Stack Exchange Network

Stack Exchange network consists of 174 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers.

Visit Stack Exchange

Questions tagged [cnn]

Convolutional Neural Networks (CNN, also called ConvNets) are a tool used for classification tasks and image recognition. The name giving first step is the extraction of features from the input data.

2
votes
1answer
20 views

Fully connected layer in deep learning

How to determine the best number of the fully connected layers in CNN? Can I use only one fully connected layer in CNN? How to determine the dimension of the fully ...
2
votes
0answers
21 views

How many Layers in the original Yoon Kim CNN implementation?

I saw some implementations of Yoon Kim's Convolutional Neural Network (Paper: http://www.aclweb.org/anthology/D14-1181).... ...in some implementations they put one more Dense(..) Layer before the ...
1
vote
0answers
17 views

Dataset containing spatial and temporal features (built on a CNN model)

A dataset contains spatial and temporal features. It contains the time series data (2 min intervals) of the sections of a map. It is 320*480 (320 map sections and 480-time intervals). Each row ...
4
votes
1answer
17 views

Training network with variable frame rate?

I would like to train a temporal network, but the video data available are in different frame rates(ex 7,12,15,30). How should I train this network, without down-sampling higher frame rate videos. I ...
1
vote
1answer
18 views

Conceptual question on CNN and any multi layer neural network (Part 2)

I have read a number of tutorials and online lectures (https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/) but none of them mention the rationale for selecting a particular design. How ...
0
votes
0answers
21 views

Implementing End-to-end deep learning model by NVIDIA

I am trying to implement an end-to-end deep learning model by NVIDIA for autonomous car steering but the behavior of the loss function and and accuracy is getting me confused. I am new to machine ...
0
votes
0answers
17 views

Conceptual questions on CNN (Part 1) [closed]

I am following the webtutorial: http://cs231n.github.io/convolutional-networks/#overview and in general I have not still understood the basic concepts of CNN. In particular the following points are ...
2
votes
1answer
35 views

Time series classification with convolutional neural network

I have to classify a time series signal with a CNN (not a LSTM or some kind of RNN). The input signal is a window with a width of 90 samples (currently, but can be changed) and 59 sensor values. So my ...
2
votes
1answer
15 views

How to determine the number of the training images in Keras after data augmentaion?

I want to create a CNN model and I am using data augmentation. I want know the number of augmented images in Keras. How to determine the number of the training images in Keras after data augmentation?...
7
votes
1answer
36 views

What is the difference between upsampling and bi-linear upsampling in a CNN?

I am trying to understand this paper and am unsure of what bi-linear upsampling is. Can anyone explain this at a high-level? https://arxiv.org/abs/1606.00915
1
vote
1answer
14 views

Spatial Transformer Networks: how is theta differentiable?

In the paper Spatial Transformer Networks, the localization network's output, theta, is differentiable, given the current input feature map. How is this theta ...
4
votes
4answers
149 views

Classification problem with many images per instance

I am working in the following kind of classification problem: I have to classify every instance as class A or class B using many images of the instance. That is, every training example has not one ...
1
vote
0answers
50 views

Preserve colour in convolutional autoencoder

at the moment i work with convolutional autoencoder and now I'am looking for paper or methods that adresses a colour preversation. Most of the AE paper use grayscale images and loss functions such as ...
5
votes
1answer
30 views

Combining spatial input with a label as input for CNN using Keras

I also asked this question on Stack Overflow. However, it has not yet been answered and I think this is a more suitable platform to place it. I'm trying to implement a network set-up similar to this ...
1
vote
1answer
25 views

What is the best way to visualize 10 Fold Cross Validation Scores?

I have trained a CNN model and I have applied 10 Fold Cross Validation because I don't have much data to train the classifier. Now I am unsure about how to visulize fold wise results. Please suggest ...
2
votes
3answers
32 views

What is the intuition behind using 2 consecutive convolutional filters in a Convolutional Neural Network?

I understand the purpose of Convolutional filters (or kernels). I visualize them as learnable feature extractors. E.g. Extract vertical edges or horizontal edges, etc. Could somebody kindly explain ...
1
vote
1answer
24 views

How to save and test CNN model on test set after training

My CNN model is trained on the training set and validated on the validation set, now I want to test it on test set, here is my code: ...
1
vote
1answer
14 views

How to interpret a drastic accuracy loss while training a neuronal net (CNN)?

How can one interpret a drastic accuracy loss after ~38 epochs? Maybe more dropout should be added to the CNN network? (x-axis shows the number of epochs)
1
vote
1answer
20 views

Adding batch normalization layer to VGG16 network

I want to use batch normalization layer to decrease overfitting in VGG16 CNN. Where should the batch normalization layer be added to the network? ...
1
vote
0answers
17 views

Relu forces to classify into only one class

I am training my network on this dataset. Following is configuration of my network: *128 size 1D vector as input *3 Convnets with 128 filters each(with filter sizes 3,4,5 respectively). *RElu *Maxpool ...
1
vote
0answers
32 views

Good lectures (books, articles, blogs) about the art of modeling a CNN network [closed]

I have been using Deep Learning with Tensorflow for a pet project only for educational purpose, and have been able to obtain quite good results with a CNN of my own. However, the process I used to ...
1
vote
1answer
22 views

How is bias added in a convolutional layer [duplicate]

In a typical neural network , bias is usally added like this v = activation ( w1*x1+ ... + Wb*b) However, I am not really sure how it is done in convolutional ...
1
vote
1answer
13 views

Determine Optimal CNN Complexity: Achieve ~100% Training Accuracy, Then Add Regularization?

I was just wondering whether it is a decent idea to first try to make a CNN complex enough to achieve close to 100% accuracy on the training dataset, and then slowly add regularization, augmentation ...
0
votes
0answers
10 views

Understanding The Vertical and Horizontal stack in conditional gated Pixelcnn paper

I found some confusion understanding the importance of vertical and horizontal stacks as a solution to the blind spot problem presented in the original pixel cnn architecture discussed in https://...
1
vote
0answers
12 views

Huge performance discrepancies at each run, with the same CNN architecture

I have set up a CNN architecture with 3 Convolutional layers with pooling (except the last one), a fully connected layer and a logistic regression layer. All the layers, except the last one, have a ...
1
vote
1answer
26 views

adding summary to tensorflow model error

I have a CNN model for classifying lung CT images, the code is written in tensorflow, I added some tensorflow summaries to my code to show graph, scalar, histogram, ... of my tensorflow model in ...
0
votes
0answers
20 views

What metrics should be used for evaluating the performance of model that is taking inputs in the form of patches?

I am trying to carry out classification using CNN for the inputs that are in the form of patches. In my point of view, accuracy and loss cannot be the only metrics by which performance of the model ...
0
votes
2answers
78 views

ValueError in CNN+RNN model in keras

I am trying to build a CNN+RNN model for a computer vision problem. below is my code ...
3
votes
1answer
34 views

Keras: Prediction performance does not match accuracy

I am using Keras/CNN to identify plankton images collected with an in situ camera. When making confusion matrices on the test sets following training I am finding that the accuracy from the ...
1
vote
0answers
28 views

How would code this in NN?

is there any python code already available to do this? - I would appreciate if you can elaborate on coding of this problem as I am new to NN.
3
votes
2answers
51 views

Why convolution over volume sums up across channels?

A simple question about convolution over volume . Say we have an image with dimensions $(n, n, 3)$ and we apply a filter of dimension $(k, k, 3)$ this outputs an matrix of dimension $(n-k+1, n-k+1)$. ...
1
vote
0answers
31 views

Error with respect to kernel weights in convolutional layer

Let's say, I have a convolution layer with: 3 input channels $X$ with dimensions 3x32x32 5 kernel filters $F$ with dimensions 3x7x7 Assuming the stride = 1, the layer produces an output $O$ with ...
1
vote
0answers
14 views

How to implement YOLO in my CNN model?

I build a CNN model using keras on the cat vs dog dataset. Now what I want is with the image classification my model should also locate that animal on that image. I searched on web and found that YOLO ...
2
votes
1answer
17 views

Pre-trained CNN for one-shot learning

I'm currently trying to learn one-shot learning using convolutional neural networks. According to this video, the CNN that I use should have been pre-trained on the MNIST. Why must the CNN be pre-...
1
vote
0answers
23 views

Intuition / Importance of intermediate supervision in deep learning

These days, I have seen many papers using intermediate supervision. Single Network When using a single neural network, multiple neurons output predictions, perhaps by processing data in different ...
2
votes
1answer
26 views

Is it appropriate to use polar images as input to CNNs? Or must they be Cartesian transformed first?

I have built a convolutional neural net which trains on data originally represented in polar space (measurements are a function of angle and distance). My pipeline begins by converting coordinates to ...
1
vote
0answers
12 views

CNN logarithmic path length

Path length between positions can be logarithmic when using dilated convolutions, left-padding for text Could anyone elaborate on the above statement quoted from CNN and RNN comparisons ?
0
votes
1answer
22 views

how to save deep learning model and test it after training?

I have a CNN model written using tensorflow for python, the model is for classifying lung CT images (cancer/no-cancer), after training the model with training and validation data and get a reasonable ...
0
votes
1answer
25 views

How to tinker with CNN architectures?

I was thinking of creating a CNN. Now it is known CNN takes long times to train so it is advisable to stick to known architectures and hyper-parameters. My question is: I want to tinker with the CNN ...
0
votes
1answer
32 views

How to verify hand written signature?

I trying to create a model for determining whether a questioned hand written signature matches known signature samples, and predict if the signature is genuine or forgeries. I'm guessing I'll have to ...
1
vote
1answer
21 views

How to do face recognition without using any kind of Neural Networks like CNN?

Is/was there any way to perform face recognition, instead of using the Convolution Neural Network which uses the technique of mapping(encoding) the face using 128-D vector and then using classifier (...
1
vote
0answers
43 views

Multi-label classification, recall and precision increase but accuracy decrease, why?

I'm training a multi-label classification model using CNN, during training I'm using 90% of my data for training and the left apart 10% for validation, but something strange happens which is the ...
1
vote
0answers
18 views

Matlab: setting static iterations per epoch in a CNN

I'm building a convolutional neural network using Matlab's neural network toolbox. I have code designed to cross-train the network with different data sets, using the previous network's layers in ...
0
votes
0answers
10 views

Performance improvement on images with a consistent background

So I have images of a particular class of object in a background of almost same colour. For example: Say birds flying in a blue background. Now we know the background is almost always constant. So is ...
0
votes
0answers
7 views

Model time conplexity and long term dependencies

In this Transformer-Attention-is-all-you-need blog article , why "The complexity of O(n) for ConvS2S and O(nlogn) for ByteNet makes it harder to learn dependencies on distant positions" ?
5
votes
1answer
44 views

Effect of NOT changing filter weights of CNN during backprop

What is the effect of NOT changing filter weights of a CNN during backpropagation? I changed only the fully connected layer weights while training on the MNIST dataset and still achieved almost 99 ...
1
vote
0answers
38 views

Combining image and scalar inputs into a neural network

I'm looking at the best way of combining CNN with image input and a scalar value. I know that one of the ways is to concatenate flatten layer with this scalar value. But flatten layer consist for ...
1
vote
1answer
9 views

Preventing fitting Regression CNN to the mean when dataset has only few outliers

I am trying to train a CNN for regression on a dataset where most of the points lie around a similar output value. There are however a few outliers that are very important but they are less ...
0
votes
1answer
14 views

why this naming convention for padding as “Same” and “Valid” in keras

I was going through CNN's and found that padding argument should be set to "Valid" if i need no padding and "Same" if i need padding. But, it doesn't make any sense to me. Why can't keras development ...