seminars
-
Learning analytics that facilitate meaningful learning experiences
Abstract: The focus of education is shifting from outcomes to processes. How do learners make sense of new challenges, and how can such sense-making be identified and supported? In this talk, I will demonstrate how the use of learning analytics and digital trace analysis, coupled with innovative design and strong theoretical foundations, allows us to… Continue Reading Learning analytics that facilitate meaningful learning experiences
-
Computational and Statistical Limits in Modern Machine Learning – Job Talk
Abstract: Modern machine learning systems operate in regimes that challenge classical learning-theoretic assumptions. Models are highly overparameterized, trained with simple optimization algorithms, and rely critically on how data is collected and curated. Understanding the limits of learning in these settings requires revisiting both the computational and statistical foundations of learning theory. A central question in learning… Continue Reading Computational and Statistical Limits in Modern Machine Learning – Job Talk
-
Modern Challenges in Learning Theory – Job Talk
Abstract: Machine learning relies on its ability to generalize from limited data, yet a principled theoretical understanding of generalization remains incomplete. While binary classification is well understood in the classical PAC framework, even its natural extension to multiclass learning is substantially more challenging. In this talk, I will present recent progress in multiclass learning that… Continue Reading Modern Challenges in Learning Theory – Job Talk
-
From Learning Theory to Cryptography: Provable Guarantees for AI – Job Talk
Abstract: 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… Continue Reading From Learning Theory to Cryptography: Provable Guarantees for AI – Job Talk
-
Selective Perception under Active Resource management for Reinforcement Learning
Abstract: Reinforcement learning in multi modal environments often relies on sensors that vary in informativeness, latency, and computational cost, making continuous full observation impractical under real resource constraints. Standard approaches typically embed sensing cost into the reward, treating all measurements as uniformly undesirable and failing to account for fixed sensing budgets that require informed allocation.… Continue Reading Selective Perception under Active Resource management for Reinforcement Learning
-
Contrastive Learning for Multi-Modal Medical Imaging
Abstract: Positron Emission Tomography combined with Computed Tomography (PET/CT) plays a critical role in oncology by integrating functional and anatomical imaging. However, the scarcity of annotated medical imaging data, combined with the complexity of the dual-modality information, presents a significant challenge for supervised learning approaches. This work explores a self-supervised learning framework for PET/CT imaging… Continue Reading Contrastive Learning for Multi-Modal Medical Imaging
-
A machine learning approach to the morphology of human brain ventricles
Abstract: Brain ventricular morphology contains crucial pathological information for neurological conditions including Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis. However, manual segmentation from MRI scans is time-consuming (3-4 hours) and limits large-scale analysis. This work presents a machine learning pipeline for automated ventricular analysis. We developed an Attention U-Net-based model achieving state-of-the-art segmentation in 20… Continue Reading A machine learning approach to the morphology of human brain ventricles
-
Reinforcement Learning Algorithm for Learning Shadow Variables in Combinatorial Optimization Problems
Abstract: Many real-world decision-making problems are modeled as Mixed-Integer Linear Optimization (MILO) problems, and are solved by exact tree-search-based optimization algorithms. Some knowledge, such as preferences for the solution, often exists but is not explicitly modeled. We examine incorporating such knowledge as additional constraints to potentially shorten optimization. A wise choice of “equality preferences” (i.e.,… Continue Reading Reinforcement Learning Algorithm for Learning Shadow Variables in Combinatorial Optimization Problems
-
Graph Representation Learning for Data Betterment
Abstract: Data integration is a fundamental challenge in data-driven systems, aiming to unify heterogeneous sources into a single, coherent view. This field serves as an umbrella for a wide range of tasks, all of which involve reasoning about how disparate data items relate and how they can be meaningfully combined. In recent years, much attention… Continue Reading Graph Representation Learning for Data Betterment
