Intermediate · Neural Networks
Convolutional filter
Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.
TL;DR. A small matrix used in convolutional operations to detect specific features like edges or textures within an input matrix.
Technical Definition
A small matrix used in convolutional operations to detect specific features like edges or textures within an input matrix.
How it works
A convolutional filter, also known as a kernel, is a small matrix of learnable weights. In a convolutional operation, this filter slides over an input matrix (e.g., an image patch) and performs element-wise multiplication followed by summation. The values within the filter are adjusted during training to detect specific patterns or features, such as edges, corners, or textures, in the input data. These detected features are then passed on to subsequent layers.
Related Concepts
- Convolution — A mathematical operation that applies a filter to an input matrix, used in deep learning to extract features efficiently.