AI Dictionary › Fondamenti AI

Support Vector Machine (SVM)

A support vector machine, or SVM, is a machine learning model that tackles classification problems by finding the boundary that separates classes with the widest possible margin, that is, the greatest distance from the closest points of each category. The intuition is that, among all the lines or surfaces that correctly separate the training data, the one with the widest margin generalizes best to new, unseen data.

Definition

How it works

The points closest to the separating boundary, the ones that actually determine where to draw it, are called support vectors, and they give the algorithm its name: they are the only data points truly relevant to the decision, all the others could be removed without changing the result. When the data cannot be separated with a straight line, SVMs use what is known as the kernel trick, a mathematical transformation that projects the data into a higher-dimensional space where it becomes separable, without ever having to compute that transformation explicitly.

Applications

SVMs were, for years, the state of the art in many classification tasks before deep learning became widespread, and they remain a solid choice for moderately sized datasets with many features: text classification, handwriting recognition, medical image categorization, bioinformatics for classifying genetic sequences. They work particularly well when the number of features is large relative to the number of available examples.

History & etymology

The theoretical foundations of SVMs originate in the work of Vladimir Vapnik and Alexey Chervonenkis in the Soviet Union starting in the 1960s, within the field of statistical learning theory. The modern formulation with maximum margin and the kernel trick, which made the algorithm practical and powerful, was developed by Vapnik together with colleagues such as Bernhard Boser, Isabelle Guyon and Corinna Cortes between the late 1980s and early 1990s at Bell Labs, after Vapnik moved to the United States.

Related terms

• Classification
• K-Nearest Neighbors (KNN)
• Feature engineering
• Hyperparameter tuning
• Regularization

More in Fondamenti AI

• LLM (Large Language Model)
• Token
• Context Window
• Temperature
• AI Hallucination
• Fine-tuning
• RAG (Retrieval Augmented Generation)
• Embedding

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

📱 Download the Android app (beta) iOS coming soon

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.