Xinchen Yan
严欣辰 (My name in Korean: 엄흔진)
Email: skywalkeryxc [at] gmail [dot] com (primary)
           xcyan [at] umich [dot] edu (for CMT)
           xcyan [at] uber [dot] com (for Uber related)

@Königssee Germany, Sept. 2018

I am currently a research scientist at Uber Advanced Technologies Group (Uber ATG), San Francisco. I received my PhD degree from University of Michigan in 2019 under supervision of Prof. Honglak Lee. Previously, I received my Bachelor's degree from ACM Honored Class (交大ACM班), Shanghai Jiao Tong University in 2014. I was a visiting undergraduate student at Rapid-Rich Object Search (ROSE) LabNanyang Technological University (Singapore) in 2014, advised by Prof. Junsong YuanI also spent time working as an undergraduate research student at MOE-MSR Joint Key Lab of Intelligence Computing (BCMI Lab), advised by Prof. Liqing Zhang.

In the past, I spent my time interning at Adobe Research (San Jose, 2017) and Google Brain & X (Mountain View, 2016, 2017, 2018, & 2019). I feel very honored to receive Adobe Research Fellowship and Google PhD Fellowship which support my PhD research.

Research Interests
My primary research interests lie in the intersection of deep representation learning and generative modeling with structured and multimodal data. More specifically, my research has been focused on three related topics: (1) controllable image generation and manipulation, (2) deep generative modeling with structured data, and (3) weakly-supervised 3D shape generation with geometry-aware structures. Besides, I have a broad range of interests in developing deep generative models for vision, graphic, and robotic applications. 

Please see my PhD dissertation talk slides Learning Deep Controllable and Structured Representations for Image Synthesis, Structure Prediction and Beyond for a high-level summary of my research.

Internship Opportunities
If you are a PhD student or a highly motivated MS/UG student interested in doing research projects at Uber ATG, please send me an email with your resume and one-page research statement (recommended).

* 10/2019: I was honored to be one of top NeurIPS 2019 reviewers.
* 05/2019: I volunteered for AI4All summer program at University of Michigan in Summer 2019.
* 03/2019: Check out Pytorch implementation of our NeurIPS 2018 paper on Github.
* 01/2019: Check out TensorFlow implementation of our ECCV 2018 paper on Github.
11/2018: Check out Project page of our NeurIPS 2018 paper.
* 08/2018: Recipient of Adobe Research Fellowship and Google PhD Fellowship. I really appreciate the support from my PhD advisor, committee members at U of Michigan, collaborators in academia and industry.
* 07/2018: Check out Project page of our ECCV 2018 paper.
* 01/2018: Check out Project page of our ICRA 2018 paper.
* 07/2017: Check out TensorFlow implementation of our NeurIPS 2016 paper on Github.
* 12/2016: Check out Torch implementation of our NeurIPS 2016 paper (unsupervised learning of 3D shape generation) on Github.
* 10/2016: Check out Torch implementation of our ECCV 2016 paper (conditional image generation) on GitHub.

Education & Research Experience (full in Mandarin)
(09/2019 -- current) AI Research Scientist@Uber ATG, San Francisco
(06/2018 -- 04/2019) Student Researcher Intern@Google Brain, Mountain View 
(06/2017 -- 09/2017) Research Intern@Adobe Systems Inc, San Jose
(11/2016 -- 06/2017) Research Intern@Google Brain, Mountain View
(08/2014 -- 07/2019) PhD@University of Michigan, Ann Arbor

Professional Activities
* Conference Reviewer: NeurIPS (2016 -- 2019), ICML (2018 -- 2020), ICLR (2018 -- 2020), CVPR (2019 -- 2020)
ICCV 2019, ECCV 2020, WACV (2019 -- 2020), CoRL 2018
* Journal Reviewer
    - IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
    - IEEE Transactions on Image Processing (TIP)
    - IEEE Transactions on Multimedia (TMM)

Invited Talks
(03/2020) Tech Talk@Stanford Univ, Palo Alto, US
(02/2020) Tech Talk@Uber AI, San Francisco, US
(08/2019) Tech Talk@Didi Lab, Mountain View, US
(04/2019) Tech Talk@Google AI, Mountain View, US
(03/2019) Tech Talk@Facebook AI Research, Menlo Park, US
(03/2019) Tech Talk@Waymo, Mountain View, US
(02/2019) Tech Talk@Uber ATG, Toronto, Canada
(01/2019) Tech Talk@Adobe Research, San Francisco, US
(01/2019) Tech Talk@Nvidia Research, Santa Clara, US

Honors & Awards
(04/2018) Google PhD Fellowship in Machine Learning (8 selected world-wide) [News from Google AI Blog]
(12/2017) Adobe Research Fellowship (10 selected in USA) [News from Adobe Research Blog]
(12/2012) Academic Excellence Scholarship (Second-class) of Shanghai Jiao Tong University in Year 2011-2012
(12/2011) Academic Excellence Scholarship (Second-class) of Shanghai Jiao Tong University in Year 2010-2011
(Year 2008, 2009, & 2010) First Prizes in National Olympiad in Informatics in Provinces

SemanticAdv: Generating Adversarial Examples via Attribute-conditional Image Editing
Haonan Qiu, Chaowei Xiao, Lei Yang, Xinchen Yan, Honglak Lee, Bo Li

* Data-Efficient Learning for Sim-to-Real Robotic Grasping using Deep Point Cloud Prediction Networks
Xinchen Yan, Mohi Khansari, Jasmine Hsu, Yuanzheng Gong, Yunfei Bai, Sören Pirk, Honglak Lee
Technical Report [arxiv]

Learning Hierarchical Semantic Image Manipulation through Structured Representations
Seunghoon Hong, Xinchen Yan, Thomas Huang, Honglak Lee 
In Advances in Neural Information Processing Systems (NeurIPS), Montreal, Canada, 2018

MT-VAE: Learning Motion Transformations to Generate Multimodal Human Dynamics
Xinchen YanAkash Rastogi, Ruben Villegas, Kalyan Sunkavalli, Eli Shechtman, Sunil Hadap, Ersin Yumer, Honglak Lee
In European Conference on Computer Vision (ECCV), Munich, Germany, 2018

* Learning 6-DOF Grasping Interaction via Deep Geometry-aware 3D Representations
Xinchen YanJasmine Hsu, Mohi Khansari, Yunfei Bai, Arkanath Pathak, Abhinav Gupta, James Davidson, Honglak Lee
In Proceedings of International Conference on Robotics and Automation (ICRA)Brisbane, Australia, 2018

* Deep Variational Canonical Correlation Analysis
Weiran Wang, Xinchen Yan, Honglak Lee, Karen Livescu
Technical Report [arxiv]
TensorFlow Implementation: [code]

Perspective Transformer Nets: Learning Single-View 3D Object Reconstruction without 3D Supervision
Xinchen Yan, Jimei Yang, Ersin Yumer, Yijie Guo, Honglak Lee
In Advances in Neural Information Processing Systems (NeurIPS), Barcelona, Spain, 2016
Torch Implementation: [data][code]
TensorFlow Implementation: [code]
Note, if you want to use your own camera matrix, please refer to the section "using your own camera" here.

* Generative Adversarial Text-to-Image Synthesis
Scott Reed, Zeynep Akata, Xinchen Yan, Lajanugen Logeswaran, Bernt Schiele, Honglak Lee
In Proceedings of the 33rd International Conference on Machine Learning (ICML), New York, USA, 2016
Torch Implementation: [code]

* Attribute2Image: Conditional Image Generation from Visual Attributes
Xinchen Yan, Jimei Yang, Kihyuk Sohn, Honglak Lee
In European Conference on Computer Vision (ECCV), Amsterdam, The Netherlands, 2016
Torch Implementation: [code]

* Learning Structured Output Representation using Deep Conditional Generative Models
Kihyuk Sohn, Xinchen Yan, Honglak Lee
In Advances in Neural Information Processing Systems (NeurIPS), Montreal, Canada, 2015. 

* A Unified Framework for Automatic Wound Segmentation and Analysis with Deep Convolutional Neural Networks
Changhan Wang, Xinchen Yan, Max Smith, Kanika Kochhar, Marcie Rubin, Stephen M. Warren, James Wrobel, Honglak Lee
In 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Milan, Italy, 2015.

* Efficient Online Spatio-Temporal Filtering for Video Event Detection
Xinchen Yan, Junsong Yuan, Hui Liang, Liqing Zhang
In ECCV Workshop on Video Event Categorization, Tagging and Retrieval towards Big Data (VECTaR 2014), Zurich, Switzerland, 2014.
C++/MATLAB Implementation: [code]

Teaching Assistant
(09/2011 - 01/2012) Teaching Assistant of C++ Programming (Fall 2011)
(03/2012 - 06/2012) Teaching Assistant of Data Structures (Spring 2012)
(09/2012 - 12/2012) Assistant Lecturer of Math Foundations in Computer Science (Fall 2012)
Xinchen Yan,
Jul 31, 2014, 2:27 AM
Xinchen Yan,
Jul 31, 2014, 1:48 AM