Questions tagged [recurrent-neural-network]

A recurrent neural network (RNN) is a class of artificial neural network where connections between units form a directed cycle.

0
votes
0answers
17 views

Standardization of time-series data with scikit's StandardScaler and MinMaxScaler

Question about the right way of using StandardScaler on time-series data. I've got a time-series data that I want to put into RNN of shape (samples, time-step, features), I'm splitting the data into ...
1
vote
2answers
26 views

Keras LSTM problem, how set up correctly a neural network for time series?

i'm trying to undestand how lstm works for predict time series with Keras. Here's my example. I use an accelerometer and i have a 128.000 time series. I thought to take: n_steps_in = 10.000 ...
2
votes
0answers
20 views

How to solve the value errors in rNN?

When I did rNN, I just got: ValueError: Error when checking input: expected lstm_2_input to have 3 dimensions, but got array with shape (99, 20) scaler = MinMaxScaler(feature_range=(0, 1)) data = ...
0
votes
0answers
8 views

how to restore training checkpoints and speak with the bot

i have done the training for my text classification model for a chatbot and now i want to test this model and chat with it , what should i do ? i have an text classfication model for a chatbot and ...
0
votes
0answers
24 views

Recurrent Neural Network - Time Series prediction: On the number of output neurons and sequence length

Context I am currently working with Gated Recurrent Units for time series prediction. Anyhow, my question should apply to all kinds of such recurrent networks. In my data, I have timestamps of 15 ...
0
votes
0answers
9 views

Multi-dimensional softmax implementation in keras/tensorflow

I have a hard time understanding how multi-dimensional softmax is implemented in keras. I want to use this for predicting the next words of a sentence. y1 y2 y3 _|______|______|...
0
votes
0answers
22 views

Load and Use Keras model built using LSTM network in C++

I have created LSTM network using Keras for next word prediction based on the context of the previous words in a sentence. I have written the code in Python, but have to deploy it with existing code ...
1
vote
0answers
34 views
+50

Tensorflow Partial Run on Golang (RNN states)

I have a GRU RNN text generation model that I imported as protobuf in Golang. model, err := tf.LoadSavedModel("poetryModel", []string{"goTag"}, nil) Similar to the code from this Tensorflow tutorial,...
0
votes
0answers
22 views

Receptive Field Size

A Convolutional Neural Network (CNN) has 3 consecutive 5×5 convolutional layers with a stride of 1 without pooling. What is the size of the receptive field for a neuron in the 3rd convolutional layer?...
0
votes
0answers
8 views

Should lowering dimension GLoVE vectors and switching from bi-GRU to uni-GRU reduce overfitting

Suppose I have a text classification problem with a small training set of around 2000 sentences as the training examples, and around possible 130 classes. My model consist of : 100d GloVe embedding ...
0
votes
0answers
17 views

How can achieve LSTM/RNN history-based prediction by using Keras backend?

For my experiment, I have a formatted csv file with 1440 columns like following: timestamps[row_index] | feature1 | feature2 | ... | feature1439 | feature1440 | --------------------------------------...
-1
votes
0answers
13 views

RNN encoder-decoder model keeps overfitting

i've trainning a machine translation model (from english to vietnamese) with RNN, LSTM with 25000 example pairs (for training set -> 20000, test set -> 5000) the model i used like below but val_acc ...
0
votes
0answers
17 views

ValueError: An `initial_state` was passed that is not compatible with `cell.state_size`

i m trying to add a custom layer to my model but i m having this illogic error ,what do i need to change ? if the code of the custom layer is needed let me know it will mean the world to me if you ...
-2
votes
0answers
15 views

What is state-of-art for text generation as of 2019-05-17?

I've being toying around with some book data and I kind of wanted to create a text generator based on book characteristics to generate new text that is somewhat relevant. To my knowledge seq-to-seq ...
0
votes
0answers
20 views

What is difference between torch.nn.GRU vs torch.nn.GRUCell [duplicate]

What is difference between torch.nn.GRU vs torch.nn.GRUCell ? For example why is here nn.GRU converted to nn.GRUCell? https://github.com/erogol/WaveRNN/blob/master/models/wavernn.py#L108 https://...
2
votes
2answers
38 views

How to reproduce RNN results on several runs?

I call same model on same input twice in a row and I don't get the same result, this model have nn.GRU layers so I suspect that it have some internal state that should be release before second run? ...
-5
votes
0answers
51 views

I really do not know how to implement this neural network [closed]

I am asking if someone can help me in implementing this type of neural network with this output feedback showed in figure. Theoretically it is a special case of a recurrent neural network generally ...
0
votes
0answers
20 views

Training RNN for abstractrive text sumamrization with sparse y_train data

I am working on an abstractive solution for a legal text summarization problem where I have all the training samples I need, but nothing to use as my "label" (y_train/y_test) tensor. My idea so far ...
0
votes
0answers
19 views

Which neural network models can learn local correlation between features better?

Following this question and this paper , let's say I have time-series data on 3 physical parameters which are in 3 matrices form R,G,B and all are in the same size like N×K and I combined them and ...
0
votes
0answers
12 views

how to use echo state network to simulate its normal behavior?

I have some data collected from bridge sensors. My project is about using esn to simulate the bridge's normal behavior with the sensor data hence making the esn capable to identify its health problem. ...
0
votes
0answers
10 views

How to find the inverse of a nonlinear function with lstm

I have a system based on a combination of two other functions y=f(g(x(t)). f(x) is nonlinear and unknown with no clear closed form.However i can learn and characterize it with an lstm. I want to ...
1
vote
1answer
23 views

Setting initial state in dynamic RNN

Based on the link: https://www.tensorflow.org/api_docs/python/tf/nn/dynamic_rnn In the example, it is shown that the "initial state" is defined in the first example and not in the second example. ...
0
votes
1answer
22 views

How do I properly train and predict value like biomass using GRU RNN?

My first time trying to train a dataset containing 8 variables in a time-series of 20 years or so using GRU RNN. The biomass value is what I'm trying to predict based on the other variables. I'm ...
0
votes
1answer
41 views

(deep learning, rnn , cnn) image image captinoing with keras

I have made image captioning tutorial, but it doesn't work. help me... image captioning is a model that explained the image a person inputs. I don't have GPU, so I have to make the same model in ...
0
votes
1answer
33 views

Passing a numpy array into placeholder tensorflow

So i have an array of question id: ques = [2 1 5 2 1 3] which is of shape (6,) and I'm using this array to be passed into an input placeholder where: class RNN: def initialize(self): ...
0
votes
0answers
17 views

How can reshape data correctly to predict data after certain timestamp base on learning from previous timestamps data in Keras?

I am trying to use 2-dimensional CNN for Prediction in Keras. I have a network which continuously performs an action and my goal is to learn till certain timestamp t=30 and predict the future ...
2
votes
0answers
21 views

Implementation of multi-variable, multi-type RNN in Python

I have a dataset which has items with the following layout/schema: { words: "Hi! How are you? My name is Helennastica", ratio: 0.32, importantNum: 382, wordArray: ["dog", "cat", "...
0
votes
0answers
9 views

How Rouge Metric tells that summary is generated using abstraction technique (just like human do)

ROUGE score is a reflection of how close the model generated summary is to the human-generated summary (abstractive way) which is basically a Recall based approach. How does it tell that generated ...
-1
votes
1answer
19 views

why LSTM is performing poor than Simple Feed-forward Neural Network for Artificial time series data

I have an artificial data set. For 2 class classification problem I have 2000 data instances with 500-time samples. All the data instances are vectors of zeros (500 zeros ) For the first 1000 data ...
0
votes
0answers
18 views

How can split data for cross validation by Keras for RNN and CNN after certain time-step?

Let's say I have time-series data on 3 physical parameters which are in 3 matrices form R,G,B and all are in the same size like N×K and I combined them and convert them into big matrice PxM by reshape ...
0
votes
0answers
16 views

how to apply the autoencoder and decoders for text classificaation using LSTM

I am a beginner to CNN,RNN models, I am trying classifying the twitter data using autoencoders, BUt I got some errors on that, please help me how to solve it. Here my input shapes: from keras....
1
vote
0answers
22 views

Several questions to Stanford lexicalized, unlexicalized and RNN parser

I am writing my Master thesis about parsing german sentences with the Stanford and other parsers. The Stanford Parser is a great tool and I got good results, so thanks to all people involved! But ...
0
votes
1answer
31 views

Understanding embedding vectors dimension

In deep learning, in particularly NLP, words are transformed into a vector representation to be fed into a neural network such as an RNN. By referring to the link: http://colah.github.io/posts/2014-...
-2
votes
0answers
28 views

LSTM - how to embed sparse data in this example?

I am trying to understand an LSTM similar to what's done in a paper about embedding medical concepts in an embedding layer before inputting to an LSTM. A screenshot of the relevant figure is below: ...
1
vote
1answer
39 views

How to manually obtain the same output as model.predict() in keras

I am trying to reproduce, via Numpy, the output that I would obtain using Keras' model.predict(). My keras model layers are the following: ...
0
votes
1answer
29 views

Recurrent neural network hidden cells

So i understand that RNN has hidden units within each RNN cell, as shown in the visualization of the second answer in the link: What is num_units in tensorflow BasicLSTMCell? However, I've been ...
0
votes
0answers
21 views

Supervised automatic xml-generation from text/html code via neural nets

I need to create a neural net that will convert a text file (e.g. text of some play) into tei-format of xml. For example, to convert Аксенов Здорово ль, кум? Лыткин Здорово, Петр Аксеныч. ...
0
votes
0answers
17 views

How to use maskingLayer in model()

I have this tensorflow graph that's like A,B -> concatenate -> RNN -> timeDistribute-> dense -> softmax (A little redundant at the end) I have two more things to introduce, 1) not all time steps are ...
1
vote
0answers
48 views

Runtime error (backprop) while training a time series model using RNN in PyTorch

Following is the error I get while training an RNN: RuntimeError: Trying to backward through the graph a second time, but the buffers have already been freed. Specify retain_graph=True when ...
0
votes
0answers
13 views

Feed 3D data for a RNN

I'm trying to implement a RNN neural network, using 3D skeleton data. I'm trying to implement the exact code mentioned in https://github.com/Sunnydreamrain/IndRNN_pytorch In its code, we were asked ...
1
vote
1answer
14 views

Tensorflow.js adding simpleRNN to model()

I want to build a semi-complex neural network, so I'm not using tf.seqential(). const model = tf.model( { inputs: [tickInput,boardInput], outputs:moveChoices, } ); which has to be created ...
0
votes
0answers
12 views

How to encode time information in LSTM prediction problem

I am trying to build a model based on a recent google paper called "Scalable and Accurate Deep Learning for Electronic Health Records". The paper describes how to embed medical data for downstream ...
0
votes
1answer
22 views

Classification sequence 4D array data with RNN-LSTM

I am currently working on a project where i am trying to create a machine learning model that is able to classify actions in a video. I already created a script that is able to detect a person in a ...
1
vote
1answer
41 views

Understanding the backward mechanism of LSTMCell in Pytorch

I want to hook into the backward pass of a LSTMCell function in pytorch so in the initialization pass I do the following (num_layers=4, hidden_size=1000, input_size=1000): self.layers = nn.ModuleList(...
0
votes
0answers
22 views

How to input discrete categorical variables into RNN?

I'm trying to write an RNN model to do learn2learn. Basically, if I have a scalar valued function f(x,y,z), where x is some element of set S1, y is some element of set S2, and z is some element of set ...
1
vote
1answer
36 views

upgrade code rnn.static_bidirectional_rnn to fit with tensorflow 2.0 API

import tensorflow as tf from tf.contrib import rnn lstm_f = rnn.BasicLSTMCell(n_hidden, forget_bias=1.0) lstm_b = rnn.BasicLSTMCell(n_hidden, forget_bias=1.0) blstm_out, state_f, state_b = rnn....
1
vote
1answer
54 views

How can a Neural Network learn from testing outputs against external conditions which it can not directly control

In order to simplify the question and hopefully the answer I will provide a somewhat simplified version of what I am trying to do. Setting up fixed conditions: Max Oxygen volume permitted in room =...
2
votes
2answers
75 views

pytorch embedding index out of range

I'm following this tutorial here https://cs230-stanford.github.io/pytorch-nlp.html. In there a neural model is created, using nn.Module, with an embedding layer, which is initialized here self....
0
votes
0answers
15 views

Can an RNN perform well with dataset of many but very short series?

I know the best way to know this is to just test it but I am trying to gain an intuition: Let's say we have 100,000 examples of a very short series of say 20 where there are a few features of input ...
-1
votes
0answers
10 views

How to create a LSTM network with a many to many relation with multiple classes

Currently we are creating a LSTM network for a project. However we have data that is divided in different classes. For instance lets say we have a time series data where we have two stocks (Microsoft ...

http://mssss.yulina-kosm.ru