This repository contains all the stuff what I am learning about ML and AI. To start with ML, I joined an online course from NPTEL called Machine Learning for Engineering and Science Application.
Aritifcial Intelligence - Replicates results of Human Being Machine Learning - Performs better with experience
Aritifcial Intelligence > Machine Learning > Deep Learning
An algorithm that uses data to answer questions. In addition to it, we can also define it as an algorithm which improves its performance as the data grows.
Book Definition - A computer program said to learn from experience E with respect to some class of tasks T and performance measure P.
Task - T (Recognizing Spam)
Experience - E (Data - Emails and label them as Spam or not spam)
Performance Measure - P (How many emails are we labelling as Spam)
As E increases, the P increases
Inputs - Data and Rules
Outputs - Answers
Inputs - Date and Answers
Outputs - Rules
- Supervised Learning
- Date Labeled by Human
- Examples - OCR, Speech Recognition, Image Labeling
- Unsupervised Learning
- Unlabeled Data
- Examples - Grouping Customers, Detecting new Diseases
- Generative Approaches
- Creating new data from the given data
- Part of Unsupervised Learning
- Example - Give 100 images of cat, then draw a new cat
- Semi-supervised Learning
- Small amount of labeled data along with unlabeled data
- Self-supervised Learning
- No labeled data
- Labels are extracted from data using Heuristics
- Reinforcement Learning
- Actions are chosen based on rewards.
- Example - Chess, Games, etc
- Generating Data
- Preparing Data - Without bias
- Choosing a Model/Algorithm
- Examples - Random Forest, ANNs, Hidden markov Models, etc.
- Training
- Evaluation
- Hyperparameter Tuning
- Prediction
- Classification Problem
- Cancer or Non-Cancer, Spam or Not Spam
- Regression Problem
- Prediction of Stock Market based on history
Scalar - 0th order tensor
- Example - n = 3
Vectors - 1st order tensor - Example - x⃗ = [1, 2, 3, 4]