Chunheng Jiang is a Computer Science PhD of RPI and a Rensselaer-IBM AIRC (Artificial Intelligence Research Collaboration) Research Extern. He will join Google as a Machine Learning Engineer in August of 2022. He has a solid background in Network/Graph Science and Deep Learning. Also, he is interested in information retrieval, recommendation, optimization and data visualization. He has involved in diverse projects ranging from network inference, machine learning, network control to airline/highway traffic congestion to nuclear reaction networks and boiling crisis in nuclear reactors. Their collaboration works have been published in high-quality conferences (AAAI, IJCAI) and journals (Scientific Reports, New Journal of Physics).
CONTACT
Email: jiangchunheng@gmail.com
Phone: +1 518-960-7682
GitHub: https://github.com/horsehour
LinkedIn: https://www.linkedin.com/in/jiangch/
EDUCATION
May 2016 - May 2022: PhD, Computer Science, Rensselaer Polytechnic Institute
May 2016 - May 2018: MS, Computer Science, Rensselaer Polytechnic Institute
Sep 2011 - Jul 2014: MS, Applied Mathematics, Southwest Jiaotong University
PUBLICATIONS
(1) Chunheng Jiang, Tejaswini Pedapati, Pin-Yu Chen, Yizhou Sun, Jianxi Gao. Neural Capacitance: A New Perspective of Neural Network Selection via Edge Dynamics, Preprint at https://arxiv.org/abs/2201.04194, 2022.
(2) Chunheng Jiang, Boleslaw Szymanski, Jie Lian, Shlomo Havlin, and Jianxi Gao. Nuclear Reaction Network Unveils Novel Reaction Patterns Based on Stellar Energies, New Journal of Physics, 2021.
(3) Xiang Niu, Chunheng Jiang, Jianxi Gao, Gyorgy Korniss, and Boleslaw Szymanski. From Data to Complex Network Control of Airline Flight Delays, Scientific Reports, 11, 18715, 2021.
(4) Xiang Niu, Christopher Brissette, Chunheng Jiang, Jianxi Gao, Gyorgy Korniss and Boleslaw K. Szymanski. Heuristic Assessment of the Economic Effects of Pandemic Control, Scientific Reports, 11, 7645, 2021.
(5) Chunheng Jiang, Jae-wook Ahn and Nirmit Desai. Acoustic Environment Transfer for Distributed Systems, In SEC-2020: The 5th IEEE/ACM Symposium on Edge Computing, IEEE, 2020.
(6) Chunheng Jiang, Jianxi Gao, and Malik Magdon-Ismail. Inferring Degrees from Incomplete Networks and Nonlinear Dynamics, in Proceedings of the 29th International Joint Conference on Artificial Intelligence (IJCAI 2020), 2020. Oral. Acceptance Rate: 12.6%. (slides)
(7) Chunheng Jiang, Jianxi Gao, and Malik Magdon-Ismail. True Nonlinear Dynamics from Incomplete Networks, in Proceedings of 34th AAAI Conference on Artificial Intelligence (AAAI 2020), 2020. Oral. Acceptance Rate: 20.6%.
(8) Jun Wang, Sujoy Sikdar, Tyler Shepherd, Zhibing Zhao, Chunheng Jiang and Lirong Xia. Practical Algorithms for STV and Ranked Pairs with Parallel Universes Tiebreaking, in Proceedings of 33rd AAAI Conference on Artificial Intelligence (AAAI 2019), 2019.
(9) Chunheng Jiang, Sujoy Sikdar, Jun Wang, Lirong Xia, and Zhibing Zhao. Practical Algorithms for Computing STV and Other Multi-Round Voting Rules, in EXPLORE-2017: The 4th Workshop on Exploring Beyond the Worst Case in Computational Social Choice, 2017.
(10) Chunheng Jiang and Wenbin Lin. DEARank: A Data-envelopment-analysis-based Ranking Method. Machine Learning, 2015, 101: 415 – 435.
(11) Zhongyou Pei, Chunheng Jiang, and Wenbin Lin. Random Walks on the Bipartite Graph for Personalized Recommendation, in 2013 International Conference on Computer Science and Artificial Intelligence (ICCSAI 2013), Yuetong Lin and Gabriel Alungbe, Eds., Chengdu, China, 2013, 97 – 102.
RESEARCH INTERESTS
Network Science and Artificial Intelligence
- Network Inference (Individual Degrees, Topology and Dynamics)
- Deep Learning / Reinforcement Learning / Neural Architecture Search
WORK EXPERIENCE
- Jun - Aug 2020: Summer Research Intern, IBM, Yorktown Heights, NY, USA
- Jul 2014 - Mar 2016: Software Engineer, Antusuoji Network Technology Co., Ltd., Chengdu, China
PROJECTS
(1) Aug 2020 - Present : Dynamical System View of Neural Network Training, Rensselaer-IBM AI Research Collaboration (AIRC), Research Extern | Python, TensorFlow
- Built a novel graph representation for various neural architectures (e.g., ResNet, DenseNet, MobileNet, VGG, etc.)
- Studied the topological properties (e.g., resilience, shortest path length) of neural architectures for robust models
- Derived approximated training dynamics to speed-up neural network training and neural architecture search (NAS)
- Achieved 10 $\sim$ 70% relative improvement over the best baseline w.r.t ranking of neural networks
- Techniques: SGD, transfer learning, neural architecture search/design, learning curve prediction
(2) Jun - Aug 2020 : Acoustic Environment Transfer, IBM/Summer Research Intern | Python, TensorFlow
- Extended neural style transfer techniques to the synthesis of audio with desired environmental sounds (UrbanSound8K)
- Developed a new metric to evaluate acoustic environment transfer models (e.g., Ulyanov, Mital, VGGish and SoundNet)
- Produced augmented data with diverse environmental sound textures to increase the robustness of audio classifiers
- Techniques: AutoEncoder, audio classification, FFT, Griffin-Lim algorithm, data augmentation
(3) Aug 2018 - May 2020 : Inferring True Dynamics from Incomplete Networks, RPI/Research Assistant | Python, Scikit-learn, MPI
- Developed a mean-field approach to infer nodes’ characteristics from incomplete networks
- Recovered true nonlinear dynamics with incomplete topology and equilibrium state information
- Designed a heuristic optimization algorithm based on our topology inference approach to solve K-SUM problem
- Solved large-scale nonlinear dynamical systems in parallel (10x speedup w/ MPI/SLURM)
- Techniques: graph sampling, link prediction, mean-field, optimization, parallel computing
(4) May - Aug 2018 : Online Early Warning Signals of Highway Traffic Breakdown, RPI/Research Assistant | Python, Scikit-learn, TensorFlow
- Preprocessed traffic time series data (denoising, smoothing)
- Designed and computed early warning signals of traffic congestion
- Trained a machine learning based composite early warning signal indicator
- Issued warning for upcoming traffic congestion
- Techniques: preprocessing, Gaussian kernel density estimation, machine learning
(5) May - Aug 2017 : Multiround Winner Determination, RPI/Research Assistant | Python, Java, TensorFlow, Keras
- Devised heuristic strategies (sampling, caching, pruning) to efficiently identify all tied winners in voting
- Developed reinforcement learning models to simulate voting procedures and improve the search efficiency
- Reduced run time by 50 $\sim$ 80% relative to the baseline strategies
- Techniques: voting, DFS w/ pruning, priority queue, reinforcement learning
(6) Sep - Dec 2016 : Matching Algorithm for OKCollege (now CollegeAI), RPI/Research Assistant | Java, Python, Keras
- Designed a bilinear model to match students and colleges’ preferences
- Trained a pairwise ranking model with synthesized dataset
- Techniques: machine learning, optimization
(7) May - Aug 2016 : Learning to Vote Fairly, RPI/Research Assistant | Java, Deeplearning4j
- Learned the fairness criteria in voting based on machine learning techniques
- Generated collections of ballots that satisfy specific fairness criteria
- Applied data augmentation to produce fair voting results with multiple fairness criteria
- Techniques: machine learning, optimization, data augmentation
REVIEWER OR PC MEMBER
ICML 2022, TheWebConf 2021, NetSci 2020, NetSci-X 2020, NERCCS 2019, COMPLEX NETWORKS 2018 ~ 2021, SciSec 2018
TEACHING ASSISTANT
- CSCI 4150: Introduction to Artificial Intelligence, RPI (Spring 17/18)
- CSCI 4100/6100: Machine Learning from Data, RPI (Fall 16)
COURSES
A COMPLETE LIST OF COURSES BEING TAKEN
SKILLS
- Language: Python, Java, C/C++, Matlab, MPI, HTML, Markdown, $\rm \LaTeX $
- Operating Systems: Linux, OS X, Windows
- Databases: MySQL, SQLite, MongoDB
- Softwares: TensorFlow, Keras, PyTorch, Pandas, Scikit-Learn, XGBoost, LightGBM, AWS/EC2, SLURM, Git, bash, vi
- Visualization: Matplotlib, TikZ, NetworkX, Graph-tool, Keynote, D3.js
POSTS
- Eco-Driving Data Imputation 05 May 2019
- Speed Prediction from Taxi Trajectory Data 10 Dec 2018
- A Short Review of Network Science 30 Nov 2018
- Learning to Vote 14 Oct 2018
HOBBIES
Mountain Biking, PingPong / Frédéric François Chopin, Michael Jackson, Charlie Chaplin, Freddie Mercury, Maksim Mrvica / Akira Kurosawa, Hayao Miyazaki, One Piece, The Phantom of the Opera