| CODE | ARI2205 | ||||||||||||
| TITLE | Interpretable Artificial Intelligence | ||||||||||||
| UM LEVEL | 02 - Years 2, 3 in Modular Undergraduate Course | ||||||||||||
| MQF LEVEL | 5 | ||||||||||||
| ECTS CREDITS | 5 | ||||||||||||
| DEPARTMENT | Artificial Intelligence | ||||||||||||
| DESCRIPTION | This study-unit will introduce students to techniques related to the interpretation and merely explanation of the decisions made by machine learning models. They will explore what interpretable AI is and why it is essential for modern data science applications. They will also be exposed to various types of interpretability techniques, including: i) intrinsic techniques associated with machine learning models that are transparent by design (white-box models), ii) post-hoc techniques associated with more complex black-box models, iii) local techniques aiming at understanding models' behaviour for a specific example, iv) global techniques focusing on the global effects of the input on the models' output, and v) model-specific versus model-agnositc techniques. Moreover, the students will learn how to apply a suite of techniques for building interpretable AI systems for several case studies. The topics that will be covered include: - Techniques for interpreting white-box models (Linear regression, Decision trees, GAMs); - Techniques for interpreting black-box tree ensembles (Partial dependence plots, Feature interaction plots); - Techniques for interpreting black-box Deep Neural Networks (Local interpretable model-agnostic explanation (LIME), Shapley additive explanations (SHAP), Anchors); - Fair AI (Interpretability techniques and fairness, Bias mitigation techniques, Dataset documentation. Study-unit Aims: The aims of this study-unit are to: - Give students the necessary fundamentals of interpretable AI techniques to gain insights and explain machine learning models' behaviour; - Provide the student with the skills to apply interpretable AI techniques for building AI systems targeting real-world problems; - Expose the students to the dominant technologies used for building interpretable AI systems. Learning Outcomes: 1. Knowledge & Understanding By the end of the study-unit the student will be able to: - Explain what interpretable and fair AI is; - Outline the various properties of interpretable AI techniques; - Discuss the pros and cons of white-box and black-box models; - Evaluate a repertoire of interpretable AI techniques to be used in different scenarios; - Recognise biased datasets and explain their implications on the employed machine learning models. 2. Skills By the end of the study-unit the student will be able to: - Interpret and communicate the outputs of white-box and black-box machine learning models; - Determine the appropriate interpretation technique to use in a variety of different contexts; - Build interpretable AI systems for real-world problems; - Use bias mitigation techniques to compile fair datasets. Main Text/s and any supplementary readings: - Thampi, Ajay. Interpretable AI: Building Explainable Machine Learning Systems. Simon and Schuster, 2022. - Gilpin, Leilani H., et al. "Explaining explanations: An overview of interpretability of machine learning." 2018 IEEE 5th International Conference on data science and advanced analytics (DSAA). IEEE, 2018. - Carvalho, Diogo V., Eduardo M. Pereira, and Jaime S. Cardoso. "Machine learning interpretability: A survey on methods and metrics." Electronics 8.8 (2019): 832. - Linardatos, Pantelis, Vasilis Papastefanopoulos, and Sotiris Kotsiantis. "Explainable ai: A review of machine learning interpretability methods." Entropy 23.1 (2020): 18. |
||||||||||||
| STUDY-UNIT TYPE | Lecture, Independent Study & Tutorial | ||||||||||||
| METHOD OF ASSESSMENT |
|
||||||||||||
| LECTURER/S | Konstantinos Makantasis |
||||||||||||
|
The University makes every effort to ensure that the published Courses Plans, Programmes of Study and Study-Unit information are complete and up-to-date at the time of publication. The University reserves the right to make changes in case errors are detected after publication.
The availability of optional units may be subject to timetabling constraints. Units not attracting a sufficient number of registrations may be withdrawn without notice. It should be noted that all the information in the description above applies to study-units available during the academic year 2024/5. It may be subject to change in subsequent years. |
|||||||||||||