EE260B: Trustworthy AI for
Autonomous Systems
Winter 2024
Course Overview
This course aims to provide you with a comprehensive understanding of the principles and practices that underpin the development of safe and reliable AI for autonomous system applications. The course objectives include exploring the foundational and advanced concepts of trustworthy artificial intelligence and machine learning, with a particular focus on their applications in autonomous systems such as autonomous vehicles, drones, and robotic systems. Students will learn to assess and ensure the safety, reliability, and ethical implications of AI algorithms in these contexts. The course will also delve into the latest research and methodologies for creating AI systems that are safe, robust, generalizable, and explainable, emphasizing the importance of trust and ethical considerations in AI deployment.
Through a combination of theoretical knowledge and practical case studies, you will be equipped to design, analyze, and implement AI solutions that are not only technically sound but also socially responsible and trustworthy. There will be a final course project for students to obtain hands-on research experience in related areas.
Topics Covered
Deep Learning (Concise Overview)
Deep Reinforcement Learning (Concise Overview)
Out-of-distribution (OOD) Generalization
Explainability and Interpretability
Uncertainty Estimation/Quantification
Safety and Robustness
Language Models, World Models, and Agent Models
Course Details
Lectures
Wednesdays & Fridays: 3:30 PM - 4:50 PM Dates: 01/10 - 03/15 Classroom: Room 308, Student Success Center (SSC)
Office Hours
Instructor (Prof. Jiachen Li): Tuesdays: 3:00 PM - 3:50 PM Location: WCH 315 Email: jiachen.li@ucr.edu
Teaching Assistant (Rohit Lal): Thursdays: 3:30 PM - 4:30 PM Location: WCH 371 Email: rlal011@ucr.edu
Prerequisites
Familiarity with Python and PyTorch coding.
Basic maths: multivariate calculus, linear algebra, probability, statistics, and optimization. These mathematical concepts are fundamental to understanding and implementing AI and ML algorithms.
Basic knowledge of machine learning and deep learning concepts and techniques.
You should be able to read conference or journal papers on robotics, deep learning, and reinforcement learning and have a decent understanding of the basic ideas and concepts proposed (not necessarily a complete understanding of every detail).
Grading Policy
Paper Presentation (15%): Evaluates understanding of research papers and the presentation of key contributions and methodologies.
Paper Reviews (20%): You are expected to submit a critical review/summary of the papers that are to be presented in the class. You will submit two paper reviews. Note that these two papers should be different from the one you present during the lecture.
Final Project (40%): Assesses research capabilities, requiring strong coding skills and analytical efforts.
Allocation: project proposal (5%); final report (25%); final presentation (10%).
Midterm Exam (15%): Assesses understanding of concepts and knowledge covered before midterm. The exam will tentatively happen on 2/16/2024 during the lecture.
Class Attendance (10%): Lecture attendance is required since it is one of the most important components of the course. Asking questions after paper presentations or final project presentations will receive a bonus (up to 3%).
References
There is no required textbook for this course. An in-depth understanding of the lecture slides and the set of selected papers will be sufficient for you to succeed in this course. Note there is no existing textbook that covers such a collection of topics and materials covered in this course.
Additional Recommended References: