Machine Learning

Algorithms that learn patterns from data rather than following fixed rules.

CurrentbeginnerBeginner-friendlyFull course available

Overview

Machine learning is the broad field of algorithms that improve at a task by learning from data, rather than following manually programmed rules. Deep learning (neural networks) is one family of ML techniques among several (others include decision trees and linear models).

What it is
A field of algorithms that learn patterns from data to make predictions or decisions.
Why it's used
For tasks where the 'rules' are too complex or numerous to hand-write, but plenty of example data exists to learn from.
Where it fits
The layer between 'AI' (the broad umbrella) and 'deep learning' (one specific ML technique family).

Core concepts

  • Training data and labels
  • Model, parameters, and training
  • Overfitting
  • Evaluation metrics

Example

Training is iterative: the model's parameters are nudged repeatedly to reduce error on example data, not solved for in one step.

// Simplified idea of "training":
// Show the model thousands of (email, is_spam) pairs.
// It adjusts internal parameters to reduce prediction error
// on those examples -- then generalizes to new emails.

Common use cases

  • Spam and fraud detection
  • Recommendation systems
  • Forecasting

Project ideas

  • Explain, in plain language, how you'd design training data for a model that predicts whether a customer will churn

Official references