I have a broad interest in both the math and the scientific worlds, and I am open to new knowledge at all time. Currently, I am a research engineer at Deepmind, with a focus on applying pioneer machine learning techniques to large-scale, real-world products.

In 2021, I defended my PhD from the University of Amsterdam, supervised by my amazing advisors Prof. Max Welling and Dr. Herk van Hoof.  My research interests include representation learning, generative modeling, and reinforcement learning.

My PhD thesis (here) is dedicated to Granvil Keith Geroge, whose spirit has supported me throughout my research career. 

My undergraduate honor thesis advisor, Professor Bourdon, also has nourished me with the heart of humanity: compassion, caring and selflessness.  

My Masters degree was in Applied Mathematics from the University of Michigan.  

 

Wendy Shang

E-mail: wendyshang at deepmind.com

Publications and Preprints

A Moment-based Approach for DVH-guided Radiotherapy Treatment Plan.  

M. Zarepisheh, M. Shakourifar, G. Trigila, P. Ghomi, S. Couzens, A. Abebe, L. Norena, W. Shang, S. Jiang and Y. Zinchenko. Physics in Medicine and Biology, 2013.

Normal Weighted Composition Operators on Weighted Dirichlet Spaces. 

L. Lu, Y. Nakada, D. Nestor, W. Shang and R. Weir. JMAA, 2014.
 

Improved Multimodal Deep Learning with Variation of Information. 

K. Sohn, W. Shang, H. Lee. NIPS, 2014.

Reproducing Kernel Hilbert Spaces Supporting Nontrivial Hermitian Weighted Composition Operators. 

P. Bourdon and W. Shang. Complex Analysis and Operator Theory, 2011. 

Talks and Posters

Discriminative Training Of Structured Dictionaries Via Orthogonal Matching Pursuit. 

Women in Machine Learning Workshop, 2014.

The Weighted Composition Operators in Hilbert Spaces. (won the poster award)

the Joint Math Meetings Student Poster Session, 2011.

Discriminative Training of Structured Dictionaries via Block Orthogonal Matching Pursuit

W. Shang, K. Sohn, and H. Lee. Siam SDM, 2016.

On the Norm Preservation Properties of Convolutional Neural Networks. 

Women in Machine Learning Workshop, 2015. 

Understanding and Improving Convolutional Neural Networks via Concatenated Rectified Linear Units

W. Shang, K. Sohn, D. Almeida and H. Lee. ICML, 2016.

Exploring Normalization in Deep Residual Networks with Concatenated Rectified Linear Units

W. Shang, J. Chiu, and K. Sohn. AAAI, 2017.

ELF: An Extensive, Lightweight and Flexible Research Platform for Real-time Strategy Games

Yuandong Tian, Qucheng Gong, W. Shang, Yuxin Wu and Larry Zitnick. NIPS, 2017.

Channel-Recurrent Autoencoding for Image Modeling

W. Shang, K. Sohn, Y. Tian. WACV, 2018.  [code and PDF]

Stochastic Activation Actor-Critic Methods. 

W. Shang,  H. van Hoof, M. Welling. ECML-PKDD 2019. [code and PDF]

selected as a contributive talk at UAI, Uncertainty in Deeplearning Workshop, 2018. 

Thesis Supervision

Convolutional Neural Networks for Crop Yield Prediction using Satellite Images. 

Helena Russello, Masters of AI, 2018. (co-advised with Dr. Santiago Gaitan, IBM)

An Empirical Comparison of Deep Conditional Generative Models with Applications to Attribute-Guided Face Synthesis. 

Kewin Dereniewicz, Masters of Data Science, 2018. (co-advised with Benjamin Timmermans, IBM)

Attentive Conditional Channel-Recurrent Autoencoding for Attribute Conditioned Face Synthesis

W. Shang, K. Sohn. WACV, 2019.  [code and PDF]

Unsupervised Domain Adaptation for Distance Metric Learning

K. Sohn. W. Shang, X. Yu, M. Chandraker. ICLR, 2019.  

Learning World Graphs for Accelerated Hierarchical Reinforcement Learning. 

W. Shang,  A.Trott*,  S. Zheng*,  C. Xiong, R. Socher. ICML Real-world Sequential Decision Making workshop, 2019. [project page]

Decompose Video Representations from Temporal Coherence and Dynamics. 

W. Shang,  S. Liu, A. Vahdat, S. De Mello, J. Kautz. 2019 [project page]

Reinforcement Learning with Latent Flow. 

W. Shang*,  X. Wang*, A. Lakshminarayanan, A. Rajeswaran, Y. Gao, P.  Abbeel, M. Laskin

Neurips 2021. [code and PDF]

selected as a contributive talk at Neurips, Deep RL Workshop, 2020. 

Agent-Centric Representations for Multi-agent Reinforcement Learning. 

W. Shang,  L. Espeholt, A. Raichuk, T. Salimans. 2020 [project page]

Experiences

Reviewer. 
Neurips, ICML, CVPR, ICCV, UAI, AISTATS, AAAI,

Neural Computation,

Transactions on Computational Intelligence and AI in Games, Transactions on Neural Networks and Learning Systems .

 

 

Computer Vision Software Engineer. 

Oculus, Menlo Park, CA, Spring 2016-Fall 2017. 

Research Intern. 

Salesforce Research, Palo Alto, CA, Spring 2019. 

Research Intern. 

NVIDIA Research, Santa Clara, CA, Summer 2019. 

Research Engineer. 

Deepmind, Mountain View, CA, Fall 2020-Now. 

Research Intern. 

Google Brain, Amsterdam, NL, Spring 2020. 

Visiting Scholar. 

UC Berkeley, Berkeley, CA, Summer 2020.