- INFORMS Annual Meeting 2021 Virtual. A Stochastic Alternating Balance k-Means Algorithm for Fair Clustering. (presentation)
- MOPTA 2021: Lehigh University, Bethlehem, PA. A Stochastic Alternating Balance k-Means Algorithm for Fair Clustering. (presentation and session chair of “Fairness in ML and optimization”)
- INFORMS Annual Meeting 2020 Virtual. Accuracy and Fairness Trade-offs in Machine Learning: A Stochastic Multi-Objective Approach. (presentation)
- INFORMS Annual Meeting 2019: Seattle, WA. The Stochastic Multi-Gradient Algorithm for Multi-Objective Optimization and its Application to Supervised Machine Learning. (presentation)
- MOPTA 2019: Lehigh University, Bethlehem, PA. The Stochastic Multi-Gradient Algorithm for Multi-Objective Optimization and its Application to Supervised Machine Learning. (presentation)
- 12th Annual Machine Learning Symposium: New York, NY. (attendance)
- DIMACS/MOPTA 2018: Lehigh University, Bethlehem, PA. (summer school and conference attendance)
- MOPTA 2017: Lehigh University, Bethlehem, PA. (attendance)
- Efficient and Sparse Neural Networks by Pruning Weights in a Multiobjective Learning Approach. arXiv preprint arXiv:2008.13590.
- Inexact Derivative-Free Optimization for Bilevel Learning. arXiv preprint arXiv:2006.12674.
- Cocchi G, Liuzzi G, Lucidi S, Sciandrone M. On the convergence of steepest descent methods for multiobjective optimization. Computational Optimization and Applications. 2020 May 5:1-27.
Amazon Alexa Local Information Team (Remote, base at San Francisco)
Applied Scientist Intern Sep. 13 – Dec. 31, 2021
- Minimizing perplexity and LM score for the Geo-GDM dynamic language model.
Autodesk AutoCAD Machine Learning Team (Remote, base at San Francisco)
Machine Learning Engineering Intern Jun. 1st – Aug. 20, 2021
- Command sequences pattern analysis: N-gram analysis, word embedding model in NLP.
- Geometry block recommendations: geometry/graph embedding, convolutional neural network.
SAS Institute Analytics R&D / Scientific Computing Group (Virtual)
Graduate Intern May 19 – Aug 7, 2020
- Improve Hidden Markov Models using Black-Box optimization solver instead of the default multi-start mode (SAS CASL)
- Propose three promising tuning configurations and conducted numerical experiments for RS-AR models (discovering hidden stock market states)
- Saved 50-90% computation time in finding the best model and parameters estimation while achieving the same or less AIC errors
- More details about numerical results and analysis were published as a SAS OR Blog. Check out here
Sinotrans Air Transportation Development Co, Ltd., Beijing, China
Data Analyst Jul. 2014 – Sep. 2014
- Optimize the workflow of the Export department at North branch
- Collected and visualized operation data from the CA/BGS cargo consignment centers (MySQL, High charts)
- Conducted data analysis and lean improvement to identify bottlenecks and reduce idle time
Online courses taken
I have taken several interesting courses on Coursera, especially in Deep Learning Specialization offered by DeepLearning.AI. Here are the list of courses!
- Sequence Models, Feb 2022.
- Convolutional Neural Networks, July 2021.
- Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization, Mar 2021.
- Neural Network and Deep Learning, Jan 2021.
- Artificial Intelligence Data Fairness and Bias, Nov 2020.