معرفی
David I. Inouye is an Assistant Professor at Purdue University's Elmore Family School of Electrical and Computer Engineering (ECE), where he conducts research on trustworthy machine learning methods that are robust to imperfect distributional and computational assumptions. His work bridges the gap between theoretical foundations and practical applications in AI systems.
Dr. Inouye's educational background includes:
- PostDoc in Machine Learning (2019) from Carnegie Mellon University under Prof. Pradeep Ravikumar
- PhD in Computer Science (2017) from The University of Texas at Austin under Profs. Inderjit Dhillon and Pradeep Ravikumar
- MS in Computer Science (2015) from The University of Texas at Austin
- BS in Electrical Engineering (2012) from Georgia Institute of Technology
- BA in Natural Sciences (2011) from Covenant College
Dr. Inouye's research focuses on developing trustworthy AI systems through multiple interconnected threads. His primary research vision centers on creating machine learning methods that maintain robust performance even when faced with distributional shifts or computational constraints. Trustworthy AI/ML forms the foundation of his work, with particular emphasis on Causal ML approaches to understand and mitigate robustness issues. He investigates how ML explanations relate to model robustness, exploring whether interpretability can enhance reliability. Another significant strand examines Robust Collaborative Learning on dynamic networks of edge devices, addressing the challenges of distributed AI systems. His work on Distribution Shift and Robustness tackles the critical problem of models failing when deployed in environments different from training conditions. Additionally, he explores Fairness in machine learning, viewing it as a specific form of robustness to sensitive attributes.
Dr. Inouye's recent publications (2022-2025) reveal a strong focus on causal approaches to robustness, with approximately 40% of his work connecting counterfactual reasoning to distribution shift problems. His research shows a clear trajectory toward practical implementations of theoretically grounded methods, with increasing emphasis on collaborative and federated learning settings. The interdisciplinary nature of his work is evident in publications spanning computer vision (25%), graph neural networks (15%), and fundamental machine learning theory (60%).
Dr. Inouye has received recognition for his research, including a Spotlight paper at ICLR 2024 (5% acceptance rate) for his work on "Benchmarking Algorithms for Federated Domain Generalization."
As an educator and mentor, Dr. Inouye has taught multiple iterations of ECE 47300 (Introduction to Artificial Intelligence) and ECE 57000 (Artificial Intelligence), as well as ECE 20875 (Python for Data Science). His research has been supported by significant funding from the National Science Foundation (NSF), the Army Research Laboratory (ARL), and the Office of Naval Research (ONR). These grants have enabled his team to pursue ambitious research questions at the intersection of theory and practice in trustworthy machine learning.
Dr. Inouye leads the Inouye Lab at Purdue University, which focuses on developing theoretically sound yet practical approaches to trustworthy machine learning. The lab maintains active collaborations with researchers at Carnegie Mellon University, The University of Texas at Austin, and industry partners working on edge computing and robust AI systems. Current projects include developing methods for robust collaborative learning across distributed edge devices and creating causal frameworks for understanding and mitigating distribution shifts in real-world applications.
