How many learnable parameters does a pooling layer typically have?

A pooling layer, such as max pooling or average pooling, typically has no learnable parameters. The purpose of a pooling layer is to reduce the spatial dimensions of the input feature maps while preserving the essential information. It does this by applying a specific function (like taking the maximum or average value) over a defined window or kernel size.

Since pooling layers only perform operations based on the input data and do not require any weights or biases that can be adjusted during training, they do not contain learnable parameters. This characteristic allows pooling layers to effectively down-sample feature maps without adding additional complexity to the model. The lack of parameters also contributes to the efficiency and speed of convolutional neural networks (CNNs), making them more suitable for larger datasets and deeper architectures.

Understanding the role of pooling layers in consolidating information while avoiding overfitting is crucial for designing effective machine learning models.