AIGP Study Guide
Module 1: Foundations of AI · BoK IV.A

The four ways machines learn

Four ML approaches: supervised (labelled), unsupervised (unlabelled), semi-supervised (small labelled + large unlabelled) and reinforcement (agent learns by trial and error). LLMs rely on semi-supervised learning is a recurring fact.

AI models must be taught before they can solve anything. Machine learning is that teaching process. Four approaches, one comparison table, then the detail the exam digs into.

The four ML approaches
ApproachTraining dataGoalExamplesWatch-outs
Supervised learning|SupervisedLabelled data, inputs mapped to known targetsPredict outputs for new, unseen dataSpam detection · fraud flagging · labelled road signsLabelled data is costly and slow · labelling can introduce bias
Unsupervised learning|UnsupervisedUnlabelled data, no predefined targetsFind hidden patterns, structures, relationshipsCustomer segmentation · anomaly detection · genetics, fault detection, marketingCheaper but less accurate, unpredictable, interpretation is subjective
Semi-supervised learning|Semi-supervisedSmall labelled + large unlabelled setBest of both worlds, cuts manual labelling costSpeech recognition · LLMs often rely on it · ChatGPT, DALL-ELabel quality and consistency · choosing an algorithm that handles both data types
Reinforcement learning|ReinforcementNo labels → an agent interacts with an environmentMaximise reward through trial and errorGame AI · robots in mazes/warehouses · AVs · predictive text · real-time ad biddingDesigning the reward mechanism · exploration vs exploitation trade-off

Inside supervised learning - two sub-types. Classification models predict specific categorical responses by labelling input data (e.g. "Spam" vs "Not spam", image recognition, medical diagnosis); SVM (Support Vector Machine)|SVM is used mostly for classification. Regression models predict a continuous numerical outcome (e.g. car price, stock prices, temperature); SVR (Support Vector Regression)|SVR most commonly produces continuous values.

Inside unsupervised learning - two sub-types. Clustering automatically groups data points sharing similar attributes (e.g. DNA samples, customer segments). Association rule learning identifies relationships between data points (e.g. people who buy X also buy Y).

Reinforcement learning mechanics

An agent acts in an environment → rewarded actions get reinforced, errors trigger penalties proportional to the error's scale. The agent is never told what to do, it learns from trial and error. Real deployment → Amazon's warehouse supply chain optimisation.

Exam flash

Scenario says "labelled" → supervised. "Find patterns, no labels" → unsupervised. "Rewards and penalties" → reinforcement. "A bit of labelled, lots of unlabelled" → semi-supervised. LLMs → semi-supervised is a recurring fact.

Key terms - quick answers

What is “Supervised learning”?
Learning from labelled data to predict outputs for new data; labelling is costly and can introduce bias.
What is “Unsupervised learning”?
Learning from unlabelled data to find hidden patterns; cheaper but less accurate and subjective.
What is “Semi-supervised learning”?
Learning from small labelled plus large unlabelled data; LLMs often rely on it.
What is “Reinforcement learning”?
An agent learns by trial and error in an environment, maximising reward via reinforcement and penalties.