WELCOME TO YUE ZHOU'S HOME PAGE
​
ABOUT
I am one among many human beings devoting to humanity and technology development. The meaning of my academic career is all about contributing myself in enhancing productivity, comprehending the objective pehnonmenon, and retrofitting the current systems. Seeking for a break-out is my attitude toward my life.
—With gratitude to my Alma Maters
ABSTRACT INTERPRETATION
April 12, 2021
Abstract Interpretation is the statical analysis techniques that apply to the code at compilation time. The logic inside is to build an abstract interpreter to rigorously connect the abstract domain with concrete domains. Based on different concrete domains, the content of each abstract interpreter is unique. I built an abstract interpreter for the integer domain(one type of the concrete domains).
ROBOTIC VISION
Sep 2020 - Sep 2021
In this multi-robot perception research, based on CNN and GNN, we have improved the performance of segmentation and depth detection on objects(both real world and simulation) in the case where multiple cameras are applied and some of them share the corrupted images. My contributions are 1. to implement the C++ scripts in Unity, Rviz, and Flightmare according to the purpose of data collection. 2. data collection. 3. image processing on RGB/segmentation transformations. 4. file managements by python scripts 5. analysis on the result.
CITIBANK BACKEND SYSTEM
May 2021 - Sep 2021
During this project, we built a fully automated back-end system for the competition website of variety of trading bots at Citibank. As the group leader, organized the collaboration and coordinated the teamwork. The back-end system has 1. a cron job scheduler for all competitions 2. an online file management system that has scripts on downloading submitted files, all types of error checks and validation, competition and result publishing. 3. communication and interactions with back-end database, website database and all participants. In this project, I was responsible for the cron job, validation process, and the read&write to front-end database(SQL).
ODE & SDE & PDE
Sep 2017 - Sep 2019
By the time when I completed my first master degree in mathematics at Courant, the thesis was presented. It coherently adopts and connects the Q-measure(math measure) and P-measure(statistical measure). I built one of the possible connections between these two measures. In P-measure, the stochastic differential Equation and Monte-Carlo simulation has been applied. In Q-measure, it starts from probability theory.
https://arxiv.org/abs/2001.05124
PROJECTS
Along the Path of Inventing, all the projects below are accomplished within the year of 2021.
BACKEND SYSTEM FOR CITIBANK'S COMPETITION PLATFORM
Citibank
A fully automated back-end  for the competition platform developed during the summer of 2021. It is under credential agreement, hence I am not allowed to share the content.
During the class taught by Professor Yann Lecun, a supervised learning model has been trained on two thousand labeled datapoints, and ten thousand unlabeled datapoints, in total of12k images. Techniques such as image augmentation and Neural Network ResNet-101 has been applied to the training process.
This project is the final project from the class CSCI-2433 Database Systems at NYU. It is an end-to-end solution from the end of customer's need and to the other end of technological solutions. In the middle of the process, the involvements are data collection(even web crawler), data management(cleaning, allocations, storage, utilizations in data lake), data training in deep learning models, and eventually a website that connects to my models and representing the result and commercial solutions.
In this project, I analyzed the pre-condition and post-condition of programming bodies under the eight categories: Empty Statement, Assignment, Break, Program, Statement List, If, If else, While. Meanwhile, multiple abstract domains have been implemented, such as Bottom&Top, Sign, Parity. Each abstract domain has its own purpose on guiding the concrete domain, which corresponds to the program body as we daily write. Building an abstract interpreter is similar to interpret the attributes of an abstract syntax tree. The purpose of adding the attributes are for predicting the range of possible values under(before/during/after) each statement(while, if, ifelse...).
This project includes a tokenizer, parser, static semantics, operational semantics on small steps implementation of the language MiniOO. The project is proposed in the ph.D.-level CS course CSCI-3110-GA at NYU.
An NP problem is proposed in each week during the semester. In order to complete the restriction of limiting the computing time(runtime) under two minutes for each problem solution, I constructed an optimized heuristic solution for each problem.
This project is also located in my github repository.
​
The problem is designed by Professor Dennis Shasha, and realized by me. It is a multiple player game developed in python3. The goal of the game is obtain as many rubies as possible in each player's turn. Do keep in mind, there is a risk.
Summary of All Relevant Courses from the Transcript
​
​