Lecture 1: What is Machine Learning?
Read the lecture transcript below to learn the fundamentals of machine learning.
Lecture Transcript
Welcome to the first lecture in our Introduction to Machine Learning series. Today we are going to answer a fundamental question: what exactly is machine learning, and how does it differ from traditional programming and from artificial intelligence as a whole?
Machine learning is a subset of artificial intelligence that focuses on building systems which learn from data rather than being explicitly programmed with rules. In traditional software development, a programmer writes specific instructions telling the computer exactly what to do in every situation. With machine learning, we instead provide the computer with large amounts of data and allow it to discover patterns and make decisions on its own. The formal definition, attributed to Arthur Samuel in 1959, describes machine learning as "the field of study that gives computers the ability to learn without being explicitly programmed."
It is important to understand the relationship between artificial intelligence and machine learning. Artificial intelligence is the broader concept of machines being able to carry out tasks in a way that we would consider intelligent. Machine learning is a specific approach to achieving AI. In other words, all machine learning is artificial intelligence, but not all artificial intelligence is machine learning. For example, a rule-based expert system that uses hand-coded if-then rules is considered AI but is not machine learning because it does not learn from data.
There are three primary types of machine learning. The first is supervised learning, which is the most common approach. In supervised learning, the model is trained on labeled data, meaning each training example is paired with a correct output or target value. The algorithm learns to map inputs to outputs by studying these labeled examples. Common tasks include classification, where the output is a category such as spam or not-spam, and regression, where the output is a continuous number such as predicting house prices.
The second type is unsupervised learning, where the model works with unlabeled data. There are no predefined correct answers. Instead, the algorithm must find hidden patterns, groupings, or structures within the data on its own. Clustering is a popular unsupervised technique where similar data points are grouped together. For instance, a retailer might use clustering to segment customers into distinct groups based on purchasing behavior without pre-defining what those groups should be. Dimensionality reduction is another unsupervised method that simplifies data while preserving its essential structure.
The third type is reinforcement learning, which is fundamentally different from the other two. In reinforcement learning, an agent interacts with an environment and learns by receiving rewards or penalties for its actions. The agent's goal is to maximize its cumulative reward over time. This approach has been notably successful in game-playing AI, such as DeepMind's AlphaGo, which defeated the world champion in the board game Go in 2016. Reinforcement learning is also used in robotics, autonomous vehicles, and recommendation systems.
To summarize: machine learning is a subset of AI that learns from data. Supervised learning uses labeled data, unsupervised learning finds patterns in unlabeled data, and reinforcement learning learns through trial-and-error with rewards. Understanding these three categories is essential for choosing the right approach to any given problem.
Supervised learning uses labeled data where each training example is paired with a correct output. The model learns to map inputs to outputs. Common tasks include classification and regression.
Unsupervised learning works with unlabeled data and discovers hidden patterns or groupings on its own. Clustering and dimensionality reduction are common unsupervised techniques.
In reinforcement learning, an agent learns by interacting with an environment and receiving rewards or penalties. The goal is to maximize cumulative reward over time. DeepMind's AlphaGo is a famous example.
Machine learning is a subset of artificial intelligence. All ML is AI, but not all AI is ML. Rule-based expert systems are AI but not ML because they do not learn from data.