I have been engaged in researching the power of AI technologies in the area of Japanese Anime/Game for over 5 years. My long-term goal is to revolutionize the Japanese Anime/Game industry. Meanwhile, as an entrepreneur, I founded Crypko.AI in 2017, which was acquired by Preferred Networks in late 2018.

Email: yanghuajin94 [at] gmail [dot] com

Academic Publications

Surrogate Gradient Field for Latent Space Manipulation
IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2021
Minjun Li*, Yanghua Jin*, Huachun Zhu (* Equal Contribution)
[paper]

Amortized Nesterov’s Momentum: Robust and Lightweight Momentum for Deep Learning
The Conference on Uncertainty in Artificial Intelligence (UAI) 2020
Kaiwen Zhou, Yanghua Jin, Qinghua Ding, James Cheng
[paper][supp][code]

Fast Soft Color Segmentation
IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020
Naofumi Akimoto, Huachun Zhu, Yanghua Jin, Yoshimitsu Aoki
[paper][github]

Towards the Automatic Anime Characters Creation with Generative Adversarial Networks
Advances in Neural Information Processing Systems (NIPS) 2017 Workshop on Machine Learning for Creativity and Design (spotlight) Yanghua Jin, Jiakai Zhang, Minjun Li, Yingtao Tian, Huachun Zhu
[paper][web site]

Talks

Crypko: A New Workflow for Anime Character Creation
12th CODH Seminar (Online) AI for Culture: From Japanese Art to Anime, NII ROIS-DS, Aug 05, 2020
[YouTube] [slides]

Exploring the Anime Characters Creation with Generative Adversarial Networks
Deep Learning: Theory, Algorithms, and Applications, RikenAIP, Tokyo, Mar 19-22, 2018
[YouTube]

Create Anime Characters using AI
AI for Manga & Anime, Anime Expo 2018, Los Angeles, 2018
[slides]

Timeline

Sizigi Studios, ML Advisor
2022.4 - present

Preferred Networks, Research Engineer
2018.11 - 2022.4
Deep Learning R&D in the area of Japanese anime.

Crypko.AI, Founder
2017.12 - 2018.10
Building next-generation cryptocollectible game with GAN and Ethereum. Director, Machine Learning Development, Backend Development. This project is acquired by Preferred Networks in late 2018.

Fudan University, B.S. with honors in Computer Science, Shanghai, China
2013 - 2018