||
跟ICML'13一篇paper(Maxout Networks)的作者讨论一个bio-inspired deep learning的想法,得到的回复是:你的想法没啥硬伤,关键是能不能跑出更好的结果。
"Other people, including me, having worked on fairly similar things and didn't really get them to work. That doesn't mean it's impossible though; just that you might need to try a few different tricks to go with it that we haven't tried."
"In general it's very hard to predict theoretically how well a deep learning method will work in advance. You just have to get your hands dirty and try a lot of them."
虽说“实践是检验真理的唯一标准”没错,但还是觉得,现阶段的DL不像科学,更像炼金术!
接下来打算好好玩玩pylearn2和cuda-convnet了。
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-10-20 02:18
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社