Provides a foundational treatment of modern machine learning at the masters level, complemented with comprehensive projects that involve implementing the state-of-the-art machine learning algorithms on several datasets. The course topics are divided into 3 categories: 1) Classics of learning, including Bayesian decision theory, maximum likelihood estimation, Fisher's discriminant analysis, nearest neighbor classification, and support vector machines, clustering and dimensionality reduction, 2) Modern approaches to learning, including neural networks, deep learning networks, generative adversarial networks, autoencoders, and transformers, 3) Engineering perspectives on learning, including the effects of quantization and hardware implementation challenges, introduction to neuromorphic hardware and learning, and introduction to secure machine learning.

This applied introduction to data science course equips students with foundational skills for working with real-world data. Building on courses in statistical methods, numerical methods, and foundations of machine learning, students will learn techniques for data handling, cleaning, transformation, and visualization, with practical applications in healthcare, finance, and marketing. Topics include data integration, interactive dashboards, advanced visualizations, and the use of predictive and classification models. Hands-on exercises and case studies support the development of analytical and modeling skills. The course also addresses ethical issues such as bias, fairness, and privacy. By the end, students will be able to clean and prepare complex datasets, create effective visualizations, build dashboards, apply machine learning models, and complete a capstone project demonstrating their abilities.

With the growing use of machine learning and generative AI in high-stakes applications, ensuring these models are trustworthy, responsible, and sustainable is critical. This course explores five pillars of trustworthy AI: (i) explainability and interpretability to support human understanding, (ii) robustness against adversarial attacks and real-world uncertainty, (iii) ethical alignment with human values, (iv) privacy-preserving methods for confidentiality and compliance, and (v) sustainability through techniques like model compression and distillation. The course also introduces large language models (LLMs) and emerging approaches to enhance their trust, accountability, and sustainability.

This course introduces graduate engineering students to the foundations and frontiers of Large Language Models (LLMs), with an emphasis on both conceptual understanding and practical skills. We begin with core topics in natural language processing, tokenization, and language representation, then explore the transformer architecture, attention mechanisms, and the full LLM pipeline--from training to deployment. Students will examine key developments such as in-context learning, emergent capabilities, and prompt engineering, as well as multimodal extensions like Vision-Language Models (VLMs). Later weeks will address ethical considerations and recent trends in unified multimodal models. Through a mix of lectures, discussions, and hands-on assignments, students will be well-prepared to engage with cutting-edge research and real-world applications in generative AI.