Yingtian Tang

I am currently a PhD student at EPFL, working on NeuroAI with Prof. Martin Schrimpf.

I had my Master's degree at University of Pennsylvania, majoring in Computer and Information Science (CIS). I was also a research assistant at the GRASP Lab, supervised by Prof. Pratik Chaudhari.

I had my undergraduate at University of Electronic Science and Technology of China, majored in Computer Science. During my undergraduate, I also worked as research assistants in Nanyang Technological University (Rapid-Rich Object Search Lab), and later Tencent (Robotics X).

Contact

yingtian.tang [at] epfl dot ch

Curriculum Vitae (Transcript) [slightly outdated]

Research Interests

I am interested in artifitial intelligence, especially how current machine learning methods could approach active perception and learning in the real world. Unfortunately, despite searching for it for a long time, it is not until last year that I finally recognized this line of research as the intersection between computational neuroscience, computational cognitive science, and machine learning. I then studied and was impressed by the research by people like James J. Gibson and Peter Dayan, and believe that machine learning will take them further (or conversely, they will guide a right way for ML research).

Before this, I have gone through multiple fields to study machine learning broadly, including graph learning, visual representation learning, reinforcement learning for optimization, and motion synthesis (in time order; please see CV for more details).

Currently, I am working on building models for perceptual representation, based on temporal slowness (specifically, Slow Feature Analysis). I would like to study how "objectness" could arise from the learned representations. Meanwhile, I am investigating the recent slot-attention-based object-centric models, to see if they are robust for complex real-world scenes.

Publications

Online Motion Style Transfer for Interactive Character Control
Yingtian Tang, Jiangtao Liu, Cheng Zhou, and Tingguang Li
Arxiv Preprint (2021)
Link

Learning-Aided Heuristics Design for Storage System
Yingtian Tang, Han Lu, Xijun Li, Lei Chen, Mingxuan Yuan, and Jia Zeng
The 2021 ACM SIGMOD/PODS International Conference on Management of Data
Link

Accurate probabilistic miss ratio curve approximation for adaptive cache allocation in block storage systems
Rongshang Li, Yingtian Tang, Qiquan Shi, Hui Mao, Lei Chen, Jikun Jin, Peng Lu, and Zhuo Cheng
2022 Design, Automation & Test in Europe Conference & Exhibition (DATE)
Link