Research

Distribution Matching

Distribution matching is a tool for trustworthy ML and has the opposite objective of classification.

Localized Learning

Localized learning avoids end-to-end learning by updating model parts via non-global objectives.

Explainable AI

Explaining distribution shifts and prediction uncertainty are important for modern AI systems.

Preprints

Vertical Federated learning (VFL) is a class of FL where each client shares the same sample space but only holds a subset of the …
Preprint Project

Selected Publications

In high-stake domains such as healthcare and hiring, the role of machine learning (ML) in decision-making raises significant fairness …
Preprint Project

As Diffusion Models have shown promising performance, a lot of efforts have been made to improve the controllability of Diffusion …
Preprint Project

There has been a growing excitement that implicit graph generative models could be used to design or discover new molecules for …
PDF Code

Answering counterfactual queries has important applications such as explainability, robustness, and fairness but is challenging when …
Preprint PDF Project

While prior federated learning (FL) methods mainly consider client heterogeneity, we focus on the Federated Domain Generalization (DG) …
Preprint PDF Project Project

Distribution matching can be used to learn invariant representations with applications in fairness and robustness. Most prior works …
Preprint PDF Project

A distribution shift can have fundamental consequences such as signaling a change in the operating environment or significantly …
Preprint PDF Project Code

Spatial reasoning tasks in multi-agent environments such as event prediction, agent type identification, or missing data imputation are …
PDF Project Project Code

A central theme in federated learning (FL) is the fact that client data distributions are often not independent and identically …
PDF Project Project Code

Unsupervised distribution alignment estimates a transformation that maps two or more source distributions to a shared aligned …
PDF Project Code

While normalizing flows for continuous data have been extensively researched, flows for discrete data have only recently been explored. …
PDF Project Code

The unsupervised task of aligning two or more distributions in a shared latent space has many applications including fair …
PDF Project Project Code

Shapley values have become one of the most popular feature attribution explanation methods. However, most prior work has focused on …
PDF Project Code

While previous distribution shift detection approaches can identify if a shift has occurred, these approaches cannot localize which …
PDF Project Code

In practical applications of machine learning, it is necessary to look beyond standard metrics such as test accuracy in order to …
PDF Project Code

We consider objective evaluation measures of explanations of complex black-box machine learning models. We propose simple robust …
PDF Project Code

We propose a unified framework for deep density models by formally defining density destructors. A density destructor is an invertible …
PDF Project Code

The Poisson distribution has been widely studied and used for modeling univariate count-valued data. Multivariate generalizations of …
Preprint PDF Code

Square Root Graphical Models (SQR), a novel class of parametric graphical models that provides multivariate generalizations of …
PDF Poster Code

We propose a novel distribution that generalizes the Multinomial distribution to enable dependencies between dimensions. Our novel …
PDF Poster Code

We develop a fast algorithm for the Admixture of Poisson MRFs (APM) topic model and propose a novel metric to directly evaluate this …
PDF Poster Code

This paper introduces a new topic model based on an admixture of Poisson Markov Random Fields (APM), which can model dependencies …
PDF Poster Slides Code

Other Publications

Distribution shift can have fundamental consequences such as signaling a change in the operating environment or significantly reducing …
PDF Project

Project

Existing research on label noise often focuses on simple uniform or classconditional noise. However, in many real-world settings, label …
PDF Code

Owing to the sheer volume of text generated by a microblog site like Twitter, it is often difficult to fully understand what is being …
PDF DOI

Noise cancellation in an MRI environment is difficult due to the high noise levels that are in the spectral range of human speech. This …
PDF DOI

Due to the sheer volume of text generated by a micro log site like Twitter, it is often difficult to fully understand what is being …
PDF DOI

Teaching

An introduction to artificial intelligence (AI).
Last updated on

A graduate-level project-based introduction to artificial intelligence (AI) with a primary focus on unsupervised learning. The lecture …
Last updated on

An introduction to artificial intelligence (AI).
Last updated on

A graduate-level project-based introduction to artificial intelligence (AI) with a primary focus on unsupervised learning. The lecture …
Last updated on

A graduate-level project-based introduction to artificial intelligence (AI) with a primary focus on unsupervised learning. The lecture …
Last updated on

This course introduces Python programming to students through data science problems. Students learn Python concepts as well as …
Last updated on

A graduate-level project-based introduction to artificial intelligence (AI) with a primary focus on unsupervised learning. The lecture …
Last updated on

This course introduces Python programming to students through data science problems. Students learn Python concepts as well as …
Last updated on

A graduate-level project-based introduction to artificial intelligence (AI) with a primary focus on unsupervised learning. The lecture …
Last updated on

Contact