TL;DRSVMs find the boundary that maximally separates classes, using kernels for nonlinearity.
A Support Vector Machine finds the hyperplane with the widest margin between classes, defined by the closest points (support vectors). The kernel trick maps data into higher dimensions so a linear boundary can separate nonlinear classes. SVMs shine on small/medium, high-dimensional datasets (like text) but scale poorly to huge data and need feature scaling.
Key points
Maximizes the margin between classes
Support vectors define the boundary
Kernel trick handles nonlinear separation
Great on small/medium high-dim data
Common mistakes
Skipping feature scaling (SVMs need it)
Using SVMs on very large datasets
Ignoring kernel/parameter tuning
Try it: Explain what "maximizing the margin" buys you.
Example code
<!doctype html><html><head><meta charset="utf-8"></head>
<body style="background:#06040d;color:#e6e0ff;font-family:monospace;padding:20px"><pre>find boundary with widest gap between classes
support vectors = closest points define it
kernel → separate nonlinear data</pre></body></html>