Research

The problem of fair division in computer science involves developing algorithms and protocols to allocate resources or divide tasks among multiple parties in a way that is deemed fair, considering various criteria such as individual preferences, constraints, and subjective notions of fairness. It is a challenging computational problem with applications in resource allocation, distributed systems, and multi-agent settings.

Published:

• Fairness and Efficiency in Online Class Matching (to appear at NeurIPS '24)

• Almost Envy-Free Allocations of Indivisible Goods or Chores with Entitlements (AAAI '24)

• Online Algorithms for the Santa Claus Problem (NeurIPS '22)

In review:

• Almost Tight Guarantees for Online Nash Social Welfare Maximization

• Fair and Efficient Allocations on Multigraphs

The problem of fair clustering in machine learning revolves around devising algorithms that partition data into clusters while ensuring equitable treatment of different demographic groups, aiming to mitigate biases and discrimination in the clustering process. It addresses the challenge of incorporating fairness considerations into the clustering framework to avoid reinforcing or exacerbating disparities present in the data.

Published:

• Fair Polylog Approximate Low-Cost Hierarchical Clustering (NeurIPS '23)

• Generalized Reductions: Making any Hierarchical Clustering Fair and Balanced with Low Cost (ICML '23)

Theoretical research in machine learning involves the development and analysis of mathematical models, algorithms, and frameworks that provide insights into the fundamental principles underlying learning systems. It aims to establish theoretical foundations, explore algorithmic properties, and understand the theoretical limits and capabilities of various machine learning methods.

Published:

• Dynamic Metric Embedding into lp Space (ICML '24)

• An Improved Relaxation of Oracle-Efficient Adversarial Contextual Bandits (NeurIPS '23)

• Analysis of a Learning Based Algorithm for Budget Pacing (AAMAS '23)

• Optimal Sparse Recovery with Decision Stumps (AAAI '23)

In review:

• Less is More: Adaptive Coverage for Synthetic Training Data

• A Game Theoretic Perspective on Missing Data Imputation

Research in computational neuroscience involves the application of mathematical and computational methods to analyze, model, and simulate complex neural systems, aiming to unravel the mechanisms underlying brain function and behavior. It encompasses the integration of neuroscience and computational techniques to advance our understanding of the brain's structure, dynamics, and information processing capabilities.

Published:

• A Mathematical Model-Derived Disposition Index without Insulin Validated in Youth with Obesity (JCEM '24)

• Estimating Insulin Sensitivity and Beta-Cell Function from the Oral Glucose Tolerance Test: Validation of a new Insulin Sensitivity and Secretion Model (AJP '23)

• A Machine Learning Approach for Predicting Impaired Consciousness in Absence Epilepsy (ACTN '22)

• The Pulse: Transient fMRI Signal Increases in Subcortical Arousal Systems During Transitions in Attention (NeuroImage '21)