User Recessive

6 votes

2 answers

2k views

Does using the softmax function in Q learning not defeat the purpose of Q learning?

reinforcement-learning q-learning

asked Dec 7, 2019 at 9:40

6 votes

1 answer

2k views

Why is the learning rate generally beneath 1?

asked Sep 25, 2020 at 3:40

5 votes

2 answers

807 views

Why is gradient descent used over the conjugate gradient method?

machine-learning optimization gradient-descent

asked Nov 17, 2021 at 1:11

3 votes

0 answers

88 views

Why do regression LSTMs learn high to low inputs significantly better than low to high?

neural-networks long-short-term-memory

asked Nov 26, 2019 at 7:05

3 votes

1 answer

117 views

In unsupervised learning, what is meant by "finding the probability of an image"?

unsupervised-learning

asked Dec 6, 2019 at 10:57

3 votes

1 answer

260 views

How are exploding numbers in a forward pass of a CNN combated?

machine-learning convolutional-neural-networks relu

asked Jun 28, 2019 at 2:59

2 votes

0 answers

169 views

How is the bias caused by a max pooling layer overcome?

machine-learning convolutional-neural-networks relu bias-variance-tradeoff

asked Jul 1, 2019 at 4:13

2 votes

1 answer

67 views

How is a neural network where the majority of inputs are 0 trained?

convolutional-neural-networks training backpropagation alexnet

asked Jul 22, 2019 at 3:23

2 votes

1 answer

89 views

Structure discrepancy of an LSTM?

neural-networks backpropagation long-short-term-memory

asked Sep 8, 2019 at 3:20

2 votes

1 answer

95 views

Why does an LSTM cycle on initialisation?

long-short-term-memory

asked Oct 4, 2019 at 2:35

2 votes

1 answer

528 views

How does a decision tree split a continuous feature?

decision-trees feature-engineering

asked Jul 9, 2021 at 6:12

1 vote

1 answer

49 views

Is the following the correct implementation of the Q learning algorithm for a neural network?

neural-networks q-learning

asked Dec 9, 2019 at 6:12

1 vote

2 answers

97 views

What are the standard problems for CNNs and LSTMs?

convolutional-neural-networks long-short-term-memory

asked Oct 14, 2019 at 4:33

1 vote

1 answer

128 views

Is it possible with stochastic gradient descent for the error to increase?

neural-networks gradient-descent objective-functions

asked Jul 10, 2019 at 3:27

1 vote

1 answer

185 views

Dropout causes too much noise for network to train

neural-networks convolutional-neural-networks dropout regularization relu

asked Jul 16, 2019 at 2:58

1 vote

0 answers

23 views

When training a seq2seq model is it better to train using the models outputs or expected outputs?

natural-language-processing recurrent-neural-networks transformer attention seq2seq

asked Apr 21 at 2:04

0 votes

0 answers

37 views

How is it possible to use batches of data from within the same sequence with an LSTM?

machine-learning recurrent-neural-networks long-short-term-memory pytorch batch-learning

asked Jul 20 at 6:24

0 votes

1 answer

45 views

Entirely linear neural network learning non-linear function

neural-networks machine-learning regression

asked Jul 27 at 4:08

0 votes

0 answers

124 views

Is attention always better then an RNN/CNN?

machine-learning natural-language-processing recurrent-neural-networks attention

asked Apr 14 at 7:05

0 votes

1 answer

225 views

How does an LSTM output the correct dimensions for classes?

long-short-term-memory

asked Aug 31, 2019 at 2:17

0 votes

2 answers

231 views

Possible model to use to find pixel locations of objects

convolutional-neural-networks object-detection regression

asked Feb 3, 2020 at 6:46

0 votes

1 answer

56 views

Positive bias causes the calculation of incorrect gradients

convolutional-neural-networks backpropagation

asked Feb 11, 2020 at 5:53

Stack Exchange Network

current community

your communities

more stack exchange communities

22 Questions

Does using the softmax function in Q learning not defeat the purpose of Q learning?

Why is the learning rate generally beneath 1?

Why is gradient descent used over the conjugate gradient method?

Why do regression LSTMs learn high to low inputs significantly better than low to high?

In unsupervised learning, what is meant by "finding the probability of an image"?

How are exploding numbers in a forward pass of a CNN combated?

How is the bias caused by a max pooling layer overcome?

How is a neural network where the majority of inputs are 0 trained?

Structure discrepancy of an LSTM?

Why does an LSTM cycle on initialisation?

How does a decision tree split a continuous feature?

Is the following the correct implementation of the Q learning algorithm for a neural network?

What are the standard problems for CNNs and LSTMs?

Is it possible with stochastic gradient descent for the error to increase?

Dropout causes too much noise for network to train

When training a seq2seq model is it better to train using the models outputs or expected outputs?

How is it possible to use batches of data from within the same sequence with an LSTM?

Entirely linear neural network learning non-linear function

Is attention always better then an RNN/CNN?

How does an LSTM output the correct dimensions for classes?

Possible model to use to find pixel locations of objects

Positive bias causes the calculation of incorrect gradients