Select a calendar:
Filter November Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for November 28, 2023
-
Thesis Proposal (Sasha Volokh)
Tue, Nov 28, 2023 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Thesis Proposal Committee Members:
William G.J. Halfond (Chair)
Nenad Medvidovic
Andrew Nealen
Mukund Raghothaman
Chao Wang
Abstract:
Modern computer games often release with significant bugs, causing consumer dissatisfaction and a loss of business and reputation for the companies involved. Testing is a key mechanism by which these issues can be caught and addressed during development. A key requirement for thorough manual and automated testing of games is knowledge of the possible player actions and their associated device inputs. In this thesis I propose novel program analysis techniques to inform both automated testing agents and human testers of the possible game actions. First, I propose a symbolic analysis technique that automatically analyzes the user input handling logic present in games to determine a discrete action space, along with the conditions under which the actions are valid, and the device inputs associated with each action. I then demonstrate how this technique can be adapted to enable effective performance in agents that automatically explore game functionalities. Next, I propose adapting this technique for game playing reinforcement learning agents. Finally, I propose methods to automatically generate in-game instructions for human testers based on the outcome of the action analysis.Location: Charles Lee Powell Hall (PHE) - 325
Audiences: Everyone Is Invited
Contact: CS Events
-
Thesis Proposal (Han Zhang)
Tue, Nov 28, 2023 @ 12:00 PM - 01:00 PM
Thomas Lord Department of Computer Science
University Calendar
Thesis Proposal Committee Members:
Sven Koenig (Chair)
Satish Kumar Thittamaranahalli
Lars Lindemann
Satyandra Kumar Gupta
Ariel Felner
Title: Speeding-up Multi-Objective Search Algorithms
Abstract: In the Multi-Objective Search problem, given a graph in which each edge is annotated with a cost vector, a start state, and a goal state, a typical task is to compute a Pareto frontier. State-of-the-art multi-objective search algorithms conform to the same best-first algorithmic framework. These algorithms are similar to best-first search algorithms, such as A*, but, most differently, they need to consider multiple nodes (with costs that do not dominate each other) for the same state. Due to the similarity between multi-objective and single-objective search algorithms, I hypothesize that one can speed up multi-objective search algorithms by applying insights gained from single-objective search. More specifically, I propose to speed up multi-objective search algorithms by (1) sacrificing solution optimality, (2) using preprocessing techniques, and (3) using efficient data structures for dominance checks.Location: Hughes Aircraft Electrical Engineering Center (EEB) - 110
Audiences: Everyone Is Invited
Contact: CS Events
-
Thesis Proposal (Hejia Zhang)
Tue, Nov 28, 2023 @ 03:00 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Thesis Proposal Committee members:
Stefanos Nikolaidis
C.C.-Jay Kuo
Jyo Deshmukh
Jesse Thomason
Daniel Seita
Title: Understanding, Learning and Planning for Long-horizon Collaborative Manipulation Tasks
Abstract: Robots that assist humans in their daily activities have to perform long-horizon manipulation tasks, such as cooking, table setting tasks, effectively and collaboratively. To successfully perform these tasks, robots have to address the problem of generating both high-level task action sequences and low-level executable motion trajectories, which is known as the Task-and-Motion Planning (TAMP) problem. In this thesis, we first explore how robots can understand and imitate human collaborative manipulation task plans by watching YouTube videos. We then study the problem of robots executing specified high-level task goals in any unstructured environments. We specifically focus on a subclass of the TAMP problem, namely the Geometric Task-and-Motion Planning (GTAMP) problem. We present a framework that allows robots to perform GTAMP tasks collaboratively. Finally, we discuss the proposed work that will potentially allow robots to collaborate with humans to perform long-horizon collaborative manipulation tasks in the real world.Location: Hughes Aircraft Electrical Engineering Center (EEB) - 110
Audiences: Everyone Is Invited
Contact: CS Events
-
CS Colloquium: Niloufar Salehi (UC Berkeley) - Designing Reliable Human-AI Interactions
Tue, Nov 28, 2023 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Niloufar Salehi, UC Berkeley
Talk Title: Designing Reliable Human-AI Interactions
Abstract: How can users trust an AI system that fails in unpredictable ways? Machine learning models, while powerful, can produce unpredictable results. This uncertainty becomes even more pronounced in areas where verification is challenging, such as in machine translation or probabilistic genotyping. Providing users with guidance on when to rely on a system is challenging because models can create a wide range of outputs (e.g. text), error boundaries are highly stochastic, and automated explanations themselves may be incorrect. In this talk, I will focus on the case of health-care communication to share approaches to improving the reliability of ML-based systems by designing actionable strategies for users to gauge reliability and recover from potential errors.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Niloufar Salehi is an assistant professor in the School of Information at UC, Berkeley and faculty member of Berkeley AI Research (BAIR). Her research interests are in social computing, human-centered AI, and more broadly, human-computer interaction (HCI). Her research is in close collaboration with partners and domain experts spanning education to healthcare to restorative justice. Her work has been published and received awards in premier venues including ACM CHI and CSCW and has been covered in VentureBeat, Wired, and the Guardian. She is a W. T. Grant Foundation scholar. She received her PhD in computer science from Stanford University in 2018.
Host: Souti Chattopadhyay
Location: Olin Hall of Engineering (OHE) - 136
Audiences: Everyone Is Invited
Contact: CS Faculty Affairs
