My name is Donald Loveland and I am a fourth year PhD candidate in the CSE department at University of Michigan, Ann Arbor, advised by Danai Koutra. My work is supported by an NSF Graduate Research Fellowship Award and a Rackham Merit Fellowship Award.

Research

My research focuses on improving the interpretability, robustness, and fairness of graph neural networks (GNNs) with the goal of mitigating harmful bias and identify practical limitations of current methods. Some of my work includes showing how disparity for different individuals can arise in GNNs, as well as different recommender system applications.

During the Summer of 2024, I was a research intern at Snap Inc. on the User Modeling and Personalization (UMAP) team, mentored by Clark Ju, Tong Zhao, and Neil Shah. Our work, currently under review, studies how to accelerate recommender system training from the perspective of matrix rank in the underlying embedding tables. During the two Summers before that, I interned as MIT Lincoln Laboratory as a machine learning researcher focusing on challenges in regards to combinatorial optimization over networks. This has led to a paper accepted for oral presentation at Complex Networks 2023, as well as another paper currently under submission.

Before my PhD, I was an applied machine learning researcher at Lawrence Livermore National Lab where I broadly worked on graph deep learning applications. Some of my specific research included graph classification/regression for molecules, inverse design of molecular graphs, and post-hoc insight extraction for graph neural networks (GNNs). I have also worked heavily in the computer vision space developing methods to extract actionable insights from convolution neural networks through generative counterfactual models.

Mentoring

As the first member of my family to go to college, I am grateful for the opportunities I have recieved during both my academic and professional career. While I am still figuring out how to properly navigate the world of academia, I have worked to make myself a resource for others who come from similar backgrounds. During my undergraduate degree, I acted as a peer mentor for younger first generation students and have continued these efforts by participating in undergraduate mentoring during my graduate studies. While at Lawrence Livermore, I also helped facilitate a two week long machine learning challenge geared towards teaching students how to perform applied research. Please feel free to reach out if you ever want to chat!