Artificial Intelligence & Machine Learning Basics

Artificial Intelligence (AI)

• Weak AI (Narrow AI): AI trained to do one specific task very well (e.g., Apple's Siri, self-driving cars, chess-playing robots). All AI that exists today is Narrow AI.
• Strong AI (General AI): A theoretical machine with generalized human cognitive abilities. If presented with an unfamiliar task, it could find a solution just like a human. This does not exist yet.

Machine Learning (ML)

1. Types of Machine Learning

A. Supervised Learning

• How it works: The algorithm is trained on labeled data. You provide the input AND the correct output (the answer key) during training.
• Example: You show the computer 1,000 pictures of cats labeled 'Cat' and 1,000 pictures of dogs labeled 'Dog'. The computer learns the features of both. Then you give it a new picture, and it predicts if it's a cat or a dog.
• Use Cases: Spam filtering, Image recognition, Price prediction.

B. Unsupervised Learning

• How it works: The algorithm is given unlabeled data (no answer key). It must figure out the patterns and hidden structures in the data on its own.
• Example: You give the computer 10,000 unlabelled news articles. The computer groups them into categories (sports, politics, entertainment) based on similar words.
• Use Cases: Customer segmentation, recommendation systems (like Netflix).

C. Reinforcement Learning

• How it works: The algorithm (Agent) learns by interacting with an environment. It learns by Trial and Error. It gets a 'reward' for doing something right and a 'penalty' for doing something wrong.
• Example: Training an AI to play Super Mario. If it moves forward, give it a point. If it falls in a pit, subtract a point. Eventually, it learns to beat the level flawlessly.
• Use Cases: Robotics, Game AI, Self-driving cars.

Deep Learning & Neural Networks

• Neural Networks: Inspired by the human brain. They consist of layers of interconnected 'neurons' (nodes).
• Structure: Has an Input Layer, multiple Hidden Layers (where the 'deep' processing happens), and an Output Layer.
• Why use it? Deep learning excels at recognizing highly complex patterns (like understanding human speech or facial recognition) where traditional ML algorithms struggle.