Dimensions in convolutional neural network
I am trying to understand how the dimensions in convolutional neural network behave. In the figure below the input is 28-by-28 matrix with 1 channel. Then there are 32 5-by-5 filters (with stride 2 in height and width). So I understand that the result is 14-by-14-by-32. But then in the next convolutional layer we have 64 5-by-5 filters (again with stride 2). So why the result is 7-by-7- by 64 and not 7-by-7-by 32*64? Aren't we applying each one of the 64 filters to each one of the 32 channels?
One filter is the sum of all the dimensions in the previous layer. This means that the 5x5 filter sums up over all 32 dimensions and in essence is a weighted sum of 32*5*5 values. However the weight values are shared across dimensions. Then there are 64 such filters. A better explanation with images can be found here: http://cs231n.github.io/convolutional-networks/.
- Why is the accuracy for my Keras model always 0 when training?
- The concept of straight through estimator (STE)
- How to convert the output of an artificial neural network into probabilities?
- Single-headed transformer model, output and target tensor sizes don't match
- How to add the bounding box values to the labels text files during prediction with a trained YOLO-V8 instance segmentation model?
- Inconsistency in input shape to neural network
- How to calculate optimal batch size?
- How to handle categorical features in neural network?
- Shapes (4,1) and (2,1) not aligned: 1 (dim 1) != 2 (dim 0)
- Are neural networks really abandonware?
- Does static quantization enable the model to feed a layer with the output of the previous one, without converting to fp (and back to int)?
- How to get an output dimension for each layer of the Neural Network in Pytorch?
- Merge weights of same model trained on 2 different computers using tensorflow
- pytorch BCEWithLogitsLoss calculating pos_weight
- Should the random noise given to a GAN kept constant?
- Why Do Two Different Ways of Adding a Drop Out Layer in Tensor Flow Function Differently?
- Fast sigmoid algorithm
- Why does my pruned model have a larger file size than my initial model?
- Dimensions in convolutional neural network
- Uninitialized nested pointers after initializing them (Not deterministic) [C++]
- When to use / not use bias term in convolutional neural networks
- Computing the Jacobian matrix in TensorFlow
- Recurrent neural network trained with torch.autograd predicts nonsense
- Learning rate not updating
- Finite basis physics-informed neural networks (FBPINNs) JAX problem
- Investigating discrepancies in TensorFlow and PyTorch performance
- Neural network to approximate the square function gives 0 output
- What's the correct way of expressing Residual Block with forward function of pytorch?
- Feature Importance of a Pytorch AutoEncoder
- How can I fix a problem with logit(expit(50)) = inf in a custom neural network model?