LoginSignup
0

More than 5 years have passed since last update.

posted at

updated at

Deep Learning > link > Can Deep Learning (and Neural Networks) be useful for regression problems where the output variable has an unknown (or varying in real time) upper bound?

Deep Learningで学習を検討しているのは「多クラス分類」ではなく「回帰」。

以下を見つけた。

気になったのは

Vicent Ribas Ripoll, I solve classification problems.
Written Aug 16, 2013

Yes they are. One only has to make sure to use the proper cost function (MSE) and adapt the network to provide a continous output (this normally means removing the softmax in the output). Some examples are:

Page on Stanford

上記のPDFファイルではPower Loadという多数の山と谷を持つ形状の学習をしている。
誤差(%RMSE)は以下

Learning Method %RMSE
Kernelized Regression 8.3%
Frequency NN 6.7%
Deep Feedforward NN 5.9%
Deep Recurrent NN 2.8%

自分が必要な精度(system of linear equationの初期値の誤差)がどれくらいのものかは実際に実験をしないと分からない。

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
What you can do with signing up
0