# Tag Info

12

3D CNN's are used when you want to extract features in 3 Dimensions or establish a relationship between 3 dimensions. Essentially its the same as 2D convolutions but the kernel movement is now 3-Dimensional causing a better capture of dependencies within the 3 dimensions and a difference in output dimensions post convolution. The kernel on convolution ...

5

3D convolutions should when you want to extract spatial features from your input on three dimensions. For Computer Vision, they are typically used on volumetric images, which are 3D. Some examples are classifying 3D rendered images and medical image segmentation

4

NOTE: All the observations and results are from the paper The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks. To answer your questions one by one: Yes there are ways to determine which filters have more impact on the output. Its a very naive way but works very good in practice. Filters with small weights impact output less (according ...

4

What are the parameters in a convolutional layer? The (learnable) parameters of a convolutional layer are the elements of the kernels (or filters) and biases (if you decide to have them). There are 1d, 2d and 3d convolutions. The most common are 2d convolutions, which are the ones people usually refer to, so I will mainly focus on this case. 2d ...

3

Short answer Theoretically, convolutional neural networks (CNNs) can either perform the cross-correlation or convolution: it does not really matter whether they perform the cross-correlation or convolution because the kernels are learnable, so they can adapt to the cross-correlation or convolution given the data, although, in the typical diagrams, CNNs are ...

3

I don't think that to understand convolution you need to dig into the nested code of huge libraries, since the code becomes quickly really hard to understand and convoluted (ba dum tsss!). Joking apart, in PyTorch Conv2d is a layer that applies another low level function, conv2d, written in c++. Luckily enough, the guys from PyTorch wrote the general idea ...

3

For a standard convolution layer, the weight matrix will have a shape of (out_channels, in_channels, kernel_sizes*) in addition you will need a vector of shape [out_channels] for biases. For your specific case, 2d, your weight matrix will have a shape of (out_channels, in_channels, kernel_size[0], kernel_size[1]). Now if we plugin the numbers: out_channels =...

3

Am I right in thinking that because there are only newImageX * newImageY patterns in the 32 x 32 image, that the maximum amount of filters should be newImageX * newImageY, and any more would be redundant? Your assumption is wrong. If you have a $32 \times 32$ images (so consider only grayscale images), then you have $256^{32 \times 32}$ possible patterns (i....

3

I haven't seen it as you describe and I don't think it would be much useful. Pooling layers are being gradually phased out of networks, because they don't seem to be that useful anymore. With the emergence of more and more conv-only architectures, I don't see that likely.

3

I have had similar thoughts about neural networks before. Convolution layers are layers of two dimensional nodes effectively passing the spacial data so why don't we use two dimensional hidden layers to receive information out of them. I'm sure someone has used this type of implementation before. I believe the papers bellow are using this. Part of the ...

3

For a 3 channel image (RGB), each filter in a convolutional layer computes a feature map which is essentially a single channel image. Typically, 2D convolutional filters are used for multichannel images. This can be a single filter applied to each layer or a seperate filter per layer. These filters are looking for features which are independent of the color,...

3

You are partially correct. On CNNs the output shape per layer is defined by the amount of filters used, and the application of the filters (dilation, stride, padding, etc.). CNNs shapes In your example, your input is 30 x 30 x 3. Assuming stride of 1, no padding, and no dilation on the filter, you will get a spatial shape equal to your input, that is ...

2

About the images inside the CNN layers: I really recommend this article since there is no one short answer to this question and it probably will be better to experiment with it. About the RGB input images: When needed to train on RGB pictures it is not advised to split the RGB channels, you can think of it by trying to identify a fictional cat with red ears,...

2

Short answer is no. You can't use a model trained for one task to predict on a totally different task. Even if the second task was another image classification task, the CNN would have to be fine tuned for the new data to work. A couple of things to note... 1) CNNs are good for images due to their nature. It isn't necessary that they'd be good for any 2-...

2

In most modern neural network frameworks, the update rules for training can be selectively applied to some parameters and not others. How to do that is dependent on the framework. Some will have the concept of "freezing" a layer, preventing parameters in it being updated. Keras does this for example. Others will do the opposite and expect you to provide a ...

2

I'd suggest you better understand edge detectors such as Robert or Sobel operators first to understand better how convolution operation on images extract features by constant value kernels. Would personally recommend Gonzales and Woods for this, as it gives a pure mathematical explanation to how and why these features are extracted. Essentially the ...

2

Yes this looks a lot like overfitting. The clue is in the low and slowly decreasing training loss compared to the large increases in validation loss. One simple fix would be to stop training around epoch 50, taking the best cross validation result to select the most general network at that point. However, anything that works to improve stable generalisation ...

2

You can also think of a convolutional neural network (CNN) as an encoder, i.e. a neural network that learns a smaller representation of the input, which then acts as the feature vector (input) to a fully connected network (or another neural network). In fact, there are CNNs that can be thought of as auto-encoders (i.e. an encoder followed by a decoder): for ...

2

Mathematically, the convolution is an operation that takes two functions, $f$ and $g$, and produces a third function, $h$. Concisely, we can denote the convolution operation as follows $$f \circledast g = h$$ In the context of computer vision and, in particular, image processing, the convolution is widely used to apply a so-called kernel (aka filter) to an ...

1

Yes, it is applied element-wise on every single value of the feature map. Assuming ReLU as your nonlinearity function, all negative values of the image feature map are set to zero, and the rest of the elements stay unchanged.

1

To offer a bit of theory, CNNs work well for many image tasks because they process spacially local information, without much care for absolute position. Essentially, every layer chops every image up into tiny crop images, and do an analysis step on the crops. The simple questions of "is this a line... corner... eye... face?" can be asked equally of every ...

1

Read on Fully Convolutional Networks (FCN). There is a lot of papers on the subject, first was "Fully Convolutional Networks for Semantic Segmentation" by Long. The idea is quite close to what you describe - preserve spatial locality in the layers. In FCN there is no fully connected layer. Instead there is average pooling on top of last low-resolution/high-...

1

The pooling operation in a CNN is applied independently to each layer and the resulting feature maps are disjoint. This is the very reason that in most schematics depicting a certain CNN architecture, we obtain three output maps from an input image (corresponding to the convolutions and pooling operations performed on the RGB channels separately). Each ...

1

The pooling operation is applied to the output of the convolution layer. More precisely, it is applied separately for each of the input channels (or slices). So, if the pooling layer receives an input volume of $H_i \times W_i \times D$, then it will produce an output volume $H_o \times W_o \times D$, so the depth of the output volume is equal to the depth ...

1

Does the next convolutional filter have a depth of 40? So the filter dimensions would be 3x3x40? Yes. The depth of the next layer $l$ (which corresponds to the number of feature maps) will be 40. If you apply $8$ kernels with a $3\times 3$ window to $l$, then the number of features maps (or the depth) of layer $l+1$ will be $8$. Each of these $8$ kernels ...

1

As a rule of thumb for image data augmentation, look at the augmented images: Can you correctly classify or measure your target label from the augmented images? Could something similar to the augmented images appear in the environment where you want to run inferences on previously unseen inputs? For your suggested augmentation of shuffling the channels, it ...

1

Yes you can vectorize a CNN. See this github file for details: https://github.com/parasdahal/deepnet/blob/master/deepnet/layers.py After looking through it it basically transposes the input to some dimension and apply matrix multiplication to the weight with some other kind of transfromation. Pls refer to the github repository for details. Hope this can ...

1

I'm going to post another guess to this question - it won't be a complete answer, but hopefully it'll provide some direction towards finding a more legitimate answer. The feed-forward networks as suggested by Vaswani are very reminiscent of the sparse autoencoders. Where the input / output dimensions are much greater than the hidden input dimension. If you ...

1

1) The math is the exact same, so from an optimization or mathematical perspective there is no difference 2) Here are my guesses to a possible answer. Habit: People may just call one over the other out of habit Generality: Across frameworks a 1d convolution op would work, while Dense of FC may need adjustments to work on the temporal axis Parallel ...

1

If you have a $h_i \times w_i \times d_i$ input, where $h_i, w_i$ and $d_i$ respectively refer to the height, width and depth of the input, then we usually apply $m$ $h_k \times w_k \times d_i$ kernels (or filters) to this input (with the appropriate stride and padding), where $m$ is usually a hyper-parameter. So, after the application of $m$ kernels, you ...

Only top voted, non community-wiki answers of a minimum length are eligible