Tyler LaBonte

About Me

I am a Senior Researcher at Microsoft Research AI Frontiers working on future capabilities of multimodal agents.

My research goal is to ensure robust, grounded, and trustworthy multimodal agents through foundational innovations in representation and generalization. Recently, I was a core member of the team behind Phi-4-reasoning-vision, a 15B multimodal vision-language reasoning model with application to computer-use agents. My PhD work focused on theory and algorithms for robustness to distribution shift.

Recent News

• June 2026: Joined Microsoft Research AI Frontiers as a Senior Researcher.
• May 2026: Successfully defended my dissertation and graduated from Georgia Tech!
• April 2026: Recognized as a top-200 reviewer by ICLR 2026.
• April 2026: Our work On the Unreasonable Effectiveness of Last-layer Retraining was accepted to TMLR.
• March 2026: Excited to release Phi-4-reasoning-vision-15B, our new multimodal vision-language reasoning model and the latest in the Phi model series from Microsoft Research.

Selected Publications

Technical Reports

1. Phi-4-reasoning-vision-15B Technical Report.
   (αβ) Jyoti Aneja, Michael Harrison, Neel Joshi, Tyler LaBonte, John Langford, and Eduardo Salinas.
   Technical Report, 2026. [blog] [huggingface] [code] [press] [press]

Conference Articles

1. Task Shift: From Classification to Regression in Overparameterized Linear Models.
   Tyler LaBonte*, Kuo-Wei Lai*, and Vidya Muthukumar.
   AISTATS 2025. [arxiv] [code] [poster]
   INFORMS Applied Probability Society Conference 2025.
   IMS Workshop on Frontiers in Statistical Machine Learning 2025 (top-10 award).
2. The Group Robustness is in the Details: Revisiting Finetuning under Spurious Correlations.
   Tyler LaBonte, John C. Hill, Xinchen Zhang, Vidya Muthukumar, and Abhishek Kumar.
   NeurIPS 2024. [arxiv] [code] [poster] [video]
3. Towards Last-layer Retraining for Group Robustness with Fewer Annotations.
   Tyler LaBonte, Vidya Muthukumar, and Abhishek Kumar.
   NeurIPS 2023. [arxiv] [code] [poster] [video]
4. Scaling Novel Object Detection with Weakly Supervised Detection Transformers.
   Tyler LaBonte, Yale Song, Xin Wang, Vibhav Vineet, and Neel Joshi.
   WACV 2023. [arxiv] [code] [poster]

Journal Articles

1. On the Unreasonable Effectiveness of Last-layer Retraining.
   John C. Hill, Tyler LaBonte, Xinchen Zhang, and Vidya Muthukumar.
   Transactions on Machine Learning Research, 2026. [arxiv] [code]
2. Quantifying the Unknown Impact of Segmentation Uncertainty on Image-Based Simulations.
   Michael C. Krygier, Tyler LaBonte, Carianne Martinez, Chance Norris, Krish Sharma, Lincoln N. Collins, Partha P. Mukherjee, and Scott A. Roberts.
   Nature Communications, 12(1):5414, 2021. [code]

Selected Awards

1. ICLR 2026 Top-200 Reviewer ($675)
2. IMS Workshop on Frontiers in Statistical Machine Learning Travel Grant ($500)
3. Simons Institute Deep Learning Theory Workshop Travel Grant ($2,000)
4. DoD NDSEG Fellowship ($170,000)
5. NSF Graduate Research Fellowship ($138,000—declined)
6. USC Discovery Scholar (research distinction for <100 USC graduates)
7. USC Trustee Scholar (Full scholarship worth $225,000)
8. USC Viterbi Fellow (Research funding worth $24,000)

Advising

1. Xinchen Zhang—Georgia Tech MS → Synovus (2024—2025)
2. John C. Hill—Georgia Tech BS/MS → Georgia Tech PhD (2022—2024)

Academic Service

1. Reviewer: NeurIPS (2023, 2024, 2026), CVPR (2026), ICLR (2024, 2026), TMLR (2025, 2026), ICML (2025)
2. Reviewer, CVPR Workshop on Demographic Diversity in Computer Vision (2025)
3. Reviewer, ICLR Workshop on Spurious Correlations & Shortcut Learning (2025)
4. Organizer, Georgia Tech ML Theory Reading Group (2021—2023, 2025)
5. System Administrator, Georgia Tech ML Theory GPU Cluster (2022—2025)
6. Student Organizer, Learning Theory Alliance Workshop (2023)

Other Activities

1. Fleet Captain, Georgia Tech Sailing Club (2023—2025)
2. House Chair, USC Hawai'i Club (2020—2021)
3. Vice President of Finance, USC Hawai'i Club (2019—2020)



Theme by orderedlist