לוח שנה משולב

Much of the thinking about neural population codes was motivated in the past decades by reports on neurons with highly stereotyped tuning functions. Indeed, neurons are often observed to exhibit smooth, unimodal tuning to an encoded variable, centered around preferred stimuli that vary across the neural population. Recent experiments, however, have uncovered neural response functions that are much less stereotyped and regular than observed previously. Some of the most striking examples have been observed in the hippocampus and its associated brain areas. The classical view has been that hippocampal place cells are active only in a compact region of space and exhibit a stereotyped tuning to position. In contrast to this expectation, however, place cells in large environments typically fire in multiple locations, and the multiple firing fields of individual cells, as well as those of the whole population, vary in size and shape. We recently discovered that a remarkably simple mathematical model, in which firing fields are generated by thresholding a realization of a random Gaussian process, accounts for the statistical properties of place fields in precise quantitative detail. The model captures the statistics of field sizes and positions, and generates new quantitative predictions on the statistics of field shapes and topologies. These predictions are quantitatively verified in multiple recent data sets from bats and rodents, in one, two, and three dimensions, in both small and large environments. Together, these results imply that common mechanisms underlie the diverse statistics observed in the different experiments. I will discuss a mechanistic model, which suggests that the random Gaussian process statistics arise due to random connectivity within the CA3-CA1 hippocampal circuit. If time permits I will present another recent work, in which we uncover simple principles underlying the spatial selectivity in a new class of neurons in the medial entorhinal cortex, with tuning curves that are significantly less regular than those of classical grid cells. 

The equivalence between vastly different complex systems allows physcisits to make predictions without analyzing their microscopic details. Conversely, by reducing systems to their minimal constituents, we can describe phenomena that seem inscrutable. I will apply these principles to the synthetic world of neural networks, taking grokking, or delayed generalization, as a case study.

While often attributed to complex representation learning, I will show that grokking arises in simple, analytically tractable settings: linear teacher-student models and logistic regression. By explicitly solving the gradient flow dynamics, we identify the ratio of input dimension to training samples (λ=d/N) as the control parameter of a phase transition from overfitting to generalization. I will present two distinct mechanisms for grokking driven by this criticality. In regression, it appears as a divergence in "grokking time" near the interpolation peak. In classification, it emerges at the phase transition between linearly separable and inseparable data, caused by the divergence of the separating hyperplane. Ultimately, I will illustrate how viewing training dynamics through the lens of statistical physics and critical phenomena provides a simple solution for the mystery of delayed generalization.

Large Language Models (LLMs) have transformed what machines can do and how systems are designed to serve them. These models are both computationally and memory demanding, revealing the limits of traditional optimization methods that once sufficed for conventional systems. A central challenge in building LLM systems is improving system metrics while ensuring response quality.

This talk presents approaches for reducing latency in LLM systems to support interactive applications, from scheduling algorithm design to deployment. It introduces scheduling frameworks that use lightweight predictions of request behavior to make informed decisions about prioritization and memory management across two core settings: standalone LLM inference and API-augmented LLMs that interact with external tools. Across both settings, prediction-guided scheduling delivers substantial latency reductions while remaining practical for deployment.

From rough sketches that spark ideas to polished illustrations that explain complex concepts, visual communication is central to how humans think, create, and share knowledge. Yet despite major advances in generative AI, we are still far from models that can reason and communicate through visual forms.

I will present my work on bridging generative models and visual communication, focusing on three complementary domains: (1) algorithms for generating and understanding sketches, (2) systems that support exploratory visual creation beyond one-shot generation, and (3) methods for producing editable, parametric images for design applications.

These domains pose unique challenges: they are inherently data-scarce and rely on representations that go beyond pixel-based images commonly used in standard models. I will show how the rich priors of vision-language models can be leveraged to address these challenges through novel optimization objectives and regularization techniques that connect their learned features with the specialized representations required for visual communication.

Looking ahead, this research lays the foundation for general-purpose visual communication technologies: intelligent systems that collaborate with humans in visual domains, enhancing how we design, learn, and exchange knowledge.

Bio:

Yael Vinker is a Postdoctoral Associate at MIT CSAIL, working with Prof. Antonio Torralba. She received her Ph.D. in Computer Science from Tel Aviv University, advised by Profs. Daniel Cohen-Or and Ariel Shamir. Her research spans computer graphics, computer vision, and machine learning, with a focus on generative models for visual communication. Her work has been recognized with two Best Paper Awards (SIGGRAPH 2022, SIGGRAPH Asia 2023) and a Best Paper Honorable Mention (SIGGRAPH 2023). She was selected as an MIT EECS Rising Star (2024) and received the Blavatnik Prize for Outstanding Israeli Doctoral Students in Computer Science (2024) as well as the VATAT Ph.D. Fellowship.