and flexibility. Once the model parameters are trained and optimized, it can be used to make predictions with new data. MLPs can be used for a wide variety of applications, such as natural language processing, speech recognition, image recognition and more.
