Overview
This lecture introduces the basics of machine learning (ML) within artificial intelligence (AI), covering its definition, main tasks, and the three primary types: supervised, unsupervised, and reinforcement learning.
Machine Learning Fundamentals
- Machine learning is a subset of AI that enables computers to identify useful patterns in data for decision-making.
- Unlike human intuition, ML uses algorithms to determine optimal patterns or weights from large datasets.
- A model is created from historical data (training data) and then used to make predictions on new data.
Main Tasks in Machine Learning
- Regression predicts a numerical value (e.g., house price, number of customers, time until machine failure).
- Classification assigns items to categories (e.g., admit/not admit a student, threat level of a computer session).
Types of Machine Learning
- Supervised learning uses labeled data (inputs with known outputs) to train models for prediction or classification.
- Example: Identifying cats in pictures or detecting if people are wearing hard hats.
- Unsupervised learning uses unlabeled data to find hidden patterns, often through clustering.
- Example: Grouping customers by age and income for marketing segmentation (e.g., k-means clustering).
- Reinforcement learning trains an agent to make decisions via trial and error, rewarding or penalizing actions toward a goal.
- Example: A robot vacuum (Roomba) learning to navigate a room by receiving feedback on its moves.
Key Terms & Definitions
- Artificial Intelligence (AI) — Field enabling machines to perform tasks typically requiring human intelligence.
- Machine Learning (ML) — Subfield of AI focused on algorithms that learn from data.
- Model — Mathematical formula or algorithm that makes predictions or classifications.
- Training Data — Historical data used to train ML models.
- Regression — Task predicting continuous numerical values.
- Classification — Task assigning items to discrete categories.
- Supervised Learning — ML with labeled data guiding the model.
- Unsupervised Learning — ML finding patterns in unlabeled data.
- Clustering — Grouping data points based on similarity.
- Reinforcement Learning — ML where agents learn optimal actions through reward and punishment.
Action Items / Next Steps
- Review examples of regression and classification tasks.
- Prepare for upcoming lectures on the practical application of these ML types.