Module 1

Combined Answer: Supervised, Unsupervised, and Reinforcement Learning

covers Q3 4 5 7 8 9 10

Aspect	Supervised Learning	Unsupervised Learning	Reinforcement Learning
Definition	Learns using labeled data (input-output pairs). Model predicts outputs based on training.	Learns using unlabeled data, finding hidden structures and patterns.	Learns by interacting with the environment, improving decisions using rewards & punishments.
Nature of Output	Predicts known outcomes (categorical or continuous).	Groups or associates data without predefined labels.	Sequence of actions to maximize long-term reward.
Techniques	- Classification (Spam detection, Face recognition) - Regression (House price prediction, Weather forecasting)	- Clustering (Market segmentation, Image grouping) - Association Rules (Market Basket Analysis: bread → butter)	- Q-Learning - Deep Reinforcement Learning - Policy Gradient methods
Examples	Email spam filtering, predicting stock prices, medical diagnosis.	Netflix movie recommendation, customer segmentation, anomaly detection.	Self-driving cars, game-playing AI (chess, Go), robots learning to walk.
Analogy	Learning with a teacher – given right answers during training.	Learning without a teacher – self-discovers patterns.	Learning from trial and error – like a child learning to walk.
Advantages	Accurate predictions, well-suited for structured problems.	Useful when labeled data is scarce, finds hidden insights.	Handles sequential decision-making, adapts to dynamic environments.
Limitations	Needs large labeled datasets, costly to prepare.	Results may be less accurate, difficult to interpret.	Training is time-consuming and resource-intensive.

Q1. Define Machine Learning. Explain how machine learning is different from conventional programming.

Answer:

Machine Learning (ML) is a subfield of Artificial Intelligence (AI) concerned with developing algorithms that allow computers to learn patterns and make predictions without being explicitly programmed.

Arthur Samuel (1959) defined ML as:

“A field of study that gives computers the ability to learn without being explicitly programmed.”

ML systems learn from experience (data) and improve their performance automatically. Instead of being coded with step-by-step instructions, they analyze data, build models, and predict outcomes.

Difference from Conventional Programming:

In conventional programming, a programmer writes rules and logic explicitly. Input data is processed according to these fixed rules to produce output.
In machine learning, the system is fed large amounts of data along with expected outcomes (for supervised tasks). The algorithm learns rules and relationships on its own to make predictions.

Aspect	Conventional Programming	Machine Learning
Approach	Rules are explicitly coded by programmers.	Model learns patterns from data.
Flexibility	Works only for tasks programmer anticipated.	Adapts to new/unseen data.
Example	Writing a program to calculate tax with given formula.	Predicting house prices from historical data.

Thus, ML is data-driven while conventional programming is rule-driven.

Q2. What are the types of Machine Learning? Explain in brief with examples.

Answer: