# Construct cost cost_sparse = n_beta * tf.reduce_sum(KL_Div(n_rho, rho_hat)) cost_J = tf.reduce_mean(tf.nn.l2_loss(pred['out'] - x)) cost_reg = n_lambda * (tf.nn.l2_loss(weights['hidden']) + tf.nn.l2_loss(weights['out'])) cost = cost_J + cost_reg + cost_sparse optimizer = tf.train.GradientDescentOptimizer(learning_rate=n_learning_rate).minimize(cost) # Initializing the variables init = tf.initialize_all_variables() # Launch the graph with tf.Session() as sess: sess.run(init) saver = tf.train.Saver() # Training cycle for epoch in range(n_num_epochs): batch_xs=gid.getPatches(n_epoch_size) # Fit training using batch data sess.run(optimizer, feed_dict={x: batch_xs}) print("Optimization Finished!") saver.save(sess, 'my-SAE') outWeights = sess.run(weights['hidden']) vis.display_network(outWeights)
cost_reg = n_lambda * (tf.nn.l2_loss(weights['hidden']) + tf.nn.l2_loss(weights['out'])) cost = cost_J + cost_reg + cost_sparse optimizer = tf.train.Optimizer(learning_rate=n_learning_rate).minimize(cost) # Initializing the variables init = tf.initialize_all_variables() # Launch the graph with tf.Session() as sess: sess.run(init) saver = tf.train.Saver() outWeights = sess.run(weights['hidden']) vis.display_network(outWeights, filename='pretrainweights') # Training cycle for epoch in range(n_num_epochs): batch_xs=gid.getPatches(n_epoch_size) # Fit training using batch data sess.run(optimizer, feed_dict={x: batch_xs}) print("Optimization Finished!") saver.save(sess, 'my-SAE') outWeights = sess.run(weights['hidden']) vis.display_network(outWeights) vis.display_network(outWeights.T, filename="weightsT")