CS231n Winter 2016: Lecture 7: Convolutional Neural Networks
Why Stack Overflow recommends it
Stack Overflow users link it when asking about or explaining how convolution filters, output depth and CNN layers behave, and as a learning resource before implementing a CNN.
Recommended by 2 people across 2 different months, 2016 to 2018.
What it is
Lecture 7 of Stanford's CS231n (Winter 2016), taught by Andrej Karpathy, covering how convolutional neural networks work: filters sliding over images, layer dimensions, pooling and common architectures. It is a recorded university lecture with slides, part of the Convolutional Neural Networks for Visual Recognition course.
What people said
Recommends Norman Bentley · 2016-05 · Stack Overflow · on “How does CNN training take place in this image?”
[LINK] After watching the following great video: [this video] I have got a good understanding on how filters are applied, architecture of different CNN's etc. I have watched a few other videos and read papers, but dense me still has a few questions hoping someone could help me understand this better. We learnt different ways to slide the filter over the image. Now, this process of getting the dot product, is this used for training the CNN or for identification with already trained filters. Or used in both training and identification. In the image above, were those filters trained? I understand white represents activations with the filters, but should the last column indicate anything to me? If th
Recommends vikram meena · 2018-01 · Stack Overflow · on “How to understand the CNN's convolutional layer's output depth is the number of convolution filter used ?”
Output depth is dependent on no of filters in current convolution layer. when you define a filter then you provide spatial dimension (X, Y) and last dimension of filter is based on input channels. Since in convolution each filter (3, 3, z) convolves( in simple words element-vise multiplication) with input (Xin,Yin,z) produces (Xout,Yout,1) dimension output. Each filter produces single layer output that's why depth of output is equal to no of filters in that convolution layer. Note: If you are interested in output size of layer then you can calculate using, This slide explains how convolution works. X_out=(Xin+2Pad-kernal)/Stride [LINK] You may want to watch the video [this video] of the lecture b
Cites Davors72 · 2016-02 · Stack Overflow · on “Convolutional neural network not converging”
I've been watching some videos on deep learning/convolutional neural networks, like here [this video] and here [LINK] , and I tried to implement my own in C++. I tried to keep the input data fairly simple for my first attempt so the idea is to differentiate between a cross and a circle, I have a small data set of around 25 of each (64*64 images), they look like this: [LINK] [LINK] The network itself is five layers: Convolution (5 filters, size 3, stride 1, with a ReLU) MaxPool (size 2) Convolution (1 filter, size 3, stride 1, with a ReLU) MaxPool (size 2) Linear Regression classifier My issue is that my network isn't converging, on anything. None of the weights appear to change. If I run it the p