Just graduated my first Ph.D. student! When I joined UM 4.5 years ago, I was fortunate to have Seok join my group from day 1. Seok's next destination is the University of Virginia (UVA) UVA_ESE where he joins as an Assistant Professor. Congrats, Dr. SeokhyunChung U-M Industrial and Operations Engineering

Super proud of my Ph.D. student Xubo Yue who just accepted an Assistant Professor job in the Mechanical & Industrial Engineering department Northeastern U.. Looking forward to celebrating your defense & graduation!

#BREAKING UC Davis student, Karim Abou Najm, 20 was identified as the person who was stabbed at Sycamore Park on his way back from an undergraduate awards ceremony. His father said, “We moved from Lebanon in 2018 hoping for safety” Heartbreaking😢😓💔 #RestInPeaceKarimAbouNajm

Heart breaking news from UC davis. Condolences to the family. Karim's father, Prof. Najdi was my a beloved professor at American University of Beirut

📢IISE Transactions Special Issue on Federated, Distributed Learning and Analytics - Manuscript submission deadline extended to July 15, 2023

In dictionary learning, it is often assumed that the data share the same (global) dictionary. This assumption is violated when data is heterogeneous. Can we provably identify what is shared and unique across datasets? We provide an answer in our new paper: arxiv.org/pdf/2305.15311…

Congrats, Dr. ( and Prof. To be ) Xubo Yue Xubo Yue When I joined UM close to 5 years ago, I was fortunate to have Xubo join my group from day 1 Now, Xubo's next destination is Northeastern University where he joins as an Assistant Professor this coming August U-M Industrial and Operations Engineering

Excited to share our new paper: "Collaborative and Distributed Bayesian Optimization via Consensus" We provide a generic framework for "Collaborative Trail & Error" that brings consensus optimization to optimal design Albert S. Berahas Xubo Yue arxiv.org/abs/2306.14348

‼️I am on the academic job market this year, seeking tenure-track positions. My research is on the foundations of learning, control, and analysis of emerging technologies with a specific focus on transportation. Check out my work here: wissamkontar.github.io

Two papers are featured in the December 2023 ISE Magazine, which are co-authored, respectively, by Petros Papadopoulos, David Coit and Aziz Ezzat and by Xubo Yue, Raed Al Kontar, and Ana María Estrada Gómez. (2/3)

Looking forward to my talk tomorrow at UT Austin: oden.utexas.edu/news-and-event… Oden Institute

Michigan Engineering Raed Al Kontar and Chinedum Okwudire were recently awarded a grant to work develop algorithms that address descriptive, predictive and prescriptive machine learning opportunities in large-scale distributed 3D printing. ioe.engin.umich.edu/2024/01/19/mic…

Our work, 'Personalized PCA: Decoupling Shared and Unique Features,' now accepted at JMLR, can extract interpretable and personalized features across different: (i) diseases, (ii) machines, (iii) human behavior, (iv) texts, (v) images, among others arxiv.org/abs/2207.08041

Researchers Raed Al Kontar and Naichen Shi developed a new statistical tool, personalized PCA, to distinguish shared and unique features across complex data from multiple sources. #IoT #DataScience #MachineLearning bit.ly/42YhUIq

Honored to get tenure at University of Michigan It has been an incredible journey with many ups and downs. Shoutout to the brilliant students and colleagues I have had the opportunity to work alongside, who have played a significant role in making this happen.

Congratulations to my PhD student, Naichen Shi (who is on the job market), for winning the INFORMS Data Mining Best General Paper Competition! A big shoutout as well to our collaborator, Salar Fattahi, for helping make this possible! Paper: arxiv.org/abs/2404.07955

LLMs can support optimization by leveraging contextual knowledge, but cannot be solely relied on as optimization agents. In our new paper we explore how combining LLMs with principled statistical models leads to more efficient Bayesian optimization. arxiv.org/abs/2505.14756

New Paper "Inv-Entropy: A Fully Probabilistic Framework for Uncertainty Quantification in Language Models" We flip the script on UQ in LLMs by learning p(x|y), asking: How many diverse inputs could have led to this output? Check it out 👉 arxiv.org/pdf/2506.09684