Anbang Ye | Machine Learning Engineer

Hi, I'm Anbang Ye.

A Self-driven, quick starter, passionate machine learning engineer and software programmer with a curious mind who enjoys solving complex and challenging real-world problems. I work on machine learning with a focus on natural language processing, computer vision and their applications.

About

I am a Computer Science Grad Student at Georgia Institute of Technology. Currently, I am working with Prof. Mark Riedl at the Entertainment Intelligence and Human-Centered AI Labs. My current research focuses on natural language processing. I have also worked on computer vision, hardware design, robotics, web and computer graphics during my bachelor's. I have 2 months of professional work experience as a research assistant. I am passionate about developing complex applications with machine learning that solve real-world problems impacting millions of users.

Languages: Python, C++, Java, C, JavaScript, HTML/CSS
Database: MySQL
Libraries: NumPy, Pandas, OpenCV, PySpark, Transformers
Frameworks: TensorFlow, PyTorch, Keras, Scikit-learn, D3
Tools & Technologies: Git, Docker, AWS, GCP

Looking for an opportunity to work in a challenging position combining my skills in Machine Learning and Software Engineering, which provides professional development, interesting experiences and personal growth.

Experience

Institute of Optics and Electronics

of Zhejiang University at Taizhou

Research Assistant

Supported research and development efforts to create commercial high-performance vision-based defect detection equipment and methods using C++ and OpenCV.
Engaged in designing equipment for detecting scratches on inner surfaces of fire sprinklers.
Designed a detection program with OpenCV and C++ to detect impurity in raw polyethylene. Detectable rate reached to 99%.
Assist in patent filing.

Tools:

Nov. 2020 – Jan. 2021 | Taizhou, China

Projects

                
Trash Sorting Sytem
                  A trash sorting sytem based on machine learning with automated annotating. 
                
AccomplishmentsAbout:Automatically labeled a semantic segmentation dataset of 3000 muli-object 
                  images and 20000 single object images of 204 categories based on only 200 manually labeled images. The Yolo-v7 model trained on the dataset and recieved 0.9059/0.9048 on MAP-50(bounding box/mask) on test data.
                  The pipeline can run and track objects in real time.
Team leader. We are hornored with National SRTP project title in 2019
Tools: Python, Pytorch, Tensorflow, RoboFlow, OpenCV

Neural Story Plot Planner
                  An ending guided neural story plot planner for logical story planning.
                
AccomplishmentsAbout:An ending guided neural story plot planner using commonsense knowledge extracted from a large language model by prompt engineering.
Tools: Python, Pytorch, GPT-J

Robust CARP
                  Robust Preference Learning for Storytelling via Contrastive Reinforcement Learning
                
AccomplishmentsTools: Python, Pytorch
About: We trained a contrastive model (CARP-CoOP) on story-critiques pairs, which can score how well a story matches a 
preference robustly and is subsequently used as the reward to fine-tune gpt-2-large via reinforcement learning (PPO).
This is a research project supported by EleutherAI
Tools: Python, Pytorch
\
                
Ocular Region Reconstruction
                  A system for reconstructing 3D deformable models, focusing on high-precision reconstruction of the ocular region.
                
AccomplishmentsAbout:A system capable of converting 3D face scans into 3D deformable models, focusing on high-precision reconstruction of the ocular region.
This is a part of a research project for Tencent.
Tools: Python, Pytorch，Yama
\
                
RoboMaster
                  2019 RoboMaster International Robotics Competition.
                
AccomplishmentsAbout:Worked on embeded system develpment, computer vision and robot control. Mainly responsible for coding the software for our drone.
Tools: C++, C, ChibiOS

Wizard Chess
                  A voice controled chess set.
                
AccomplishmentsAbout:A voice controled chess set. Move chess pieces with vocal command
Tools: C++, C
\
                
“Build a Cleaner Government” Cockpit
                  An interactive UI designed for Lishui City Survey Team of National Bureau of Statistics of China.
                
AccomplishmentsAn interactive UI designed for Lishui City Survey Team of National Bureau of Statistics of China. Watch the video demo for more details.
Tools: HTML, CSS, JavaScript, ECharts

Publications

● Anbang Ye, Christopher Cui, Taiwei Shi, Mark O. Riedl. Neural Story Planning. In Proc. of Creative AI Across Modalities AAAI 2023 Workshop (Hybrid). [Preprint]

● Louis Castricato*, Alexander Havrilla*, Shahbuland Matiana, Michael Pieler, Anbang Ye, Ian Yang, Spencer Frazier, and Mark Riedl. Robust Preference Learning for Storytelling via Contrastive Reinforcement Learning. Preprint, 2022. [Preprint]

● Anbang Ye, Bo Pang, Yucheng Jin, Jiahuan Cui. A Yolo-Based Neural Network with VAE for Intelligent Garbage Detection and Classification. In 2020 3rd International Conference on Algorithms, Computing and Artificial Intelligence, 439–445.[Paper]