M2M Day 354: The easy way to implement a machine learning model
This post is part of Month to Master, a 12-month accelerated learning project. For October, my goal is to defeat world champion Magnus Carlsen at a game of chess.
Today, I finished writing the small Python script that converts chess games downloaded from the internet into properly formatted data needed to train my machine learning model.
Thus, today, it was time to start building out the machine learning model itself.
Rather than starting from scratch, I instead looked for an already coded-up model on Github. In particular, I needed to find a model that analogizes reasonably well to chess.
I didn’t have to look very hard: The machine learning version of “Hello World” is called MNIST, and it works perfectly for my purposes.
MNIST is a dataset that consists of 28 x 28px images of handwritten digits like these:
The dataset also includes ten labels, indicating which digit is represented in each image (i.e. the labels for the above images would be 5, 0, 4, 1).
The objective is to craft a model that, given a collection of 28 x 28=784 values, can accurately predict the correct numerical digit.
In a very similar way, the objective of my chess model, given a collection of 8 x 8 = 64 values (where each value is represented using 12-digit one-hot encoding), is to accurately predict whether the chess move is a good move or a bad move.
So, all I need to do is download some example code from Github, modify it for my purposes, and let it run. Of course, there are still complexities with this approach (i.e. getting the data in the right format, optimizing the model for my purposes, etc.), but I should be able to use already-existing code as a solid foundation.
Here’s the code I found:
from __future__ import absolute_import
from __future__ import division
from __future__ import print_functionimport argparse
import sysfrom tensorflow.examples.tutorials.mnist import input_dataimport tensorflow as tfFLAGS = Nonedef main(_):
# Import data
mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)# Create the model
x = tf.placeholder(tf.float32, [None, 784])
W = tf.Variable(tf.zeros([784, 10]))
b = tf.Variable(tf.zeros([10]))
y = tf.matmul(x, W) + b# Define loss and optimizer
y_ = tf.placeholder(tf.float32, [None, 10])# The raw formulation of cross-entropy,
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)sess = tf.InteractiveSession()
tf.global_variables_initializer().run()
# Train
for _ in range(1000):
batch_xs, batch_ys = mnist.train.next_batch(100)
sess.run(train_step, feed_dict={x: batch_xs, y_: batch_ys})# Test trained model
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
print(sess.run(accuracy, feed_dict={x: mnist.test.images,
y_: mnist.test.labels}))if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--data_dir', type=str, default='/tmp/tensorflow/mnist/input_data',
help='Directory for storing input data')
FLAGS, unparsed = parser.parse_known_args()
tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)
Tomorrow, I’ll take a crack at modifying this code, and see if I can get anything working.
Read the next post. Read the previous post.