Questions tagged [lstm]

Long Short Term Memory. A neural network architecture that contains recurrent NN blocks that can remember a value for an arbitrary length of time. A very popular building block for deep NN.

0
votes
0answers
4 views

How does the BPTT work in LSTM? How does the truncate time lag determined?

In LSTM, how does the BPTT transport along time step? Is it the truncate BPTT? If so, how does the network know how to determine the truncate time lag?
0
votes
0answers
11 views

Compute validation loss every n batches with Keras

I am training a recurrent neural network model with a large dataset that takes up to 9 hours/epoch to train. The per batch training loss quickly improves before the end of a single epoch. I would like ...
0
votes
0answers
8 views

How should I structure my input data to a Keras LSTM network for a very long time series?

I am trying to use multivariate time series sensor data to predict when a machine will fail. My problem is there is often millions of time units between failures and I am unsure how to structure the ...
0
votes
0answers
5 views

Does the weight between hidden and hidden layers in LSTM updated during training?

I've seen the paper 'Long Short term memory', Sepp Hochreiter. The thesis says that LSTM has an characteristic called constant error carousel(CEC), which means that the error signal along the time ...
-1
votes
0answers
5 views

Multivariate time series input in Keras lstm timestep

I solving a problem of time series classification , where i have 3 sensor data, each have 3 axis and i have calculated mean, median, standard deviation ,min , max with pandas rolling window. So my ...
0
votes
0answers
12 views

How to use ConvLSTM2D followed by Conv2D in Keras python

I am trying to use the following model in Keras, where ConvLSTM2D output is followed by Conv2D to generate segmentation-like output. Input and output should be time series of the size (2*WINDOW_H+1, ...
0
votes
0answers
14 views

Keras predict_proba predicts the same probabilities for each input (LSTM)

My Keras sequential model produces nearly the same predictions for all inputs, when I use predict_proba.It seems like the model learns nothing, bescause the loss stays nearly the same during ...
0
votes
0answers
5 views

Truncated backpropagation in LSTM, errors in gradient value

Hochreiter in his seminal LSTM paper of 1997 postulates a backpropagation version. When calculating gradient over some network weight he truncates derivatives in the way that only derivatives from ...
0
votes
0answers
28 views

LSTM implementation from scratch in python

I'm currently investigating how lstm networks is implemented from scratch. Right now I have implemented a really simple model in Tensorflow with the following code:with tf.variable_scope(scope or '...
0
votes
1answer
18 views

Keras: Understanding the number of trainable LSTM parameters

I have run a Keras LSTM demo containing the following code (after line 166):m=1model=Sequential()dim_in=mdim_out=mnb_units=10model.add(LSTM(input_shape=(None, dim_in),...
-1
votes
0answers
12 views

ValueError is produced when I try to run my recurrent neural network [duplicate]

I am following this tutorial: RNN w/ LSTM cell example in TensorFlow and Python. I am trying to run the following code in my Jupyter Notebook:import tensorflow as tffrom tensorflow.examples....
1
vote
1answer
22 views

Multivariate and multi-series LSTM

I am trying to create a pollution prediction LSTM. I've seen an example on the web to cater for a Multivariate LSTM to predict the pollution levels for one city (Beijing), but what about more than one ...
0
votes
0answers
33 views

Forecasting new value using LSTM python

I have built an LSTM model that can forecast the future prices. I have tested the same with ground truth value that exists already to know the accuracy of the model. Now I wanted to use the same model ...
0
votes
0answers
29 views

Stacking LSTM layers of different sizes using MultiRNNcell tensor flow

This is the code I have used to stack multiple LSTM cells of sizes 64 and 32 from a list num_hidden=[64,32]stacked_cells=tf.nn.rnn_cell.MultiRNNCell([get_cell(num_hidden[_],keep_prob,z_prob_cells,...
-1
votes
0answers
18 views

LSTM -RNN : How to get continuous range output instead of categorical ?

I am trying to solve a problem where have to predict a value between a range for a sentence :Dataset looks like this:Index_no text_sentence value 01 ...

153050per page
angop.ao, elkhabar.com, noa.al, afghanpaper.com, bbc.com, time.com, cdc.gov, nih.gov, xnxx.com, github.com,