Hanzhang (Henry) Yin

Currently, I am a Master’s student in the Department of Computer Science at Brown University in Providence, Rhode Island, where I work on machine learning and computer vision related researches. I completed my undergraduate studies at the University of Rochester, earning dual degrees in Computer Science and Applied Mathematics.

Email  /  CV  /  Linkedin  /  X  /  Github

Research

My research interests span computer vision, generative and compression models, vision for 3D understanding, and applied AI in healthcare. Most of my research is about applying machine learning and deep computer vision to healthcare and visual computing. Projects span Parkinson's hospice risk prediction, image/video compression & generation, and early-stage 3D mesh motion prediction. Some works are highlighted.

Built a PD-specific prognostic model using MDS 3.0 linked with the Vital Status File. Benchmarked Logistic Regression, LightGBM, Random Forest, Naïve Bayes, and introduce an advance clustering method base on Logistic Regressin and LightGBM Model on Admission and Longterm Dataset. Moreover, we developed a Bi-LSTM deep learning model focusing on a newly provided time series data. We compared against Medicare hospice guidelines and analyzed feature contributions to highlight PD-unique signals.

Investigating codec-style pipelines where compressed latents serve as a structured prior for diffusion-based denoising and generation. Evaluating on ImageNet‑1k with FID/PSNR and studying robustness for denoising and generative editing in the loop.

This survey focus on the LoRA variants (1) Downstream adaptation, (2) math tasks performance, and (3) Reasoning skills given different difficulty levels assessing their abilities on Large Language Model. Specifically, we will r un s everal b enchmark tests on different variants of LoRA (including LoRA itself) using model LLaMA-3 (8 bit) on dataset GSM8k and GSM-Plus, dataset that primarily focusing on leverage model's mathematical inference ability, and analyze their behavior.

The paper recounts and revises the development of linear albegra system from ancient Chinese referenced on the acient Chinese text: The Nine Chapters on the Mathematical Art. The two main related Chapter that will be examined are: “Excess and Deficits” and “The Fangcheng Procedure” in both their mathematical concepts and some representitive questions. Furthermore, this paper will compare mordern approaches on linear algebra system to the ancient Chinese strategies, finding potential insight that reveals the process of how mathematics evolved.

Developed and enhanced front-end webpage designs using JavaScript and Fluent UI, integrating external NLP AI chatbot plugins to facilitate event scheduling within the Microsoft Teams app. Resolved security vulnerabilities in the user login verification module by redesigning regular expressions and hashing strategy, improving the efficiency of account verification between the website and database API.

Projects

Selected projects from coursework, competitions, and independent development. These span computer vision, NLP, and robotics.

Mini-CLEVR-VQA-Experiment is a lightweight VQA playground built on a synthetic Mini-CLEVR dataset. It ships with two strong, small-parameter baselines:

• ResNet-18 + SBERT
• CLIP ViT-B/32 + LoRA (r=8)

This is a QR Code scanner purely relied on traditional Computer Vision Algorithms. Here is the design pipline: (1) Implemented Sauvola binarization and connected-component analysis to detect finder patterns; (2) computed bounding boxes via right‑angle triangle geometry for pose estimation. (3) Real-time (~20 FPS) decoding with Kalman filtering for jitter-reduced tracking.

This 3D indie game was created by a seven-member team during a 48 hrs Game Jam hosted by GameChangers 益·游未尽. The objective is simple: guide all characters to the cashier's red point to escape the shop!

Small 3D indie game DEMO made by Unity.
Description: Enter a world of mysteries and puzzles as a reckless adventurer seeking hidden truths. Pay close attention to environmental details to progress. Level 1 is available now.

Teaching

• University of Rochester Teaching Assistant:
• CSC242: Introduction of Artificial Intelligence
• MATH 141: Calculus I
• MATH 162: Calculus II
• CSUG Tutor:
• Computer Science
• Mathematics

Other Fun Stuff

• I am a Trumpet Player for 8 years long! Play classical music the most, currently learning jazz.
• I am a ethuiast gamer who love playing FPS (*CS2* Especially), Adventure, ARPG, Sandbox, and a gazillion types of games. Here is my steam profile link. If you want to play with me, feel free to sent out a friend request!
• I am a "basketball" and "badminton" lover (although I am suck at this sport). Sometimes I will also go the gym for workout.
• I am a deadheart F1 fans, if you are also a fan of Red Bull ltd., we are friend :)