This program is used to demonstrate the use of a multi-layer perceptron used on a song prediciton dataset found at kaggle.com ( https://www.kaggle.com/c/kkbox-music-recommendation-challenge). The purpose of this is to demonstrate the use of tensorflow to create a feed forward network that creates a straightforward graph for new users.
This scirpt uses python 3.5 and tensorflow 1.1.0 Supporting libraries are : pandas for data structuring sklearn for model accuracy and splitting our train and test set
I reccomend using a python virtual environment specifically for use of tensorflow
def hidden_layer(x, channels_in, channels_out,activation = None, pk = None, drop = False,name='hlayer'):
with tf.name_scope(name):
W = tf.Variable(tf.zeros([channels_in, channels_out]),name = 'Weights')
b = tf.Variable(tf.zeros([channels_out]), name = 'Bias')
if activation is 'relu':
act = tf.nn.relu(tf.matmul(x,W) + b)
if activation is 'sig':
act = tf.nn.sigmoid(tf.matmul(x,W) + b)
if activation is 'soft':
act = tf.nn.softmax(tf.matmul(x,W) + b)
if activation is 'tanh':
act = tf.nn.tanh(tf.matmul(x,W) + b)
else:
act = tf.matmul(x, W) + b
if drop is True:
act = tf.nn.dropout(act, pk)
return actIs the basis of building our network, it is a simple funciton I put together to allow you to build easy MLP layers.
x: Input
channels_in/out: The amount of tensors going and and out of a node respectively
activation: This can be relu,sigmoid,softmax or the hyperbolic tangent
pk: Percent keep if dropout is used on a layer
drop: Weather or not to use layer dropout
Each run builds a clean tensorboard graph that is used to demonstrate the framework of the network it will be put into a folder graph12 and can be run from command line by:
activate tensorflow virtual environment
tensorboard --logdir graph12
From here you can go to your localhost:6006 and under the "graphs" section you may view your interactive board
Jacob Biloki : bilokij@gmail.com
This project is licensed under the MIT License