Machine Learning Concepts and Systems

This unit introduces core machine learning concepts, the hardware and data practices that enable them, and practical application areas such as NLP, robotics, drones, industrial systems, and space operations. Students explore probabilistic reasoning, neural network basics, accelerators (GPUs and DLPs), data preprocessing, and evaluate real-world feasibility and operational constraints to prepare for advanced application and system design in subsequent units.

Learning Objectives

• Analyze the roles and tradeoffs of hardware accelerators (GPUs and DLPs) for training and inference by comparing performance, cost, and energy characteristics for representative ML tasks
• Apply probability concepts and basic neural network architectures to design, train, and evaluate simple supervised and unsupervised models using standard metrics and validation techniques
• Demonstrate data curation and preprocessing workflows (cleaning, labeling, feature selection/engineering) for ML pipelines tailored to scenarios such as robotics, drones, and industrial sensing
• Evaluate the feasibility, ethical considerations, and operational constraints of AI applications (NLP, computer vision, robotics, drones, space systems), including historical impacts like AI winters and strategies for human augmentation