Winter School on Deep Learning

The Winter School on Deep Learning 2025 (WSDL 2025) offers self-motivated students in-depth knowledge and exposure to research in Deep Learning, organized by the Electronics and Communication Sciences Unit, at the Indian Statistical Institute, Kolkata. The school offers hands-on courses to apply the latest Deep Learning methods to real-world problems. The school also covers the essential concepts of deep learning and the fundamentals of mathematics and programming. Come join us as we take the journey from GenAI, LLMs, and Beyond: A Fusion of Frontiers! Course coverage: - Basics of Python and Libraries of Importance. - Basics of Deep Learning Library: PyTorch. - Rudiments of Probability Theory for Machine Learning. -Essentials of Matrix Calculus and Linear Algebra for Machine Learning. -Bird’s Eye View of Machine Learning. - Primer on Text, Video, and Image Data processing. -Gradient-based Optimization techniques. -Rudiments of Artificial Neural Networks and Backpropagation of Error. -Tree-based Classifiers and ensemble techniques. -Steps towards Deep Learning: Activation Functions, Normalization techniques, Regularization methods, and loss functions. -Convolutional Neural Networks. - Architectures of Deep Neural Network Models. -Generative Deep Neural Network Models: GANs, VAEs, and Diffusion. -Recurrent Neural Networks and Backpropagation through Time. -Attention mechanism and Transformers: Where Text meets Vision. -Large Language Models, pre-training, task adaptation, and fine-tuning. -Introduction to Graph Neural Networks. -Explainable Artificial Intelligence -Deep Clustering techniques. -Emerging Learning strategies: Semi-supervised, few-shot, zero-shot, etc. -Adversarial Attacks, Defence, and Robust Deep Neural Network models. -Theory of Deep Learning: Special Focus on GenAI. -Deep Reinforcement Learning. -Real-world Applications: From Problem to a Solution