We propose a novel algorithm that focuses on the less studied relationship between the unlabeled data. Our algorithm achieves state-of-the-art performance on standard benchmarks
We propose a novel algorithm for semi-supervised classification that achieves state-of-the-art performance on standard benchmarks and outperform previous works on transfer setting by a large margin.
This tutorial shows how to correctly implement EMA for PyTorch
We designed and evaluated a novel framework for robot mediation of a support group
We designed a decentralized multi-task learning framework and a novel optimization algorithm to collaboratively train models over distributed devices by only sharing their gradient periodically