לוח שנה משולב

Abstract: How voluntary, self-paced intentions emerge in the brain and translate into action remains one of the most fundamental open questions in neuroscience. Leveraging rare access to intracranial neuronal recordings from human motor cortex, we built a real-time, online closed-loop system that allowed us to study the formation of voluntary actions under competitive conditions.We show that participants have only limited capacity to voluntarily steer their motor-cortex activity when doing so is strategically advantageous-revealing tight constraints on intentional control at the neural population level. Yet the commitment to act can be decoded reliably from motor-cortex activity roughly 250 ms before movement onset, at a time point when participants report already being consciously aware of their decision. We also find that brain–computer interfaces trained in one cognitive context transfer seamlessly to another, despite substantial differences in neural trajectories and force profiles-suggesting a shared underlying representational structure for volitional actions in motor cortex.Offline analyses further uncovered the specific neural patterns that signal commitment to action, shedding new light on how early voluntary actions can be reliably predicted from motor-cortex activity. We will conclude by discussing how these and related results inform emerging efforts to track and interpret intentions in advanced AI systems (ai-intentions.org).

Glucose can enter the blood following a meal, and/or can be produced by the liver and kidneys at times of need such as fasting. An elevation in blood glucose beyond steady state levels leads to secretion of the hormone insulin, leading to increase in glucose uptake into muscle and adipose tissues. Diabetes Mellitus arises when insufficient levels of insulin are secreted into the blood, manifesting as a chronic elevation in blood glucose levels. A reduction in glucose levels can lead to secretion of a large number of hormones, such as glucagon, cortisol and adrenaline, which cause endogenous glucose production and secretion into the blood,  maintaining homeostasis of glucose levels. In this talk we will use mathematical modeling and biochemical measurements to study the dynamics of hormone secretion in healthy individuals and Diabetes patients, and (hopefully) provide an answer to a key question: does the "body" measure glucose levels and regulates glucose levels accordingly by secreting insulin/glucose, as expected by a standard negative feedback system, or does it estimate future glucose levels and secretes hormones/glucose in a feedforward mechanism?Students interested in meeting the speaker after the seminar may sign up here:LINKFOR THE LATEST UPDATES AND CONTENT ON SOFT MATTER AND BIOLOGICAL PHYSICS AT THE WEIZMANN, VISIT OUR WEBSITE: https://www.bio

Ensuring that AI systems behave as intended is a central challenge in contemporary AI. This talk offers an exposition of provable mathematical guarantees for learning and security in AI systems.

Starting with a classic learning-theoretic perspective on generalization guarantees, we present two results quantifying the amount of training data that is provably necessary and sufficient for learning: (1) In online learning, we show that access to unlabeled data can reduce the number of prediction mistakes quadratically, but no more than quadratically [NeurIPS23, NeurIPS25 Best Paper Runner-Up]. (2) In statistical learning, we discuss how much labeled data is actually necessary for learning—resolving a long-standing gap left open by the celebrated VC theorem [COLT23].

Provable guarantees are especially valuable in settings that require security in the face of malicious adversaries. The main part of the talk adopts a cryptographic perspective,  showing how to: (1) Utilize interactive proof systems to delegate data collection and AI training tasks to an untrusted party [ITCS21, COLT23, NeurIPS25]. (2) Leverage random self-reducibility to provably remove backdoors from AI models, even when those backdoors are themselves provably undetectable [STOC25].

The talk concludes with an exploration of future directions concerning generalization in generative models, and AI alignment against malicious and deceptive AI.

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.

We will present a recent result with Joe Chen on the low temperature (2+1)D integer-valued Discrete Gaussian (ZGFF) model. The level lines were conjectured to have cube-root fluctuations near the sides of the box, mirroring the Solid-On-Solid picture. The new results resolve this and further recover the joint limit law of the top level-lines near the sides of the box, which is a product of Ferrari—Spohn diffusions.

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.