Timezone: »
Conditional Generative Adversarial Networks are known to be difficult to train, especially when the conditions are continuous and high-dimensional. To partially alleviate this difficulty, we propose a simple generator regularization term on the GAN generator loss in the form of a Lipschitz penalty. The intuition of this Lipschitz penalty is that, when the generator is fed with neighboring conditions in the continuous space, the regularization term will leverage the neighbor information and push the generator to generate samples that have similar conditional distributions for neighboring conditions. We analyze the effect of the proposed regularization term and demonstrate its robust performance on a range of synthetic tasks as well as real-world conditional time series generation tasks.
Author Information
Yunkai Zhang (UC Berkeley)
Yufeng Zheng (University of California, Berkeley)
Amber Ma (Apple)
Graduated from Columbia University with a Masters degree in Operations Research in 2018, and a bachelor's degree in mathematics from the University of Toronto. Keen on using statistical methods to solve business problems, especially time series and pricing related tasks. Worked as a Data Scientist in Research & Applied Science, Computational Economics team at Wework. The project I currently work on focuses on providing pricing and discount suggestions for sales agents, as well as developing cutting edge algorithms to quantify geographic information to help with the business's site selection (demand analysis). Previously, I interned at New York Life's Macroeconomics research group and worked on developing models by implementing ideas from papers in order to predict the recession probability and the default rate, both models outperformed the existing models' performances.
Siyuan Teng (University of California, Berkeley)
Zeyu Zheng (University of California Berkeley)
More from the Same Authors
-
2022 Poster: An Adaptive Deep RL Method for Non-Stationary Environments with Piecewise Stable Context »
Xiaoyu Chen · Xiangming Zhu · Yufeng Zheng · Pushi Zhang · Li Zhao · Wenxue Cheng · Peng CHENG · Yongqiang Xiong · Tao Qin · Jianyu Chen · Tie-Yan Liu -
2022 Poster: A Simple and Optimal Policy Design for Online Learning with Safety against Heavy-tailed Risk »
David Simchi-Levi · Zeyu Zheng · Feng Zhu -
2022 Poster: Gradient-Free Methods for Deterministic and Stochastic Nonsmooth Nonconvex Optimization »
Tianyi Lin · Zeyu Zheng · Michael Jordan -
2021 Poster: Stochastic $L^\natural$-convex Function Minimization »
Haixiang Zhang · Zeyu Zheng · Javad Lavaei