Kaiming He is a computer scientist and the Douglas Ross (1954) Career Development Professor of Software Technology and associate professor in MIT’s department of electrical engineering and computer science. His research is in artificial intelligence and machine learning, with a focus on computer vision. Before MIT he was a research scientist at Facebook AI Research and earlier at Microsoft Research Asia.
He is best known as the lead author of ResNet, short for “deep residual learning,” introduced in 2015. ResNet solved a stubborn problem: as neural networks were made deeper, they paradoxically became harder to train and performed worse. His team’s fix was the “residual connection,” a shortcut that lets a layer learn only the change it needs to make to its input rather than the whole transformation. This simple idea let networks grow to hundreds of layers, and ResNet won the major image-recognition competition of its year. He is also a lead author of related landmarks such as Faster R-CNN and Mask R-CNN.
For the library’s reader, He is the figure behind one of the most quietly consequential ideas in modern deep learning. The residual connection is now a standard building block, found not only in vision models but inside the transformer architecture that powers large language models. ResNet is part of the answer to why, after 2015, “deeper” stopped being a liability and became the default path to better performance.