Ping (iris) Yu Ph.D. candidate Department of Computer Science New York State University at Buffalo
Short Bio
I am a third-year Ph.D. candidate at the University at Buffalo of New York, supervised by Dr. Changyou Chen. In general, my research interest includes:
· Natural language process including multi-lingual models and text generation · Multimodal research including the interaction between natural language and computer vision · Deep generative models such as VAE and GAN · Deep reinforcement learning
My previous work focuses on learning better feature representations for long sequences and generating sequences with deep generative models, with applications to sentence generation and action generation. Before that, I received a master's degree from the Department of Electrical and Computer Engineering at the University of Michigan, Ann Arbor in 2018, and my bachelor's degree from the Department of Electrical and Computer Engineering at Central South University, Changsha, China in 2016.
News: - May.2021: I will work with Facebook ai on multi-lingual and multi-modal research.
- Nov.2020: I finished the internship with Amazon working on a multilingual language model for the natural language understanding unit for Alexa Conversation System.
- Aug.2020: I finished the internship with Tencent AI research working on a document summarization system to automatically write summaries for posts on the game forums, which contains noisy data from the special domain.
- July.2020: One paper has been accepted to ECCV 2020.
- May.2020: One paper has been accepted to ICML 2020.
- Dec.2019: One paper has been accepted to ICLR 2020.
- Nov.2019: One paper has been accepted to AAAI 2020.
- Aug.2019: I finished an internship with PlayStation working on physics-based character motion.
- Jun.2019: My recent work for text generation was orally presented in the imitation workshop, ICML 2019.
2020Structure-Aware Human-Action Generation [PDF] [Code] [Blog] Ping Yu, Yang Zhao, Chunyuan Li, Changyou Chen the European Conference on Computer Vision (ECCV), 2020. Provided a way to leverage structural information in long sequences
Feature Quantization Improves GAN Training [PDF] [Code] Yang Zhao, Chunyuan Li, Ping Yu, Changyou Chen International Conference on Machine Learning (ICML), 2020. A simple pug-in to improve BigGAN, StyleGAN, and U-GAT-IT for large-scale image generation/translation
Learning Diverse Stochastic Human-Action Generators by Learning Smooth Latent Transitions (Spotlight) [PDF] Zhenyi Wang, Ping Yu, Yang Zhao, Yufan Zhou, Ruiyi Zhang, Changyou Chen Association for the Advancement of Artificial Intelligence (AAAI), 2020.
Bayesian Meta Sampling for Fast Uncertainty Adaptation [PDF] Zhenyi Wang, Yang Zhao, Ping Yu, Ruiyi Zhang, Changyou Chen International Conference on Learning Representations (ICLR), 2020.
Discretized Bottleneck in VAE: Posterior-Collapse-Free Sequence-to-Sequence Learning [PDF] Yang Zhao, Ping Yu, Suchismit Mahapatra, Qinliang Su, Changyou Chen 2019Self-Enhanced Inverse Reinforcement Learning for Text Generation (Oral) [PDF] Ping Yu, Ruiyi Zhang, Yizhe Zhang, Chunyuan Li, Changyou Chen Imitation, Intent, and Interaction, International Conference on Machine Learning (ICML), 2019.
2016Low-power wireless sensor network protocol of mobile health based on IPv6 [PDF] Lulu Wang, Shuai Hao, Ping Yu, Zhiwu Huang 35th Chinese Control Conference (CCC), 2016.
2015Modeling of point search area and rescue path for maritime air crash [PDF] Ping Yu, Jing Wang, Zhentao Liu, Haoran Bian 34th Chinese Control Conference (CCC), 2015.
Professional Service Program Committee & Reviewer: IJCAI 2021; AAAI 2021; ACL 2021; NeurIPS 2021. |
|