what is the keep_prob: 0.8 meaning?

Admin 1
Viewed 289

What is the Keep_prob:0.8

is that correct in Active Learning?

train_summary, _ = sess.run([merged, correct], feed_dict={x: batch_xs, y_: batch_ys, keep_prob: 0.8})#train accuracy 
train_writer.add_summary(train_summary, batch_idx) 
#print("step %s, training accuracy %s" % (batch_idx, train_accuracy))

please explain keep prob for train and test sets

for active learning suitable value of Test keep_prob? 

test_summary, _ = sess.run([merged, correct], feed_dict={x: test_xs, y_: test_ys, keep_prob: 0.8}) #test_accuracy
test_writer.add_summary(test_summary, batch_idx)
#print("test accuracy %g" % test_accuracy)
conv-neural-network deep-learning python
1 Answers

The keep_prob value is used to control/adjust the dropout rate used when training the neural network. Essentially, it means that each connection between layers will only be used with probability 0.8 when training. This helps in reduces overfitting.

The keep_prob value is fed in via a placeholder so that the same graph can be used for training (with keep_prob = 0.8).Also, while evaluating make sure you don't use keep_prob as we don't want to switch off nodes randomly while testing/evaluating.

user_43939700
Related