Support Vector Machine (SVM)

Supervised learning models with associated learning algorithms that analyze data for classification and regression analysis.

How it works

An SVM finds the decision boundary (hyperplane) that maximises the margin — the distance between the boundary and the nearest training examples from each class (the 'support vectors'). In linearly inseparable cases, the kernel trick maps data into a higher-dimensional feature space where a linear boundary can separate the classes. Common kernels include RBF (radial basis function), polynomial, and sigmoid. SVMs minimise a regularised loss that balances margin maximisation and classification error.

Why it matters

SVMs were the state of the art for classification tasks in the decade before deep learning. They have strong theoretical foundations in statistical learning theory and work well in high-dimensional spaces with limited training data, making them still valuable in genomics, text classification with hand-crafted features, and anomaly detection. Understanding SVMs provides intuition for concepts like margins, regularisation, and the role of data geometry that carry over into modern deep learning.