Is it possible to learn a face detector using only unlabeled images?
Quoc V. Le, Andrew Ng, Jeff Dean et al., June 2012 (Proceedings of the 29th International Conference on Machine Learning (ICML 2012) / arXiv). Researchers at Google and Stanford describe a network trained on 10 million YouTube frames using 16,000 cores for three days, later known as the Google cat p
Building High-level Features Using Large Scale Unsupervised Learning
AI and science · Before ChatGPT
Who else?
What it refers to: A 9-layered sparse autoencoder trained without labels, which the paper reports learned a face detector and also responded to cat faces and human bodies.
Other voices
- “Learn representations of data with multiple levels of abstraction” Yann LeCun, Yoshua Bengio and Geoffrey Hinton,
- “Simply by waiting for faster GPUs and bigger datasets” Alex Krizhevsky, Ilya Sutskever and Geoffrey Hinton,
- “Simple models and a lot of data trump more elaborate models” Halevy et al.,
- “Whatever you see there, I want more of it” Alexander Mordvintsev et al.,
- “Darker-skinned females are the most misclassified group” Joy Buolamwini,
