Understanding what individual units in a neural network represent is a cornerstone of mechanistic interpretability. A common approach is to generate human-friendly text explanations for each neuron to describe their functionalities—but how can we trust that these explanations are faithful reflections of the model’s actual behavior?
Understanding what individual units in a neural network represent is a cornerstone of mechanistic interpretability. A common approach is to generate human-friendly text explanations for each neuron to describe their functionalities—but how can we trust that these explanations are faithful reflections of the model’s actual behavior? In work published in the 2025 International Conference on […]
MIT News || A new study led TILOS Foundations team member Stefanie Jegelka and graduate student Behroozi Tahmasebi presents the first provably efficient method for machine learning with symmetric data, which could advance neural network design for applications ranging from drug discovery to climate modeling.
MIT News || A team of MIT researchers, including TILOS Foundations team member and associate professor Stefanie Jegelka, and postdoctoral scholar Yifei Wang, has developed a theoretical framework to study how information flows through the machine learning (ML) architecture that forms the backbone of LLMs. Their work has uncovered the root cause of “position bias” […]
CoinDesk || National Science Foundation funding cuts threaten to devastate U.S. crypto research, say 10 leading professors, including TILOS Chips team co-lead Farinaz Koushanfar, the Nemat-Nasser Endowed Chair Professor of Electrical and Computer Engineering at UC San Diego, and founding co-director of the UC San Diego Center for Machine Intelligence, Computing, and Security (MICS).
The Robot Report || The Congressional Robotics Caucus had not convened formally since 2019, according to TILOS Robotics team co-lead Henrik Christensen, who is also the Qualcomm Chancellor’s Chair of Robot Systems and a distinguished professor of computer science at the Department of Computer Science and Engineering at UC San Diego.
RSS Pioneers is an intensive workshop for senior PhD students and early career researchers in the robotics community. Held in conjunction with the main Robotics: Science and Systems (RSS) conference, each year the RSS Pioneers brings together a cohort of the world’s top early-career researchers.
Imagine developing a finer control knob for artificial intelligence (AI) applications like Google Gemini and OpenAI ChatGPT. Mikhail Belkin, a professor with UC San Diego’s Halıcıoğlu Data Science Institute (HDSI) – part of the School of Computing, Information and Data Sciences (SCIDS) – has been working with a team that has done just that. Specifically, […]
The IEEE Information Theory Society Thomas M. Cover Dissertation Award is awarded annually to the author of an outstanding doctoral dissertation. Neha Sangwan has been selected as the recipient of the 2025 Thomas M. Cover Dissertation Award for her dissertation entitled “Communication with Byzantine Users”. This award is given annually to the author of an […]
All presentations from last week’s Horizons for Optimization in AI Workshop are now available to view!
