Seminars

Jie Yang presents BRIDGE, a multilingual evaluation benchmark built from real clinical tasks and more than one million samples derived from electronic health records. The benchmark addresses the gap between simplified examination-style tests and the complexity of clinical data, while enabling comparison of rapidly changing language models. Evaluation of 95 models involved over 24,000 experiments and 39 million predictions. Results vary substantially by model family, task and language, with some open-source systems matching proprietary models. The study also finds that chain-of-thought prompting frequently reduces accuracy on these tasks and examines stigmatizing language produced during model reasoning at large scale.

Paul S. Weiss (UCLA) explores low-energy interactions and balanced cycling in natural systems as models for efficient synthetic devices and recycling. Because the relevant energies fall below visible-photon energies, tunnelling spectroscopy and imaging are used to probe these processes. Replicating biochemical cycling and low-energy transport could reduce the energy needed for devices and materials recovery. The work seeks missing physical principles, including spin conservation in chiral molecules and polarizability in biological and synthetic systems. This perspective has led to thermal-control advances in scale, speed and magnitude. Atomic-resolution imaging, spectroscopy and functional measurements help build precise structures while revealing heterogeneous nanoscale behaviour. The talk relates these advances to the physical, electronic, mechanical, thermal and chemical connections between materials and their surroundings that determine future nanomaterial measurements and devices.

This public discussion examines the factors that help preserve brain health during aging. Researchers working on neuroscience, brain stimulation, the microbiome and brain health consider how stress, everyday behavior and social and environmental conditions can affect resilience over a lifetime. The panel discusses emerging evidence about early vulnerability, possible interventions through lifestyle changes and stress management, and more targeted approaches. It also addresses the limits of current knowledge and the uncertainties that remain when translating these findings into ways to protect cognitive health.

Matthias Vigl examines whether the computational scaling strategies behind modern machine learning can benefit particle physics. Empirical scaling laws relate model performance to computing resources and help balance model size against training-data volume. For a fixed dataset, increasing model capacity beyond the interpolation threshold can improve generalization through double descent, but gains eventually encounter limits imposed by the data. The talk then considers scaling data and model size together, deriving compute-optimal relations for transformer-based jet taggers trained on as many as billions of simulated jets. It studies how training hyperparameters and input representations change these relations. The observed behaviour suggests that larger computing budgets could yield useful gains in particle physics, especially with a move toward more general-purpose foundation models.

Jan Kosinski explains how AlphaFold-style methods predict protein complexes and why their reliance on evolutionary information makes interactions between host and pathogen proteins especially difficult. The talk reviews large-scale studies, interpreting both successful predictions and failures rather than treating reported success rates as universal performance. It then examines ways to improve predictions through broader sampling, altered sequence alignments and experimental restraints, including crosslinking mass spectrometry, with influenza A virus as a worked research example. The session is suitable for students and researchers interested in host-pathogen interactions and structural bioinformatics; basic knowledge of protein structures and sequence alignments is useful, but prior AlphaFold experience is unnecessary.

Navigation and memory require coordination between map-like allocentric representations and body-centered egocentric actions. Aaron Wilber examines how the parietal–retrosplenial–anterior thalamic–hippocampal network performs these transformations, and how disrupted coordination contributes to spatial deficits in aging and Alzheimer’s disease. Rat sequence-task experiments reveal bidirectional hippocampal–parietal interactions across route-centered, place and egocentric representations. Because that task does not isolate reference-frame use, the laboratory developed a freely available task separating egocentric, allocentric and transformation conditions. Validation indicates that parietal cortex and anterior thalamus are required for transformation; ongoing work examines population states associated with each condition. The talk then connects impaired hippocampal–cortical communication during sleep to navigation deficits in 3xTg-AD mice, reporting recovery of circuit function and cognition after early, circuit-targeted 40 Hz entrainment. Related investigations examine this network during waking navigation. In TgF344-AD rats, egocentric impairments appear before transformation deficits, addressing the shortage of rodent studies of reference-frame coordination in Alzheimer’s disease.

How do humans stabilize an inverted pendulum, and why does a balanced pole eventually fall? Drawing on 25 years of fingertip pole-balancing research, this talk examines neural correction delays: longer poles move more slowly relative to the nervous system’s response time. Delay-differential models can stabilize the upright position, yet skilled people still experience falls. The proposed explanation is microchaos arising from interactions among delay, sensory dead zones and frequency-dependent force encoding. A region of transient falling solutions lies next to stable microchaotic dynamics. Such microchaos is absent in virtual frontal-plane balancing tasks, while models of standing postural sway lack the corresponding transient regime. The comparison suggests that human falls, unlike pole falls, are more plausibly associated with medical events or slips and trips.

Robot missions require both collision-free trajectories and logical task ordering, such as collecting a resource before entering a restricted region. Augmented graphs of convex sets combine continuous trajectory optimization and discrete task sequencing in one problem. A layered graph constructed from an exact convex partition of free space selects an optimal task order and trajectory through a shortest-path calculation, exact up to finite Bézier parameterization. Its structure corresponds to Bellman–Held–Karp dynamic programming for the travelling-salesperson problem, retaining singly exponential worst-case complexity and improving exponentially on general temporal-logic tools. A library of specifications covers ordered collection, alternative keys, combined prerequisites, timing and conditional logic with correctness guarantees. Numerical benchmarks demonstrate exponential speedups and near-global optimality. Possible extensions include moving environments, safety-aware planning with conformal prediction sets and a large galactic-survey benchmark.

To orchestrate complex behaviors in mammals, the cerebral cortex must continuously transmit its processing results to subcortical regions. While internal cortical processing relies on highly selective sparse codes, these descending cortical output streams employ an enigmatic dense code, characterized by high firing rates and low feature selectivity. This dense population code presents a profound paradox: how can downstream regions extract precise, cognitively relevant signals when most cortical output neurons can be active at any moment? We address this paradox by testing the hypothesis that cortex does not lose information when switching to a dense code. Instead, it utilizes a mechanism for sparse-to-dense coding transformations discovered by my laboratory: thalamocortical synapses target specifically the dendritic initiation zone for calcium action potentials (APs), which enables cortical output neurons to transmit multiple information streams simultaneously via a multiplexed 1-2-3 AP syntax. I will provide first evidence that this remarkable synaptic specificity and coding syntax generalize across long-range pathways. Information coupling via an anatomically and biophysically distinct dendritic nexus may hence be a ubiquitous mechanism for sparse-to-dense coding transformations in cortex. Dissecting these mechanistic origins of cortical output streams is only now possible due to a unique in vivo – in silico approach, perfected over two decades, that my laboratory developed for bridging the gaps between dendritic and population-level computations

Yixuan He (Arizona State University) considers graph neural networks for social, financial and biological networks in which edge signs and directions contain information needed for clustering and prediction. After an introduction to graph neural networks, the talk explains how signed and directed methods preserve relationships lost when every connection is treated as unsigned and undirected. The seminar then asks when a representation learned before the final task can transfer to a different graph or to a setting with few labels. Two theoretical perspectives address transfer across graph domains and the use of link prediction for linear community detection. TopoDIG and TopoSIGN combine encoders that preserve edge structure with persistent topological features, enabling pre-training and prompt-based adaptation on directed and signed networks.

Richard Huang investigates how DNA nanostructures deform under uneven hydrodynamic forces, a question relevant to drug delivery, vaccines and biosensors. Coarse-grained molecular dynamics and mechanical models first examine linear and circular DNA in extensional flow. Drag concentrates tension near the middle of linear duplexes, disrupting base pairs and causing overstretching. Deforming minicircles straighten along their sides while concentrating bending into opposing tips and localized kinks. Models based on DNA elasticity and hydrodynamic-force distributions explain these responses, then extend to wireframe DNA origami to study force transmission and how local structural transitions produce whole-particle deformation. The results inform structures designed for stability or controlled deformation in flow. Online via Zoom. Friday 2 October 2026, 12:00–13:00 EDT (America/New_York; UTC−4). Open the Zoom link on the official event page; the audience explicitly includes the public. Richard Huang is affiliated with MIT. Organized by MIT Department of Mathematics — Computational Research in Boston and Beyond Seminar.

Antonio Ocello tests the idea that heavy-tailed noise improves diffusion and flow-based generative models by matching heavy-tailed data and encouraging diverse samples. Replacing Gaussian noise also changes the estimation task. A combined theoretical and experimental study derives sampling-error bounds for representative heavy- and light-tailed diffusion models, finding that heavy-tailed noise makes statistical estimation harder and produces less favourable bounds. Experiments with synthetic and real data recover the predicted trade-off. The results question whether heavy-tailed initialization reliably improves exploration of rare regions. This is joint work with Hamza Cherkaoui and Hélène Halconruy. Online via Microsoft Teams. Friday 2 October 2026 at 15:30 CEST / 13:30 UTC / 14:30 BST (Europe/Paris). Follow the Microsoft Teams link beneath this occurrence on the seminar page; the public route offers browser or app access. Ocello is affiliated with ENSAE Paris. Organized by GAMEX — Generative AI Modeling for Extreme Events, with GLE²N and CIRCE support; the series welcomes scientific exchange beyond the network.

Srinivasan Keshav (University of Cambridge) presents an interface designed to help ecologists use AI for remote sensing despite unfamiliarity with embeddings and algorithms. The Tessera Embeddings Explorer lets practitioners privately explore a region of interest, create habitat maps and perform regressions using their domain expertise and ground-truth knowledge. The talk demonstrates the tool and explains the design principles and the software architecture that follows from them, connecting usable ecological workflows with AI representations. Keshav is the Robert Sansom Professor of Computer Science at Cambridge.

Picower Institute Colloquium on the Brain and Cognition featuring Christopher Harvey, PhD, of Harvard University, held in Singleton Auditorium (46-3002) at MIT Building 46, 43 Vassar Street.

Alok Puranik (Jane Street) examines data mixing for large language models: allocating model capacity and computation among training sources. A small model’s preferred data mix may differ from that of a larger model, yet the cost of large-model training forces many decisions to depend on smaller experiments. Scaling laws can predict performance as size and computation increase only when trends remain stable or change predictably. The talk uses experiments from Puranik’s work and frontier laboratories to examine data-mix questions that violate these assumptions, explain why small-scale results can mislead, and explore better methods for choosing training data. His research at Jane Street concerns scaling laws for sequence models in trading.

Scott J. Russo examines how chronic psychosocial stress alters communication between the immune system and the brain in depression and anxiety. Human observations and mouse experiments connect persistent increases in circulating monocytes and inflammatory signals with stress-related illness. The talk describes stress-responsive neurons that project to bone marrow and regulate monocyte production, release and trafficking. It also considers how stress disrupts the blood–brain barrier, allowing peripheral inflammatory proteins to affect reward regions such as the nucleus accumbens. Together, these findings examine how peripheral immune cells act at the neurovascular interface to change brain circuits and behaviour, providing potential mechanisms relevant to treatment-resistant mood disorders. Online via Rutgers Brain Health Institute Zoom. Thursday 1 October 2026, 12:00–13:00 EDT (America/New_York; UTC−4). Open the organizer’s event page and follow its Zoom link; the meeting ID and passcode are supplied there. Hosted by Yong Kim as part of the BHI Plenary Seminar Series. This listing advertises the verified remote option.

Cryopreservation constrains manufacturing and clinical delivery of immune cells such as NK and T cells. DMSO protects against freezing but becomes toxic at physiological temperatures; associated infusion morbidity can require washing or divided dosing that diminishes therapeutic effectiveness. This talk presents an alternative in which reversible, cytocompatible hydrogels encapsulate cells before freezing without DMSO. The speaker examines evidence that encapsulation cushions osmotic stress, suppresses kinetic and phase transitions, and reduces injury caused by extracellular ice growth and fusion. The discussion covers recovery efficiency, function after thawing, control of biological stress responses, and the challenges of extending the approach to clinical-scale cell preservation.

Ivo D. Dinov examines augmented intelligence as a partnership between people and computational systems. The talk traces the technical and philosophical development of neural networks, then introduces complex-time, or kime, representations and the inference methods they support. Applications include medical imaging, healthcare and economic forecasting. It also considers the academic mission, societal consequences of AI and tensions between rapid technical development and human values, framing mathematical AI methods as extensions of human reasoning. Free hybrid seminar hosted by Wayne State University's Institute for AI and Data Science, AIDaS: CAD Seminar Series. Wednesday 30 September 2026, 14:30–15:30 EDT (America/Detroit; UTC−4). Attend in Room 1146, Faculty/Administration Building, 656 W. Kirby, Detroit, MI 48202, USA, or use the Zoom link on the official event page. Researchers, practitioners, students and faculty are welcome; the organizer encourages in-person attendance.

Navid Azizan presents methods that make neural models obey physical, safety and operational constraints at deployment. Hard-constrained neural networks, or HardNets, enforce input-dependent constraints by construction while preserving universal approximation within the feasible function class. Applications include models of chaotic dynamics with bounded trajectories, energy-constrained operator learning, safe reinforcement learning and control with formal guarantees. The talk then turns to enforcing constraints during sampling from pretrained diffusion and flow-matching models. Formulating generation as trajectory optimization allows receding-horizon control to guide outputs toward feasibility without retraining or imposing excessive restrictions on the sampling process. Examples from fluid dynamics, robot planning and control, PDE control and language-guided image editing illustrate how constraints can define admissible behavior while retaining expressive learning and generation.

Lenka Zdeborová introduces the Rules-and-Facts model, in which some observations follow a shared rule while others are isolated exceptions requiring memorization. The framework studies when a learner can acquire the rule and retain those exceptions simultaneously. Its results emphasize how capacity is organized and deployed, rather than capacity alone. The talk examines how regularization and the geometry of kernels or learned feature maps can reserve resources for memorization without undermining rule learning. It connects the balance between abstraction and memory to architectural and algorithmic choices, providing a theoretical account of neural networks that both generalize and recall specific facts.