symmetric_functions_neural_network

We show that TensorFlow DNN.classifier fails to learn symmetric polynomials: To run the script I used conda package manager to activate tensorflow environment. Below is the information about the version of conda I used

 active environment : tensorflow
	active env location : /anaconda3/envs/tensorflow
					shell level : 1
		 user config file : /Users/ra312/.condarc

populated config files : /Users/ra312/.condarc conda version : 4.5.11 conda-build version : 3.4.1 python version : 3.6.4.final.0 base environment : /anaconda3 (writable) channel URLs : https://repo.anaconda.com/pkgs/main/osx-64 https://repo.anaconda.com/pkgs/main/noarch https://repo.anaconda.com/pkgs/free/osx-64 https://repo.anaconda.com/pkgs/free/noarch https://repo.anaconda.com/pkgs/r/osx-64 https://repo.anaconda.com/pkgs/r/noarch https://repo.anaconda.com/pkgs/pro/osx-64 https://repo.anaconda.com/pkgs/pro/noarch package cache : /anaconda3/pkgs /Users/ra312/.conda/pkgs envs directories : /anaconda3/envs /Users/ra312/.conda/envs platform : osx-64 user-agent : conda/4.5.11 requests/2.18.4 CPython/3.6.4 Darwin/17.7.0 OSX/10.13.6 UID:GID : 501:20 netrc file : None offline mode : False

Creating a tensorflow environment in conda conda create -n tensorflow You can use either conda or pip to install TensorFlow.

Running the script

python my_estimator.py

Test set accuracy: 0.788

Prediction is "0" (99.7%), expected "0"

Prediction is "3" (53.6%), expected "1"

Prediction is "2" (54.1%), expected "2"

Prediction is "2" (39.4%), expected "3"

Labels description: "0" -> x1 + x2 + x3 + x4

"1" ->x1x2 + x1x3 + x1x4 + x2x3 + x2x4 + x3x4

"2" -> x1x2x3 + x1x2x4 + x1x3x4 + x2x3x4

"3" -> x1x2x3*x4