Exclusively available to PGP-AIML learners

Reinforcement Learning (RL) is one of the three fundamental paradigms of Machine Learning, inspired by psychology and neuroscience, focused on developing agents that take actions in an environment to achieve goals. Unlike supervised learning, RL does not rely on labelled input/output pairs; instead, agents learn by balancing exploration and exploitation to discover optimal policies. In this unit, students will explore and implement solutions to a range of RL problems and Markov Decision Processes (MDPs), including advanced variants and techniques. Key Highlights: • Work with MDP variants such as discrete-time MDPs, Semi-MDPs (SMDP), continuous-time MDPs, POMDPs, and MOMDPs • Apply core RL techniques, including multi-armed bandits, value iteration, policy gradient, temporal difference learning, and reward design • Understand advanced concepts such as on-policy vs off-policy learning, eligibility traces, feature construction, and continuous action spaces • Explore emerging areas, including deep RL, multi-agent systems, transfer learning, hierarchical and curiosity-driven learning

This module focuses on the end-to-end process of designing, developing, and operationalising AI solutions. It highlights how AI solution development differs from traditional software engineering while equipping learners with the skills to apply engineering principles, conduct experiments, and effectively manage stakeholder expectations throughout the AI lifecycle. Key Highlights: • Understand the process and key characteristics of developing AI solutions and how they differ from traditional software development • Design, develop, deploy, and maintain AI solutions using modern tools, frameworks, and libraries • Apply engineering principles and the scientific method with appropriate rigour in experimentation • Manage stakeholder expectations and guide the operationalisation of AI solutions from inception to deployment and ongoing maintenance

This module explores how robotics systems sense and interact with their environment using computer vision and speech processing techniques. Students will analyse existing algorithms, study their real-world applications, and investigate state-of-the-art machine learning methods to develop innovative solutions in this domain. Key Highlights: • Understand how computer vision and speech processing enable sensing and interaction in robotics • Analyse existing algorithms and their applications in real-world robotic scenarios • Investigate state-of-the-art machine learning techniques used in vision and speech domains • Develop novel solutions integrating computer vision and speech processing for robotics applications

Deep learning is a transformative technology in data science and artificial intelligence. This module enables students to build practical knowledge of deep learning concepts and applications, focusing on model development for both structured and unstructured data, such as images, videos, and text. Key Highlights: • Understand core deep learning theories, including computational graphs and representation learning • Build deep learning models for structured and unstructured data • Learn key techniques such as convolutional neural networks, recurrent networks, and neural embedding methods • Explore real-world applications of deep learning widely adopted across industries

This module examines the importance of aligning AI systems with human values as their deployment in society continues to grow. It focuses on the ethical, safe, explainable, and interactive aspects of AI, explored through both philosophical perspectives and practical implementation. Key Highlights: • Understand the need for aligning AI with human requirements, including ethics, safety, and explainability • Explore concepts such as artificial general intelligence, superintelligence, consciousness, and ethical decision-making • Study methods including safe exploration, constrained AI, interpretability, transparency, and interactivity • Analyse emerging areas such as AI pedagogy, industry standards, and black, grey, and white box systems, with an emphasis on ongoing research beyond the unit content

This module focuses on the foundational mathematical concepts essential for understanding and solving problems in artificial intelligence. It enables students to interpret, apply, and evaluate mathematical techniques used across AI applications. Key Highlights: • Explain the role and application of mathematical concepts associated with artificial intelligence • Identify and summarise key mathematical concepts and techniques required to solve AI-related problems • Verify and critically evaluate results, and communicate findings to a range of audiences • Read and interpret mathematical notation and clearly communicate problem-solving approaches