Introduction to Machine Learning (ML)

Introduction to machine learning starts with a simple idea: systems that improve at a task by learning patterns from data, instead of being fully hand-coded for every case. Another useful definition: machine learning is the process of deducing unknown values from known values.

This guide covers what machine learning is, a short theory example, and the main types of machine learning problems — supervised, unsupervised, and reinforcement learning — including classification vs regression.

What Is Machine Learning?

Machine learning is a branch of artificial intelligence that gives a computer system the ability to progressively learn and improve performance on tasks without being explicitly programmed for every detail of those tasks.

In practice, you collect examples (data), choose a model, and train it so that it can make useful predictions or discover structure on new inputs it has not seen before.

The Theory Behind Machine Learning

Remember the second definition: determine unknown values given a set of known values.

Example: Class 5 students usually score between 70% and 100%. Class 3 students usually score between 40% and 60%. A new student scores 45%. Which class would you place them in?

Most people choose Class 3, because 45% falls in the Class 3 range. That intuition is classification: assign a new observation to a category using patterns learned from labeled examples.

A Simple Classification Example

Given a training dataset of N observations {x_n} (n = 1, …, N) and corresponding target values {t_n}, the goal of machine learning is to predict t for a new value of x.

The usual approach is to construct a function y(x). For a new input x, y(x) is the prediction for t. We also care about uncertainty — often written as p(t | x), the probability of t given x. That captures how confident the model is.

That is the core theory in plain language: make predictions about what is not yet known, using structure learned from what is known. For a deeper walkthrough of classification, see Classification in Machine Learning.

Types of Machine Learning Problems

Machine learning problems are commonly grouped into supervised learning, unsupervised learning, and reinforcement learning. Related settings include semi-supervised learning and active learning.

Supervised Learning

In supervised learning, the training set includes input vectors {x} and corresponding target values. The task is to find a function f(x) such that f(x) ≈ t for new inputs.

Supervised problems are usually either classification or regression.

Classification: Targets come from a finite set of categories. The model assigns each input to one of those discrete classes (for example, spam vs not spam).

Regression: Targets are continuous values in real-valued space (for example, predicting a house price). For a practical regression walkthrough, see Machine Learning 101 — Basics of Logistic Regression and related regression tutorials on the blog.

Unsupervised Learning

Unsupervised learning looks for patterns or structure in the input data without target labels. Common goals include:

• Clustering — group similar examples (see What is K-Means in Clustering)
• Density estimation — model how data is distributed
• Dimensionality reduction — map high-dimensional data to fewer dimensions (for example 2D/3D) while keeping important structure

Reinforcement Learning

Reinforcement learning is a third major class of problem. An agent takes actions in an environment and receives rewards or penalties. Over time it learns a policy — which actions to take in which situations — to maximize expected reward.

Classic applications include game playing and robotics. Reinforcement learning is not the same as supervised or unsupervised learning, and it is not a subtype of semi-supervised learning.

Semi-Supervised and Active Learning

Semi-supervised learning uses a mix of labeled and unlabeled data when labels are expensive to obtain.

Active learning lets the model (or a training loop) choose which unlabeled examples should be labeled next, to improve performance with fewer labels.

Frequently Asked Questions

What is machine learning in simple terms?

Machine learning is a field of AI where systems improve at a task by learning patterns from data instead of being fully hand-coded for every case.

What are the main types of machine learning?

The three core types beginners should know are supervised learning, unsupervised learning, and reinforcement learning. Semi-supervised and active learning are important related settings.

What is the difference between classification and regression?

Classification predicts discrete labels or categories. Regression predicts continuous numeric values.

Is reinforcement learning supervised or unsupervised?

Neither. Reinforcement learning learns from rewards and penalties through interaction with an environment, not from a fixed labeled dataset alone.

Next Steps and Further Reading

You now have an overview of machine learning: definitions, the prediction framing, and the main problem types. Next, go deeper with these beginner tutorials:

• Basics of Neural Networks
• Basics of Perceptron in Neural Networks
• Classification in Machine Learning
• What is K-Means in Clustering
• Introduction to Support Vector Machines (SVM)
• All Machine Learning tutorials

Want structured ML training? Join Alkademy for instructor-led machine learning and data science classes with hands-on projects.