This course offers an advanced and comprehensive study of machine learning techniques with an emphasis on theoretical foundations, algorithmic development, and practical implementation. The course begins with advanced machine learning using TensorFlow, introducing tensor operations, computational graphs, activation and loss functions, backpropagation, model training strategies, evaluation methods, and considerations for production-level deployment.
The curriculum further examines ensemble learning methods, including bagging, random forests, boosting, AdaBoost, and CatBoost, highlighting their role in improving model accuracy and robustness through the combination of multiple learners. Reinforcement learning is covered in detail, encompassing the formal learning framework, policy learning components, value-based methods, Q-learning, Deep Q-Networks (DQN), and advanced architectures such as double and dueling DQNs, along with discussions on generalization, feature selection, and bias–variance trade-offs.
The course also addresses model evaluation and hyper-parameter tuning techniques, including cross-validation, performance metrics, learning and validation curves, grid search, and statistical validation methods. Finally, it introduces machine learning deployment practices, focusing on model serialization, development of web-based applications using Flask, and deployment to cloud platforms such as AWS and Google Cloud.
Upon successful completion of the course, students will be equipped to design, evaluate, optimize, and deploy advanced machine learning models for real-world applications in academic and industrial contexts.
- Teacher: Dr. Smriti Mishra