Questions tagged [tensorflow]
A Python library for deep learning developed by Google. Use this tag for any on-topic question that (a) involves tensorflow either as a critical part of the question or expected answer, & (b) is not just about how to use tensorflow.
533
questions
0
votes
0answers
7 views
Obtaining Model Gradients in TensorFlow.JS [closed]
I'm familiar with getting gradients of a model in Python TensorFlow, i.e. via
gradients_of_model = tf.gradients(loss_function, model.trainable_variables)
with ...
0
votes
0answers
19 views
Issues with Tensorflow and CUDA Causing Error [closed]
I'm working on a project that finds an object on an image using
Mask RCNN. Although the algorithm was made for Tensorflow-gpu 1.13 through ...
1
vote
0answers
28 views
Did I go wrong somewhere with my ANN?
Data for this is available here: https://tmpfiles.org/download/51656/Data.csv
This is my ANN model:
...
0
votes
0answers
4 views
Validation Error less than training error? custom metrics affected from droputs?
I have a neuronal network trained on some data. My testing loss is less than my training loss.
As this question is well answered regarding some points here I ask myself, if a custom metric that is ...
1
vote
1answer
16 views
Conceptual question on Image semantic/ instance segmentation networks
I am trying to understand better how the image semantic/ instance segmentation work.
I understand going from the concept of the perceptron that Deep Neural Networks have one or both of the following:
...
0
votes
0answers
13 views
Validation loss diverging when training a simple CNN for text classification
I'm training a CNN for text classification on the IMDb movie reviews dataset. The dataset contains 25000 training and 25000 testing samples of movie reviews, each half positive and half negative. The ...
1
vote
0answers
10 views
Tfp.sts unexpected wights prior behavior
I am trying to build a time series structural model and get the coefficients of the external factor.
here is the structure of the model
...
1
vote
1answer
34 views
How to understand network neural network architecture from a research paper
Hello everyone I have the following architecture from the DELP-DAR research paper (https://www.sciencedirect.com/science/article/pii/S0167865519303216)
and I dont really understand two things, first ...
1
vote
1answer
24 views
Validation and Test accuracy at random performance, whereas Train accuracy very high
I am trying to build a classifier in TensorFlow2.1 for CIFAR10 using ResNet50 pre-trained over imagenet from keras.application and then stacking a small FNN on top of it:
...
1
vote
1answer
25 views
Combine ReLU with TanH is a good idea?
I have a CNN implementation for the Generator of a GAN, internally, the architecture is using ReLU for non-linearities, but at the output, the paper of the architecture specifies Tanh must be used.
...
3
votes
1answer
60 views
Variational autoencoder with L2-regularisation?
I have built a variational autoencoder (VAE) with Keras in Tenforflow 2.0, based on the following model from Seo et al. (link to paper here). The VAE is used for image reconstruction.
Note that the ...
0
votes
0answers
17 views
Should the lambda for L1 norm regularizer inversely be proportional to the number of trainable weights?
Say I want to implement Conv2D in keras and for each Conv2D layer, if I apply 20 filters of [2,3] filter on an input with depth of 10, then there will be 20*(2*3*10+1) = 1220 trainable weights.
the ...
2
votes
1answer
42 views
Neural Network Loss Function for Predicted Probability
I am building a neural network (using tensorflow/keras) that attempts to classify into one of 5 categories (0, 1, 2, 3, 4). I have been using sparse categorical cross-entropy as my loss function. This ...
0
votes
0answers
10 views
How to improve neural network training against a large data set of points with varying magnitude
I am currently using TensorFlow and have simply been trying to train a neural network directly against a large continuous data set, e.g. $y = [0.014, 1.545, 10.232, 0.948, ...]$. The loss function in ...
0
votes
1answer
18 views
Finding patterns in binary files using deep learning
I am a newbie in deep learning and wanted to know if the problem I have at hand is a suitable fit for deep learning algorithms. I have thousands of fragments each of about 1000 bytes size (i.e. ...
0
votes
2answers
38 views
Manipulate keras multiple loss
Lets assume that we have a model model_A and we want to build up a backpropagation based on 3 different loss functions. The first loss (...
1
vote
0answers
10 views
Is Determinism important for Hyperparameter Tuning?
When training the Model on GPU, different results are retrieved for the same hyperparameters. This effect can be shut down by using CPU or Tensorflow 2.1. with deterministic settings.
The Post on ...
1
vote
0answers
6 views
How to build a sentence classifier with tensorflow, that has two sub bi-lstms one for sentence embedding and other for sentence classification?
I want to build a model that takes a document, creates sentence embedding for each sentence using a bi-LSTM, then use the sequence of sentences embedding as input to another bi-LSTM that outputs a ...
0
votes
0answers
23 views
How do I apply Min Max scaling for numerical forecast when both dependent and independent volumes are increasing over time?
I'm want to build a numerical regression to forecast.
From my initial analysis, it shows linear models (glm) out performs the typical decision tree models (xgboost, ranger...etc). I hypothesized that ...
1
vote
1answer
40 views
Masked Autoencoder MADE implementation in TensorFlow vs Pytorch
I am following the course CS294-158 [1] and got stuck with the first exercise that requests to implement the MADE paper (see here [2]). My implementation in TensorFlow [3] achieves results that are ...
1
vote
0answers
24 views
Keras Neural Network: use loss from one output in the loss of the other output [closed]
I would like to use the loss from one of my NN auxiliary outputs as part of the loss for the other output.
$L($total$) = L($main output$) - \lambda L($auxiliary output$)$
I am unsure how to access ...
0
votes
0answers
6 views
Where the embeddings should be implemented in the RNN model?
Hi All (it's my first question here so welcome everyone),
I wrote simple RNN model in tensor flow and I cannot figure out where the embeddings should be inserted inside, please find my code and below ...
0
votes
1answer
22 views
Building a neural network with two training paths in Keras
I am trying to build a NN in Keras with two different output paths where the first path informs the second. The first path passes its loss to the end of the second path, like so:
Pass through layer A ...
0
votes
0answers
15 views
3D image segmentation: 2D slice-wise vs full 3D model?
I need to segment a volume of $N\times N \times N$ pixels. I can do it in two ways, using a fully 3D convolutional neural network (e.g.: Conv3D in Keras), or I can segment $N$ 2D slices (e.g. Conv2D) ...
2
votes
1answer
21 views
Which is the error of a value corresponding to the maximum of a function?
This is my problem: I use data observed with MUSE (which is an astronomical instrument provides cubes, i.e. an image for each wavelength with a certain range, link
) to extract a measure of redshift. ...
0
votes
0answers
10 views
Handling missing target values in vector regression problem with Keras
I'm doing a vector regression problem and many of the target vectors have a few components which are missing (so these components are recorded as nans in the ...
0
votes
0answers
12 views
Why would we use different output and recurrent activations in an LSTM? [duplicate]
I am developing a LSTM network with Tensorflow. The Tensorflow LSTM documentation shows that there are different activations for the output of the timestep (tanh) and for the data flowing back in to ...
2
votes
0answers
39 views
Interpretation of Tensor Flow CNN results with big dips in accuracy while training
I am trying to classify images using a CNN in tensor flow. I am doing 10 fold cross validation. At each fold, the training set is 900+ images and the validation set is 100 images. It is only two ...
0
votes
0answers
15 views
Use transfer learning with subclassing api of tensorflow keras
I am able to use transfer learning with sequential and functional api but I want to use ResNet50 from tf.keras.applications with subclassing model api of tf.keras. Is this possible? If yes, Can anyone ...
0
votes
0answers
5 views
One step prediction of time series using LSTM
I want to predict stock prices using LSTM. I have successfully trained my model and have it saved. Now that I've loaded it back in, how would I use model.predict() ...
0
votes
0answers
24 views
WGAN-GP stability loss
I am training a Conditional WaveGAN (1D DCGAN for audio) using WGAN-GP whose generator is of an auotencoder architecture. The network is trained to take an audio input, compress it, then decompress it ...
0
votes
0answers
19 views
Gradient Boosting and neural networks
Is there any Python package that implements a boosted neural network ? Any pointer is appreciated.
A sample reference about the boosting and NN can be this one.
1
vote
1answer
33 views
Low memory time series input for deep learning
Background
I have some data that looks like this:
...
1
vote
2answers
32 views
Should I normalize all data prior feeding the tensorflow models?
Appreciate your wisdom on this,
My understanding is most of the tutorials recommend normalizing / scaling the data prior feeding the tensorflow models. Doesn't normalization require that data ...
0
votes
0answers
17 views
Autodiff implementation for policy gradient methods (reinforcement learning)
Maybe this is better posted on stackoverflow, but I thought I would start here . I am trying to understand how RL policy gradient style algorithms are implemented in autodiff frameworks like ...
0
votes
0answers
34 views
tensorflow high validation acuraccy but bad predictions
I asked this over at stackoverflow but this seems to be the specific place to ask machine learning questions so hopefully I can get more answers. Anyways I am trying to make a multiclassed multilabel ...
0
votes
0answers
21 views
Keras model with Google BERT -> very low accuracy [duplicate]
I'm attempting to fine-tune Google BERT to be able to classify some text to a single integer label (multiclass classification). I have the model up and running, however the predicted labels are all ...
0
votes
1answer
56 views
Difference between batch_size=1 and SGD optimisers in Keras
I have a question that to implement stochastic gradient descent we set the batch size=1 what ever be the optimizer.So what does sgd optimizer do that is different then setting batch size=1 using any ...
0
votes
0answers
10 views
LSTM for non-timeseries data - Input Shape
I am dealing with a non-time series problem. I have a total of 20 features. 19 for input and 1 for output I have my input as follows:
...
0
votes
0answers
10 views
Regression with CNN Loss won't pass certain values
I'm working on a regression problem which since my dataset is images i prefered to use widely used CNNs.In general,i have to predict a single value between 0 and 500 for a 100x100 grid based on the ...
0
votes
1answer
27 views
Tensorflow how to make a training data
I'm a begginer in programming I beg your pardon for that question, I just want to learn.
The function
tf.keras.datasets.mnist.load_data()
returns
...
0
votes
0answers
9 views
Tacking out of vocab words in neural machine translation with fasttext
I am trying to tackle out of vocab words in my machine translation model build using Tensorflow's official implementation of NMT . I am very much confused how to do it since I am new to NLP. I tried ...
0
votes
0answers
42 views
CNN Mask - IoU Versus Keras “Accuracy”
I am using Keras/Tensorflow to do an FPN CNN model for basic binary mask detection, and have the trainer set to track the metrics of 'IoU' and and the built-in default Keras 'accuracy' metric. IoU is ...
0
votes
0answers
30 views
Is tensorflow LKJ distribution correct?
I am trying to verify that this distribution is propoprtional to a power of the determinant of the matrix:
https://www.tensorflow.org/probability/api_docs/python/tfp/distributions/CholeskyLKJ
Which ...
1
vote
1answer
16 views
Why does padding affect my results?
I'm learning deep learning, and have been looking at a tutorial for autoencoders using MNIST. It's quite straightforward, and I think I understand it quite well so far. This is the code from the ...
0
votes
0answers
7 views
How to train model with custom train/test split ratio?
what i mean train/split ratio is that i've very few images to train model on so is it possible to allocate 100% images to train split and ignore test split entirely?
link to project here https://...
0
votes
0answers
13 views
Predictions all coming out the same
Sorry for this basic question, I'm a total noob.
I have input data in the format of 6000 x 3 x 256, that is 6000 rows with 3 features, which themselves contain 256 ...
1
vote
1answer
24 views
What is the actual use of GAN model? Is it only used to generate the data that closely resembles original dataset?
I am very new to tensor flow. I came across GANS. From what I understand in GANs there are 2 models, Generator and Discriminator. Generator job is to generate the data that will be able to fool the ...
0
votes
0answers
7 views
Weird validation model loss with a stable accuracy
I've trained a CNN based model for activity recognition. The model's validation loss is spiking as seen in the image below
What is strange is that the accuracy of my model's validation remains ...
0
votes
1answer
13 views
Slower training after transfer learning
Before I used a model to categorize cars, bikes and bicycles that looked like this:
...
