If you ever asked yourself, what did suddenly changed the importance of neural networks (ahm... deep learning), this is a great run down. The turning point in m

If you ever asked yourself, what did suddenly changed the importance of neural networks (ahm... deep learning), this is a great run down. The turning point in my view was 2010 paper by Schmidhuber where he conclusively proves that deeper networks have been learning all the time and excelling compared to SVM/k-means. They were just too slow so needed to run for 24+ hours. GPU made this 40X faster and suddenly deep networks were getting trained in couple of hours.