I think you have to give emphasize on optimizing algorithm which can increase performance. Choose deep learning algorithm as per your data and requirement.

You may find something useful at this link

rtinsights.com/top-deep-learning-tools/

MATLAB is the best tool in dealing all such computing/visualization tasks. I would like to provide you some useful links that may be helpful as per your query + More. I am sure by following all these links deeply, you might be the leader in this field.

https://www.mathworks.com/help/deeplearning/examples/train-a-convolutional-neural-network-for-regression.html;jsessionid=2c56a20c391fba5cd00177e8cd34

----

https://www.manopt.org/downloads.html

https://www.mathworks.com/products/deep-learning.html

----

https://stats.stackexchange.com/questions/35185/dimensionality-reduction-svd-or-pca-on-a-large-sparse-matrix?rq=1

---

https://www.google.com/search?source=hp&ei=BoaOXfSzLdOZmgeD0YGoDg&q=bayesian+deep+learning+tutorial&oq=Bayesian+deep+&gs_l=psy-ab.1.1.0l10.3273.7300..11240...0.0..0.231.2792.0j7j7......0....1..gws-wiz.......0i131.UnmbgzRX37Q

---

https://keras.io/examples/imdb_cnn/

----

https://towardsdatascience.com/deep-neural-networks-for-regression-problems-81321897ca33

---

Manifold Learning: The Theory Behind It - Towards Data Science

Models and layers | TensorFlow.js

Cluster Analysis - MATLAB & Simulink

https://www.mathworks.com/help/deeplearning/ref/trainingoptions.html

https://machinelearningmastery.com/how-to-develop-convolutional-neural-network-models-for-time-series-forecasting/

Thanks a lot for everyone participated in this discussion. I will respond to each reply, if required, to enrich the discussion

Very useful link ...May I ask you Mr. Mokhaled N. A. Al-Hamadani if you have used Python and managed to address the two constraints I mentioned (Several hidden layers and Visualization dynamically while running the code)?

Re your recommendation Mr. Mohsen Saroughi to use Keras, is it possible to integrate it with Simulink? I have used Mathworks solution (ANN toolbox + Simulink) and it can be easily to integrate the trained ANN with Simulink to test it along with other system components (e.g. nonlinear system mode, control law, and models for actuators). The concern is if I use Keras, will it be possible to integrate the trained net with Simulink?

I hope Mr. İhsan Tuğal that you read the above comments and provide from your experience with these tools on how to integrate a trained ANN with Mathworks Simulink? This is very important to decide on any external tool as much of the work is done using Matlab and Simulink.

If you are interested in the dynamic evaluation of your NN architecture, Pytorch is what you are looking for. Python has a variety of visualization tools to help you MATLAB is quite rigid comparatively. you will know the difference once you start building using High-end APIs (Pytorch, Tensorflow, Keras, Fastai).

The ANN that is being used to make a prediction is not affected by the framework used to implement it.

There are many open-source frameworks like Tensorflow, Pytorch, Keras, et.c available in python to implement an ANN. The Pytorch framework provides methods to see how the gradient is backpropagating through layers.

You can go through each of the frameworks to find a suitable one suiting your problem.

Mr. Purnendu Mishra , how about other tools mentioned above by Mr. Sanaullah Mehran Ujjan and Mr. Mohsen Saroughi ? Do they provide visualization as well?

How about integration with Simulink?

Thanks to you all for the helpful remarks

Um bom começo é usar Pytorch, Tensorflow e Keras. Essas ferramentas podem ajuda-lo em sua tarefa.

Mohammad Fahad Al-Malki none of the tools mentioned in this thread have inherent Simulink integration capability. you will either have to move your model completely on simulink. or write an C++ lnterface pakacge.

I don't think that visualising the internal states of the neural network will in any way be helpful to understanding with your network fails to predict correctly.

Likewise, the choice of language for constructing the network shouldn't affect the accuracy. Python is, however, very user friendly and Keras is as out-of-the-box easy as it gets, so it reduces the chance of user error. I would take some time to review your input data, and see if there's some simple issue with that that could be causing your problems.

Keras and PyTorch are popular open-source deep learning libraries. In the present time the Pytorch is at the top.

Mohammad Fahad Al-Malki I don't think Keras or Pytorch provide integration with Simulink.

Tensorflow and Keras are best libraries using Jupyter notebook for the implementation of any model of Deep Learning

https://www.ibm.com/uk-en/cloud/deep-learning

Thanks a lot for your contribution.

The main constraint preventing usage of other than Matlab platform is the integration of any of these excellent tools with Simulink. I hope to see some light shed on this aspect.

I think you have to give emphasize on optimizing algorithm which can increase performance. Choose deep learning algorithm as per your data and requirement.

Dear Mohammed Ashikur Rahman , the point I am after is how to integrate your work in Python with Simulink.

Tensorflow, Keras, Caffe, and PyTorch are the python based frameworks for deep learning.

deep-learning toolbox in Matlab.

Deep visualization toolbox for visualizing internal layer-wise output

in order to speed up the training of multilayer neural networks, use better optimization techniques like Adam, RMS Prop, or AdaGrad.

For visualization the internal working of NNs, use

https://github.com/yosinski/deep-visualization-toolbox

Thanks a lot Mr. Mohammad Khalid Pandit for your contribution.

The remaining subject of concern is to integrate Python (or any other toolbox) with Matlab/Simulink where one may have in Matllab environment some mathematical model and controller designed using e.g. Robust Control Toolbox and wish to integrate them all.

TensorFlow, Keras, PyCharm