Haocheng Dai

Ph.D. student

Scientific Computing and Imaging Institute

& School of Computing, University of Utah

haocheng@cs.utah.edu

Vitae / GitHub / YouTube / Flickr / Instagram

I am a Ph.D. student in computer science, advised by Dr. Sarang Joshi with both SCI Institute and Kahlert School of Computing, 🏜️ University of Utah. I also work closely with researchers from FSU, UCLA, UVa, and Yale. I was an applied scientist intern at 🍌 Amazon, focusing on 🎨 diffusion models for text inpainting (2023 summer) and πŸ“‹ multimodal transformers for document understanding (2022 summer).

Before I joined the U, I received my B.Eng. in computer science from Tongji University and have studied at ✑️ Israel Institute of Technology and πŸ₯ Institut de MathΓ©matiques de Toulouse as an exchange student, focusing on image analysis and Riemannian geometry, respectively.

My interest lies in developing specialized and trustworthy machine learning tools tailored for computer vision problems in πŸ§‘β€βš•οΈ healthcare settings, so as to improve medical treatment, diagnosis and understanding of the disease. My research extends to, but is not limited to:

πŸ‘¨β€βš–οΈ trustworthy machine learning (fairness and robustness);
πŸ“ geometric deep learning and shape modeling;
πŸ‘οΈ multimodal learning and vision language models;
πŸ”­ physics-informed machine learning.



Publications & Preprints


High-Fidelity CT on Rails-Based Characterization of Delivered Dose Variation in Conformal Head and Neck Treatments.
  • Haocheng Dai, Vikren Sarkar, Christian Dial, Markus Foote, Ying Hitchcock, Sarang Joshi, Bill Salter.
  • Applied Radiation Oncology (ARO), 2023
  • πŸ‘¨β€βš–οΈπŸ§‘β€βš•οΈ / Paper / Code / Slides / Citation

pmb



Detect AI-generated Images Uploaded for Risk Evidence Collection in CSSW.
  • Haocheng Dai, Siwei Chen, Bei Xiao, Yangho Chen.
  • Amazon Machine Learning Conference (AMLC), 2023
  • πŸ‘¨β€βš–οΈπŸ‘οΈ / Paper / Citation

ipmi



Neural Operator Learning for Ultrasound Tomography Inversion.
  • Haocheng Dai*, Michael Penwarden*, Mike Kirby, Sarang Joshi. (*equal contribution)
  • International Conference on Medical Imaging with Deep Learning (MIDL), 2023
  • πŸ§‘β€βš•οΈπŸ”­ / Paper / Code / Slides / Poster / Citation

midl



Modeling the Shape of the Brain Connectome via Deep Neural Networks.
  • Haocheng Dai, Martin Bauer, Tom Fletcher, Sarang Joshi.
  • International Conference on Information Processing in Medical Imaging (IPMI), 2023
  • Oral Presentation
  • πŸ§‘β€βš•οΈπŸ“πŸ”­ / Paper / Code / Slides / YouTube / Citation / Media Coverage

ipmi



Understanding Visual Documents from Customer Self-Service Workflow using Multimodal Transformer.
  • Haocheng Dai, Jia-Kai Chou, Siwei Chen, Bei Xiao, Yangho Chen.
  • Amazon Machine Learning Conference (AMLC), 2022
  • πŸ‘οΈ / Paper / Citation

ipmi



Integrated Construction of Multimodal Atlases with Structural Connectomes in the Space of Riemannian Metrics.
  • Kris Campbell, Haocheng Dai, Zhe Su, Martin Bauer, Tom Fletcher, Sarang Joshi.
  • Machine Learning for Biomedical Imaging (MELBA), 2022
  • πŸ§‘β€βš•οΈπŸ“ / Paper / Code / Citation

melba



Structural Connectome Atlas Construction in the Space of Riemannian Metrics.

ipmi



Services


I have served as a reviewer for several journals and conferences, including ACM MM, CVPR, ARO, MedIA, MELBA, MICCAI, MIDL, Scientific Reports, and the ICLR Workshop on AI for Differential Equations in Science.



Miscellaneous


I made a handful of notes for better understanding in machine learning, mathematics of imaging, metric estimation, image registration, and solving large systems of linear equations.

My ErdΕ‘s Number = 4:
Haocheng Dai -> Sarang Joshi -> Ulf Grenander -> Oved Shisha -> Paul ErdΕ‘s;
Haocheng Dai -> Mike Kirby -> Frank Stenger -> Ambikeshwar Sharma -> Paul ErdΕ‘s.

I'm an amateur photographer, vlogger and also a loyal reader of πŸ“° the New York Times, you can find the highlight front pages I collect by the years of 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, and 2023.



Footprints


ipmi