On the hardness of learning under symmetries
The researchers investigate the problem of learning equivariant neural networks through gradient descent. Despite the known symmetries improving the performance of learning pipelines in fields like biology and computer vision,…
Continue reading