221EIA021: Machine Learning (2024)

https://meet.google.com/pmp-dybr-mzb 

Module 1: Definition and Applications of Machine Learning

• Definition of ML: Definitions by Arthur Samuel and Tom Mitchell

• Scope and Application of ML

• ML Algorithms

  • Supervised Learning:

    • Regression and Classification

    • Linear Regression with one variable:

      • Model representation

      • Cost function

      • Gradient descent for linear regression

      • Learning rate

  • Unsupervised Learning:

    • Cocktail party problem

  • Reinforcement Learning and Recommender Systems

Module 2: Linear Algebra and Advanced Regression

• Linear Algebra:

  • Inverse and Transpose of Matrices

  • Significance of Eigenvalues and Eigenvectors

  • Singular Value Decomposition

• Linear Regression with Multiple Variables:

  • Multiple Features

  • Polynomial Regression

  • Normal Equation

  • Non-invertibility

• Logistic Regression for Classification:

  • Hypothesis representation

  • Decision boundary

  • Multiclass classification

Module 3: Neural Networks

• Neural Networks:

  • Representation:

    • Non-linear hypothesis

    • Neurons and the brain

    • Weights, effective inputs, threshold, and activation function

  • The McCulloch-Pitts Neuron Model:

    • Single-layer neural network

    • Multi-layer neural network

  • Learning:

    • General learning rule for one neuron

    • Perceptron and Delta Learning Rules

    • Backpropagation Algorithm

Module 4: Hypothesis Evaluation and Statistical Theory

• Evaluating Hypotheses:

  • Estimating hypothesis accuracy

  • Sample error and true error

• Basics of Sampling Theory:

  • The Binomial Distribution

  • Estimators, Bias, and Variance

  • Confidence Intervals

• Central Limit Theorem:

  • Deriving confidence intervals

  • Hypothesis testing

• Paired r-test for Comparing Learning Algorithms

Module 5: Support Vector Machines, Dimensionality Reduction, and Anomaly Detection

• Support Vector Machines (SVM):

  • Large margin classification

  • Applications of SVM

  • Linear and Non-linear SVM

  • Kernels

• Dimensionality Reduction:

  • Data Compression

  • Principal Component Analysis (PCA)

  • Choosing the number of principal components

• Unsupervised Learning and Anomaly Detection:

  • k-Means Algorithm:

    • Optimization objective

    • Choosing the number of clusters

    • Hierarchical Agglomeration

  • Scope of Anomaly Detection:

    • Gaussian Distribution

    • Developing and Evaluating an Anomaly Detection System

    • Anomaly Detection vs. Supervised Learning

    • Choosing features

    • Multivariate Gaussian Distribution

    • Anomaly detection using Multivariate Gaussian Distribution

References

1. Ethem Alpaydin, Introduction to Machine Learning, Second Edition, The MIT Press, Cambridge, Massachusetts, London, England, 2010.

2. Tom M. Mitchell, Machine Learning, McGraw Hill Education, 2017.

3. Christopher M. Bishop, Pattern Recognition and Machine Learning, Springer, 2011.

4. Trevor Hastie, Robert Tibshirani, Jerome H. Friedman, The Elements of Statistical Learning, Springer, 2nd Edition, 2009.

5. Max Kuhn and Kjell Johnson, Applied Predictive Modelling, Springer, 2013.

6. Sebastian Raschka and Vahid Mirjalili, Python Machine Learning, Packt Publishing, 2017.

CO-PO Mapping

Module 1: Definition and Applications of Machine Learning (6 Lectures)

• Lecture 1: Introduction to Machine Learning (ML)

• Lecture 2: Overview of ML Algorithms

• Lecture 3: Supervised Learning – Regression and Classification

• Lecture 4: Cost Function and Gradient Descent for Linear Regression

• Lecture 5: Learning Rate and Optimization in Gradient Descent

• Lecture 6: Introduction to Unsupervised Learning

Module 2: Linear Algebra and Advanced Regression (8 Lectures)

• Lecture 7: Basics of Linear Algebra

• Lecture 8: Eigenvalues, Eigenvectors, and their Significance

• Lecture 9: Singular Value Decomposition

• Lecture 10: Linear Regression with Multiple Variables

• Lecture 11: Normal Equation and Non-invertibility

• Lecture 12: Introduction to Logistic Regression for Classification

• Lecture 13: Decision Boundary in Logistic Regression

• Lecture 14: Multiclass Classification

Module 3: Neural Networks (7 Lectures)

Lecture 15: Introduction to Neural Networks

Lecture 16: McCulloch-Pitts Neuron Model

Lecture 17: Multi-layer Neural Networks

Lecture 18: Learning in Neural Networks

Lecture 19: Perceptron Learning Rule

Lecture 20: Delta Learning Rule

Lecture 21: Backpropagation Algorithm

Module 4: Hypothesis Evaluation and Statistical Theory (7 Lectures)

• Lecture 22: Evaluating Hypotheses:

• Lecture 23: Sample Error vs. True Error

• Lecture 24: Basics of Sampling Theory

• Lecture 25: Estimators: Bias and Variance

• Lecture 26: Confidence Intervals and the Central Limit Theorem

• Lecture 27: Hypothesis Testing

• Lecture 28: Paired r-Test for Comparing Learning Algorithms

Module 5: SVM, Dimensionality Reduction, and Anomaly Detection (7 Lectures)

• Lecture 29: Support Vector Machines (SVM)

• Lecture 30: Linear and Non-linear SVM, Kernels

• Lecture 31: Dimensionality Reduction

• Lecture 32: k-Means Algorithm

• Lecture 33: Hierarchical Agglomeration and Scope of Anomaly Detection

• Lecture 34: Gaussian Distribution and Anomaly Detection System Development

• Lecture 35: Multivariate Gaussian Distribution for Anomaly Detection

• Review Paper: 15 marks

• Assignments : 5 marks

• Seminar: 5 marks

• Quiz(es): 5 marks

• Internal Examination: 10 marks

CO  5

Rubrics for Assessment of Review Paper 

Attendance and Internal Marks

• SlideDeck_Module 1

• SlideDeck_Module 2 

Assignment 1: Predicting House Prices using Regression

Submit your assignments here. Create sepearete folders for each student. Folder name should be <YourFirstName>_<YourRollNo.>

Question. All questions carry equal marks. 

Q1: CO2, Q2: C)3, Q3: CO3, Q4: CO2, Q5: CO5