Machine learning
Definition
Machine learning (ML) is the study of algorithms that improve with experience (data). Key paradigms include supervised learning (learning from labeled examples), unsupervised learning (finding structure without labels), and reinforcement learning (learning from rewards).
ML is preferred over hand-coded rules when the problem is too complex to specify explicitly or when data is abundant. It sits between classical AI (symbolic rules) and deep learning (large neural networks); many real-world systems combine ML models with pipelines and business logic.
How it works
Train: You choose a representation (e.g. linear model, tree, or neural network) and an objective (loss for supervised/unsupervised, reward for RL). An optimizer (e.g. gradient descent) updates the model parameters to minimize the loss or maximize the reward on the training data. Model: The result is a fitted model (weights, structure) that captures patterns in the data. Predict: At inference time, you feed new inputs into the model to get outputs (labels, scores, or actions). Evaluation uses train/validation/test splits to estimate generalization and avoid overfitting.
Use cases
Classical ML shines when you have structured or tabular data and clear labels or targets.
- Spam classification, fraud detection, and other supervised classification tasks
- Recommendation systems and collaborative filtering
- Forecasting and time-series prediction
External documentation
- Google ML crash course
- Scikit-learn – User guide — Classical ML in practice