2 ms·
Many are dismissing it because it is just neural network. Which were basically at academic works since 80’s or earlier. It is only that now we have enough hard
by ozim 4y ago
Many are dismissing it because it is just neural network. Which were basically at academic works since 80’s or earlier. It is only that now we have enough hardware and material for training neural nets that they start to be useful.
- btilly 4y agoThere is a lot more than that. 1. We've learned that relu is a really good function to use for neural networks. 2. We've learned that regularization is essential to having large networks without overfitting. 3. We've learned how to scale models up and down so that you can train something on a giant cluster then run it on your phone. 4. We've learned how to apply this to lots of specific kinds of problems. etc, etc, etc. We do NOT simply do what we did in the 80s then throw more hardware at it. Instead we are getting better at making the hardware do something useful.