Re:Google acquires neural network startup that may help it h...
by gcr on Wednesday March 13, 2013 @01:13PM
(#5369695)
It's my understanding that deep learning networks have to be tuned almost perfectly to perform well. Not my area ef expertise, but I'm not convinced just yet.
[ Parent
| on HN ]
Re:Google acquires neural network startup that may help it h...
by SatvikBeri on Wednesday March 13, 2013 @01:54PM
(#5369963)
It's true that Deep Belief Networks take a lot of work compared to, say, Random Forests, but they make good Neural Networks much easier.
Before Deep Learning, Neural Networks required a lot of hand tuning. Without that kind of tuning, Neural Networks with more than 1 or 2 hidden layers tended to perform very badly, because they would get stuck in plateaus or poor local optima.
In contrast, (one facet of) Deep Learning focuses on using unsupervised learning to better initialize NNs, which in turn allows for much higher levels of accuracy. In essence, Deep Learning helps pick the NN parameters automatically. So it's actually decreasing the amount of hand tuning needed.
Re:Google acquires neural network startup that may help it h...
by dia80 on Wednesday March 13, 2013 @01:25PM
(#5369774)
In fact almost the opposite! Unsupervised pre-training learns the statistical invariants in the inputs and these become the features.
This is in contrast to hand engineering features for your particular domain which, unsurprisingly, doesn't generalise well.
Re:Google acquires neural network startup that may help it h...
by dumitrue on Wednesday March 13, 2013 @02:08PM
(#5370051)
Basically not true (anymore). If you take the model that is described by Hinton et al and apply it to some vision task for which you generated a lot of data, it will invariably work better than the competition, almost out of the box (yeah you have to try a few learning rates, but that basically the case for any ML algorithm). It's pretty amazing actually.