Course

Unit 1

10 Hours

Basic motivation, examples of machine learning applications, Supervised and Unsupervised Learning – Review linear algebra, vector spaces, linear transformations, eigenvalues, and vectors – Review of statistics and probability theory, random variables, and probability distributions. Basic concepts of fuzzy sets – Operations on fuzzy sets –Fuzzy relation equations – Fuzzy logic control – Fuzzification – Defuzzification – Knowledge base – Decision making logic – Membership functions – Rule base.

Unit 2

10 Hours

Multiple Variable Linear regression, Multiple regression, Logistic regression, K-NN classification, Naive Bayes classifiers, and Support vector machines. K-means clustering, Hierarchical clustering, High-dimensional clustering, Dimension ReductionPCA, Ensemble techniques Decision Trees, Random Forests, Bagging, Boosting-Value based methods Q-learning. Reinforced learning.

Unit 3

10 Hours

Introduction – history of neural networks – multilayer perceptrons –Back propagation algorithm and its variants – Different types of learning, examples, Deep learning – Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM). Generative Adversarial Networks (GANs). Model Evaluation and Validation – Cross-validation techniques, Evaluation metrics for regression and classification tasks, Bias-variance tradeoff, Feature Engineering, and Model Optimization.

Lab Practice

12 Sessions

• Pattern recognition-based online monitoring system for machinery fault diagnosis using support vector
  machine.
• Decision tree assisted selection of materials for electric vehicle chassis.
• Predicting the optimal input parameters for the desired print quality using an artificial neural network.
• Exploration of the K-NN algorithm to predict fatigue strength of steel from composition and processing parameters.
• Prediction of remaining useful life of machine component using Support Vector Regression and LSTM
• Generation of 3D CAD model for mechanical parts using Generative Adversarial Networks (GAN)
• A deep learning approach for detection of obstacles for autonomous driving systems using CNN.
• A multi-sensor information fusion for fault diagnosis of a mechanical system utilizing discrete wavelet features.
• Physics-informed machine learning-based fault diagnosis of machine elements.
• Prediction of weld quality using image processing techniques.

Course Objectives

• Provide a strong foundation of fundamental concepts in Artificial Intelligence
• Elobarate different AI and machine learning techniques for design of AI systems.

Course Outcomes

• CO1: Understand the basics of probability and statistical learning for artificial intelligence
• CO2: Apply AI and ML techniques which involve perception, reasoning and learning
• CO3: Analyze a real world problems and solve it using machine learning and deep learning techniques
• CO4: Develop ML models using advanced techniques for various automation applications

CO-PO Mapping

Skills Acquired

Formulate engineering problems as a machine learning problem; Select appropriate solution methods and strategies to
solve machine learning problems; Solve engineering design-related machine learning problems using software tools.

1. Chandra S.S.V Artificial Intelligence and Machine Learning, Prentice Hall India Learning Private Limited; 4th edition, 2018.
2. Tom M. Mitchell, “Machine Learning”, McGraw Hill, 1997.
3. Ethem Alpaydin, “Introduction to Machine Learning”, MIT Press, 2015.
4. C. M. Bishop, “Pattern Recognition and Machine Learning”, Springer, 2006.
5. C. Muller and S. Guido, “Introduction to Machine Learning with Python”, O’Reilly Media, 2017.
6. Goodfellow, YoshuaBengio and Aeron Courville,” Deep Learning”, MIT Press, First Edition, 2016.
7. Guttag, John., “Introduction to Computation and Programming Using Python: With Application to Understanding Data”, Second Edition. MIT Press, 2016.