creators

Kaiming He - ResNet author, MIT and Google DeepMind

Kaiming He wrote ResNet, Mask R-CNN and Masked Autoencoders. He is an MIT EECS associate professor and a part-time distinguished scientist at Google DeepMind.

Kaiming He is an associate professor in MIT's Department of Electrical Engineering and Computer Science, where he holds the Douglas Ross (1954) Career Development Professorship of Software Technology, and a part-time distinguished scientist at Google DeepMind. He took a bachelor's degree at Tsinghua University in 2007 and a PhD at the Chinese University of Hong Kong in 2011, then worked at Microsoft Research Asia until 2016 and at Facebook AI Research until 2024, when he moved to MIT. He is the first author of "Deep Residual Learning for Image Recognition", the 2015 paper that introduced ResNet and the residual connection and made networks hundreds of layers deep trainable; Google Scholar's own reports named it the most cited paper of the preceding five years in both 2020 and 2021. He also co-wrote Mask R-CNN for instance segmentation, Focal Loss and RetinaNet for dense detection, and Masked Autoencoders for self-supervised pretraining of vision transformers. His recent work is on generative modelling. "Mean Flows for One-step Generative Modeling" was an oral at NeurIPS 2025 and "Back to Basics: Let Denoising Generative Models Denoise" was accepted at CVPR 2026. He teaches MIT's deep learning course, 6.7960.

Specialization

computer vision, deep learning architectures, self-supervised learning, generative models

Country

United States