Answered
Custom Layer Learnable Parameters Initialization.
This is now supported via the optional method initialize: https://www-integ2.mathworks.com/help/deeplearning/ug/define-custom-de...

2 years ago | 1

| accepted

Answered
Disable the gpu.
As soon as you start MATLAB type setenv CUDA_VISIBLE_DEVICES -1 MATLAB will no longer recognise any GPU for computation.

2 years ago | 2

Answered
Algorithm to extract linearly dependent columns in a large scale [-1,1] matrix ( 10^5 by 10^6)
You might consider using distributed arrays on an HPC cluster.

2 years ago | 0

Answered
Accuracy problems using mldivide on GPU
The answer is not less accurate, it is just as inaccurate. What you have are 60001 massively overdetermined systems generated by...

2 years ago | 1

Answered
When training a neural network does every training image need to have every pixel class?
The answer to question 1 is no. The answer to question 3 is that probably 126 training images is not sufficient, you might need ...

2 years ago | 0

Answered
How to use Sum and Dot function in GPU computation with arrayfun?
You cannot perform vector operations in GPU arrayfun. The a and b arguments to your function are not the whole array, they are t...

2 years ago | 0

| accepted

Answered
Double precision in deep learning
This is possible with a |dlnetwork| but it is (currently) more of a workaround than anything else. Once your dlnetwork is ready ...

2 years ago | 0

| accepted

Answered
When I want to train Fully convolutional neural network, I have the following error
I suspect your best bet here is to upgrade MATLAB to a more recent version. Many of these bugs will be fixed with newer versions...

2 years ago | 0

Answered
Error saying ''Dot Indexing is not supported for variables of this type".
This could be a bug, especially if you didn't modify the example code. What is the data you passed to |adamupdate|?

2 years ago | 0

Answered
YoloV4 - Out of memory
Generally the best solution here is to reduce the size of the input data. Still, these object detector networks do seem to be...

2 years ago | 1

| accepted

Answered
Why is my GPU code faster with the profiler on in RTX GPUs?
This is due to an optimization which is not performing ideally under memory pressure. If you reduce the size of your input you'l...

2 years ago | 0

| accepted

Answered
Conflicting behaviour of arrayfun() with gpu: example that works and example of error
The function normcdf isn't supported by GPU arrayfun because it accepts varargin. For a list of supported functions see the docu...

2 years ago | 0

| accepted

Answered
How to initialize a string variable, and pass it to the matlab function using GPU coder
MATLAB and Simulink code generation do not currently support string. Edit: Sorry, my bad, it does support scalar strings, but n...

2 years ago | 0

| accepted

Answered
need to plot the accuracy vs epoch graph
Add Plots="training-progress" to your training options. FWIW, you shouldn't use ReadFcn for resizing images, it dramatically sl...

2 years ago | 1

Answered
Update BatchNorm Layer State in Siamese netwrok with custom loop for triplet and contrastive loss
Interesting question! The purpose of batch norm state is to collect statistics about typical inputs. In a normal Siamese workflo...

2 years ago | 0

| accepted

Answered
gpu arrayfun don't support linspace or NaN array
You cannot create an array inside a call to GPU arrayfun, only scalars.

2 years ago | 0

Answered
GPU Support for RTX 4090
Forgive me for needing to correct Walter, but the last three versions of MATLAB _will_ natively support the 4000 series because,...

2 years ago | 2

Answered
mexcuda gives unsupported GNU version error
R2022a uses CUDA 11.2, not 11.7. I suspect that the actual compiler that ends up being used is the version of nvcc shipped with ...

2 years ago | 0

| accepted

Answered
GPU speed up for pcg() is disappointing
I'm guessing LL' is extremely dense, which will explain why the solver stalls. On the GPU the preconditioning is (currently) per...

2 years ago | 0

| accepted

Answered
How to implement Siamese network with the two subnetworks not share weights
You can try gathering the weights back from each network after you've used it, as in net = dlupdate(@gather,net). This should sa...

2 years ago | 0

Answered
Speed up inference or/and training of a 3D deep neural network (U-net) for a regression task
Have you tried using dlaccelerate? As well as ensuring any Custom Layers are using the Acceleratable mixin?

2 years ago | 1

| accepted

Answered
Matrix multiplication optimization using GPU parallel computation
The Windows Task Manager lets you track GPU utilization and memory graphically, and the utility nvidia-smi lets you do it in a t...

2 years ago | 1

Answered
How to increase MiniBatchSize
It depends on what you're doing. Some ideas: * Get a new GPU with more memory * Use a smaller model * If your model accepts...

2 years ago | 0

Answered
Matlab trainNetwork CNN training pauses iterating intermittently at random then continues
Is the pause associated with a validation measurement being added to the training plot? With 7 times as much validation data it ...

2 years ago | 0

Answered
problems with @arrayfun on GPU
This is a bug. I have reported it. Thanks for finding it! In the meantime, you can work around the issue by using a local funct...

2 years ago | 0

| accepted

Answered
A problem when using "multi-gpu" as "ExecutionEnvironment" for training a CNN
Most likely this is this issue, which is fixed in the latest update to R2022a. You can also try downgrading your GPU drivers.

2 years ago | 0

| accepted

Answered
Perform mldivide between 3x3 matrix M and every RGB pixel in a image in GPU
I feel like I'm missing something - this is just a single backslash with multiple right-hand sides, or to avoid permutation a si...

2 years ago | 1

Answered
Library not loaded: @rpath/libcudart.10.2.dylib
This problem should now be fixed at Apple, please reboot and report here if you are still experiencing issues.

2 years ago | 0

Answered
Warning: GPU is low on memory
A 3-D U-net is a very large model. Try reducing |patchSize|, |patchPerImage|, |miniBatchSize| and |inputSize|.

2 years ago | 0

| accepted

Answered
How to run lane detection optimized with GPU coder project on matlab
https://www.mathworks.com/help/gpucoder/ug/lane-detection-optimized-with-gpu-coder.html

2 years ago | 0