Advanced Topics in Knowledge Representation and Reasoning

COMP64602 Advanced Topics in Knowledge Representation and Reasoning 15 Semester 2 Postgraduate Taught Level 6 Louise Dennis Unit coordinator 1 FHEQ level (Framework for Higher Education Qualifications) ' Masters/Integrated Masters P4 ' European Credit Transfer & Accumulation System Rating : 7.5 Knowledge Representation and Reasoning (KRR) has been fundamental to the study of Artificial Intelligence, Autonomous Systems and Information Systems since the inception of the field. It is founded on concepts of logical reasoning drawn from mathematics and philosophy. This unit draws on a foundation in logic to show how these KRR techniques can be applied when information is uncertain, where contradictory knowledge and probabilistic reasoning are required to be considered, or changing, where it must be used to plan a series of actions over-time, and how it can be used to manage interactions between multiple agents. Building on Logics for Knowledge Representation and Reasoning, this unit explores a range of advanced topics in Knowledge Representation and Reasoning, including important techniques for using a knowledge base in applications such as link prediction, web agents, and robotic autonomous systems. These topics include dealing with contradictory knowledge (e.g., birds fly, penguins are birds, penguins don't fly) and using knowledge bases for planning and in multi-agent systems (e.g., one of several robots with their own beliefs, goals, and understanding of the world). The content of the module will focus on three important themes: techniques for handling contradictory, uncertain and changing information such as argumentation theory, non-monotonic logics and probabilistic reasoning; techniques which allow computational agents to use knowledge to plan sequences of actions, accounting for uncertainty and possible failures; and techniques which enable distributed agents to coordinate their actions without revealing all their own information and goals. This unit aims to introduce students to state-of-the-art techniques for representing and reasoning about knowledge, as well as important computational paradigms for using such knowledge to achieve tasks. 1. Describe and discuss the applicability of emerging Knowledge Representation and Reasoning topics. 2. Represent uncertain and contradictory information for computational processing.3. Represent problems as AI planning problems. 4. Apply reasoning algorithms for uncertain knowledge.5. Apply the agent paradigm to the design and implementation of computational systems. 6. Apply and evaluate planning algorithms. Analytical skills Innovation/creativity Problem solving Written communication <ul><li>Uncertain and Probabilistic Reasoning</li><li>Non-monotonic Reasoning</li><li>Multi-Agent Communication and Coordination</li><li>AI Planning</li><li>Reasoning about Agent Programs</li></ul> Weekly asynchronous resources in the form of video lectures and assigned reading. Weekly formative coursework in the form of quizzes and lab exercises. Weekly optional drop-in labs or tutorials for assistance with coursework. Weekly synchronous lectures to summarise the material and provide an opportunity for Q&A. 1 Written exam 70% 2 Written assignment (inc essay) 15% 8 Practical skills assessment 15% Cohort-level feedback after marking.Individualised feedback via marking rubrics.Individualised feedback on request in Labs.Autograded weekly quizzes. COMP64401 Logics for Knowledge Representation and Reasoning Pre-Requisite Recommended Y Stuart J. Russell and Peter Norvig, Artificial Intelligence: A Modern Approach, Pearson, 2021. Gerhard Weiss, Multiagent Systems, MIT Press, 2016. Assessment written exam 2 Lectures 10 Practical classes & workshops 10 0 128 Summative Coursework 16 hoursFormative Coursework 20 hoursFormative Quizzes 6 hoursVideos 10 hoursDirected reading 10 hoursIndependent study, consolidation and revision 66 hours