לוח שנה משולב

Living systems utilize fundamental physics in the form of mechanical forces and geometric cues to move and change shape.  A central question motivating our research is: how does biological matter utilize mechanical forces to form ordered structures and change shape? As a prototype of active biological materials capable of self-organized shape change, we explain experimental findings on cytoskeletal gel extracts by our collaborators at the Bernheim laboratory. Despite having identical composition of the biopolymer actin, molecular motor myosin and the crosslinker fascin, these gels contract and buckle into different shapes depending on the initial gel aspect ratio: thinner gels tend to wrinkle, while thicker gels tend to form domes. By incorporating motor-generated active stresses, alignment of active fibers, and stress-dependent myosin binding kinetics into a network-fluid (poroelastic) model, we qualitatively capture the observed trends in gel contraction dynamics measured using particle image velocimetry (PIV). We then show how a geometric elastic model for thin sheets can relate the 3D buckled shapes to strain rates predicted by the poroelastic model. Our findings have implications for shape changes during tissue morphogenesis and bio-inspired soft materials design.

The field of artificial intelligence (AI) is undergoing a paradigm shift, moving from neural networks trained for narrowly defined tasks (e.g., image classification and machine translation) to general-purpose models such as ChatGPT. These models are trained at unprecedented scales to perform a wide range of tasks, from providing travel recommendations to solving Olympiad-level math problems. As they are increasingly adopted in society, a central challenge is to ensure the alignment of general-purpose models with human preferences. In this talk, I will present a series of works that reveal fundamental pitfalls in existing alignment methods. In particular, I will show that they can: (1) suffer from a flat objective landscape that hinders optimization, and (2) fail to reliably increase the likelihood of generating preferred outputs, sometimes even causing the model to generate outputs with an opposite meaning. Beyond characterizing these pitfalls, our theory provides quantitative measures for identifying when they occur, suggests preventative guidelines, and has led to the development of new data selection and alignment algorithms, validated at large scale in real-world settings. Our contributions address both efficiency challenges and safety risks that may arise in the alignment process. I will conclude with an outlook on future directions, toward building a practical theory in the age of general-purpose AI.

Short bio:

Noam Razin is a Postdoctoral Fellow at Princeton Language and Intelligence, Princeton University. His research focuses on the fundamentals of artificial intelligence (AI). By combining mathematical analyses with systematic experimentation, he aims to develop theories that shed light on how modern AI works, identify potential failures, and yield principled methods for improving efficiency, reliability, and performance.

Noam earned his PhD in Computer Science at Tel Aviv University, where he was advised by Nadav Cohen. Prior to that, he obtained a BSc in Computer Science (summa cum laude) at The Hebrew University of Jerusalem under the Amirim honors program. For his research, Noam received several honors and awards, including the Zuckerman Postdoctoral Scholarship, the Israeli Council for Higher Education (VATAT) Postdoctoral Scholarship, the Apple Scholars in AI/ML PhD fellowship, the Tel Aviv University Center for AI and Data Science excellence fellowship, and the Deutsch Prize for PhD candidates.

The presence of bacteria in solid human tumors has been documented for over a century. However, only in recent years has a more comprehensive characterization of this low-biomass microbiome been undertaken. We have been characterizing the presence of bacteria and fungi across a wide range of human tumor types and have begun to dissect their functional roles and clinical relevance, including their impact on responses to therapy. In this seminar, I will provide a brief overview of the current understanding of the multi-kingdom tumor microbiome landscape and present our findings on its potential effects on cancer therapy.