About
I am Qi Yang . I am now working as a NLP Algorithm Engineer in a International Digital Economic Academy(IDEA).
Before that, I acquired a Master degree under the supervision of Professor Ke Tang and graduated in 2022.
My existing research experiences includes Reinforcement Learning and Evolution Computation and Graph-based Casual Machine Learning.
Before that, I received my B.Eng degree from Huazhong University of Science and Technology, Academy of Artificial Intelligence and Automation.
During my undergraduate studies, I completed my thesis under the supervision of Prof. Min OuYang.
A PDF version in English could be save on the left side bar.
Personality: INTJ; Self-motivated; Nerd (Love reading including science fiction, history, psychology and a little bit philosophy);
Skills
- Code & Programming Proficient in Python 4 years, Know about C/C++, Go
- AI/ML Framework TensorFlow, Pytorch, Matplotlib, Numpy, Pandas, Gym, Transformers, HuggingFace etc.)
- Workflow Linux, Shell, Git, Docker, Conda, Singularity
- Database and Cloud SQL, Access, MySQL, Spark, Alicloud, AWS
- Others Solidity, HTML/CSS/ Javascript
- Soft skills Fast Reading and Learning, Grant and Patent Writing, Leadership , Collaboration
Background
International Digital Economy Academy(IDEA)
Research Intern/ NLP ALgorithm Engineer, Cognitive Computing and Natural Languge Centre(CCNL)
Jan. 2022 - now.(ShenZhen , GuangDong)
- Research Area: Large Pre-trained Langugage Model; Transfer Learning; Derivative-free methods;
- Designed and developed the semi-supervised cooperative knowledge-based QA system.
- Researched on parameter-efficient transfer learning algorithms for the large model.
- Contributed to the open-source to the open-source project Fengshenbang 1.0 and pre-training framework Fengshen
Southern University of Science and Technology
Master of Engineering, Department of Computer Science and Engineering
Sep. 2019 - Jun.2022(ShenZhen , GuangDong)
- Research Area: Reinforcement Learning; Evolutionary Algorithm; Non-convex Optimization;
- Lab: NICAL (Supervisor: Ke Tang)
- Publication : Published 2 published paper (SCI,EI), 1 working paper, 1 invention patent.
HuaZhong University of Science and Technology
Bachelor of Engineering, Academy of Artificial Intelligence and Automation
Sep. 2015 - Jun.2019 (Wuhan, Hubei)
- GPA : 3.47 , especially reached 3.8 in the last two years
- Language : IETLS 6.5 band (Reading 8.5 band), CET-4 and CET-6 certification;
- Award: Outstand Graduate (2019) and Scholarship for Self-improvement (2018)
- Dissertation: Research on Mechanism of Scaling Law in cities (Funded by National Nature Science Foundation
of China)
- Coursework : Math(Probability Theory, Calculus, Linear Algebra, Engineering Mathematics, Operation Research,Equation of Mathematical Physics), Cybernetics(Digital & Analog circuits, Signal Analysis, Control Theory, System Identification, Complexity Science, Systems Engineering, Information System), CS(Data structure & Algorithm, Computer Network, Principle of Micro-Computer, Distributed System, C
Programming, Internet Infrastructure), AI(Artificial Intelligence, Machine Learning, Robot Principle, Optimization) etc
- Social Activities: Elected to President for Science Fiction Association of HUST and Science Fiction Association Union of Wuhan in 2016
Award
- Awarded as Outstanding Graduate of Huazhong University of Science and Technology (2019)
- Received RenMin Scholarship for Self-motivated (2018)
Publications
- Qi Yang, Peng Yang, Ke Tang, Parallel Random Embedding with Negatively Correlated Search, In: proceeding of The Twelfth International Conference on Swarm Intelligence, 2021 (EI,Oral in ICSI2021).
- Peng Yang, Qi Yang, Ke Tang, Xin Yao, Parallel Exploration via Negatively Correlated Search, Frontier of Computer Science, 2020 (SCI,Poster in ECOLE2021).
- Qi Yang, Peng Yang, Ke Tang, Active Reinforcement Learning over MDPs, 2021(unpublished).
- Qi Yang, Peng Yang, Ke Tang, A method to train dynamic obstacle avoiding strategy based on random embedding evolution. Invention Patent. No.2021108449413.
Projects
Research on Generalizable and Sample-Efficient Reinforcement Learning by Selective Sampling
Adviser: Ke Tang Nov. 2020 - now
- Proposed and developed an active learning framework to selectively sampling representative and valuable training sets, which reduced training budget to 50% and achieved competitive performance.
- Implemented 6 relevant SOTAs on improving generalization and Advanced the procedural content generation environment of RL.
Research on Cooperative Co-evolution Algorithm in Reinforcement Learning
Adviser: Ke Tang Sep. 2019 - Nov.2019
- Developed a group of derivative-free RL optimization algorithms.
• to alleviate the performance deterioration in million-scale problem(exceeds SOTA more than 40%).
• to encourage parallel behavior exploration in multi-modal problem(score 2 to 3 times as many as SOTA).
- Designed and developed an open-source project for Evolutionary-based RL (including NCNES , NCS and NCSRE algorithms).
- Wrote 2 SCI/EI indexed papers, technical reports (demand from Huawei) and applied 1 patent as 1st author.
Research on Mechanism of Scaling Law in cities (Dissertation of Bachelor)
Adviser: Prof. Min OuYang Dec.2018 - Jun.2019
- Collect micro and macro statistic data of over 10 cities in 4 countries.
- Build a mathematic model and simulation for scaling law in complex cities system and estimate several significant indexes.
Life prediction and Health Diagnosis of Rolling Bearing with Support Vector Machine(Course Project)
Adviser: Prof. Ye Yuan May. 2018 - Jun. 2018
- Train a SVM-based model to predict the life of rolling bearing and recognize inferior bearing
- Achieve a 93% accuracy on Industry Real Rolling Bearing Dataset