Semantic segmentation of medical images is pivotal in disease diagnosis and treatment planning. While deep learning has excelled in automating this task, a major hurdle is the need for numerous annotated masks, which are resource-intensive to produce due to the required expertise…

Symmetries play a significant role in machine learning. In scientific applications, they often arise as constraints imposed by physical laws. More broadly, symmetries emerge whenever objects admit multiple ways to express them (for example, in graph machine learning). In addition…

In this talk, we will model this structure using probabilistic context-free grammars – tree-like generative models from linguistics. I will present a theory of denoising diffusion on this data, predicting a phase transition that governs the reconstruction of features at various…

Medical diagnosis can be naturally framed as a classification problem: inferring an underlying pathology from observed (e.g., imaging) data. A common failure mode in classification is shortcut learning, where models exploit spurious or confounding correlations. Shifts in patient…

Capturing path- and rate-dependent behaviors of solids, such as creeping, plastic deformation, damage, and fracture, often requires interpreting and quantifying relationships among the histories of variables, such as dislocation density and porosity. This relational information…

The Chatterjee Lab at the University of Pennsylvania develops generative algorithms to design functional molecules directly from sequence. Our work begins with a question: can short peptides be de novo designed to bind undruggable targets like disordered oncogenic fusions…

Professor Jain will describe how we map dark matter from large surveys of galaxies via gravitational lensing. Testing theories of cosmology with maps of galaxies and dark matter is a rich area for machine learning. He will describe the ‘old-fashioned’ physics-based approaches and…

This talk will present new insights into the learning dynamics of gradient flow in the feature-learning regime. For separable data, we show that gradient flow learns a simple ReLU network with two neurons corresponding to the two classes. For data drawn from an orthogonal…

Neural activity is complex, dynamic, nonlinear, and high-dimensional, and the datasets we collect from the brain are equally varied. This richness makes analysis difficult, and models trained on narrow conditions rarely generalize. Rather than seeking to simplify the problem, we…

In scientific machine learning, researchers construct simulators of physical systems which are either computationally expensive or intractable to model with conventional simulation tools. Despite impressive steps in the last couple years, these techniques generally fail to meet…