Unsupervised learning of clutter-resistant visual representations from natural videos | Read Paper on Bytez