We present a plug-in replacement for batch normalization (BN) called exponential moving average normalization (EMAN), which improves the performance of existing student-teacher based self- and semi-supervised learning techniques. Unlike the standard BN, where the statistics are computed within each batch, EMAN, used in the teacher, updates its statistics by exponential moving average from the BN statistics of the student. This design reduces the intrinsic cross-sample dependency of BN and enhance the generalization of the teacher. EMAN improves strong baselines for self-supervised learning by 4-6/1-2 points and semi-supervised learning by about 7/2 points, when 1%/10% supervised labels are available on ImageNet. These improvements are consistent across methods, network architectures, training duration, and datasets, demonstrating the general effectiveness of this technique.

AI and Blockchain seem to be the new hammer -- making everything look like a nail, but are AI and Blockchain really necessary to solve most of the world's problems? What could AI do for Blockchain? AI LA's Third Thursday Talk Topic is Blockchain and we plan to get down and dirty discussing viable use cases and debunking all of the hype around two of the most overused buzzwords with some of the top minds working in both fields. Thank you to our amazing partners Phase Two, Semio.ai, LAtoken, and ObEN for helping us keep this event FREE!