How can I use TensorFlow library from matlab environment?

I want to use or access TensorFlow library from Matlab environment, especially for machine learning. How can I use TensorFlow library from matlab ?

[Royi Avital]

Really hope to see MATLAB support for Caffe / MxNet / CNTK / TensorFlow.

Who knows, maybe will have something soon.

[David Willingham]

Hi, MATLAB supports Framework Interoperability with ONNX and Tensorflow-Keras. More information on this can be found here.

Regards,

David Willingham

Deep Learning Product Manager, MathWorks

[John Daniels]

Hi David,

I currently have a CNN-LSTM model that I would like to call from MATLAB. It was developed in Keras 2.2.2. I have attempted a number of approaches that have all failed so far.

1. I tried to use the model as a function that is called from Matlab however, MATLAB crashes whenever the tensorflow/keras library is included in the script.
2. I tried to port the model over using the Tensorflow-Keras (importKerasLayers and importKerasNetwork) however the Conv1d layer is not available and custom layers can't seem to work given the 'channel last' format. In addition, it seems CNN-LSTM layers are not compatible.

Are there any solutions to this issue?

Best,

John.

[David Willingham]

Since 2021a, MATLAB supports Framework Interoperability with TensorFlow with a direct importer via importTensorFlowNetwork and importTensorFlowLayers. I'd recommend using these functions over importKerasNetwork or importKerasLayers.

Answer 1

[David Willingham]

Hi,

MATLAB supports Framework Interoperability with TensorFlow with a direct importer via importTensorFlowNetwork and importTensorFlowLayers.

Regards,

David Willingham

Deep Learning Product Manager, MathWorks

[Jack Ferrari]

On a related note, with R2022a and later, you can run cosimulations and deploy applications using MATLAB or Simulink with TensorFlow Lite (.tflite) models. With TFLite models, an import process is unnecessary. Instead, you can directly load .tflite model files into MATLAB with loadTFLiteModel.

This functionality requires the Deep Learning Toolbox Interface for TensorFlow Lite. More info on requirements can be found here: Prerequisites for Deep Learning with TensorFlow Lite Models.

Answer 2

[Louis Yu]

I've went about working on a middle-man solution for new users to Tensorflow that typically utilize Matlab. It's a simple GUI interface that auto-codes the user inputs in the Matlab GUI into a python script that can be run utilizing the Tensorflow Python Interface.

See the file exchange @ Tensorflow for Matlab

For more details on reasons for this implementation, a paper was written over the effort conducted "Benchmarking Open Source Machine Learning Frameworks and Development of Auto-coding Interface for Entry-Level Engineers" .

There is also a github repository I utilized for some test-cases that include Tensorflow snippets: Github Repository for Benchmarking

[Kenn]

@Louis Yu, congrats on your effort, Tensorflow for Matlab. It looks like it has high potential but there's something not quite right. I mentioned several bugs I found, which make it look like the code could never have worked. Perhaps an earlier, incomplete version got uploaded? Mind taking a look?

[Louis Yu]

Updates have been made to the file exchange upload. I've added a few helpful pieces for new users.

Answer 3

[Tom Anderson]

TensorFlow is a free Python library developed by Google Brain. As of April 2017, it has APIs in other languages (C++, Java and Go), but they are experimental.

MATLAB is a proprietary programming language developed by Mathworks (non-free). It has interfaces to other languages, including Python. Mathworks offers its own Neural Network Toolbox™. There are also plenty of other deep learning MATLAB toolboxes, many listed at http://deeplearning.net/software_links/.

However, running TensorFlow with MATLAB is not supported by TensorFlow or Mathworks. As we can see, the MathWorks File Exchange does not have http://au.mathworks.com/matlabcentral/fileexchange/?utf8=%E2%9C%93&term=tensorflow. Neither does GitHub https://github.com/search?utf8=%E2%9C%93&q=matlab+tensorflow&type=. Until there is a bridge between the two, I would suggest that we are best advised to learn TensorFlow in the supported language (Python). Alternatively, if you already have paid for the MATLAB licenses (Neural Network Toolbox $1,500 Individual) (or belong to an organisation that has), you may consider learning MATLAB and the Neural Network Toolbox. This field is moving quickly, so be prepared to keep paying for the new version of Neural Network Toolbox (updates are not included in the price).

[Christopher Stokely]

The MATLAB Neural network toolbos for an individual is about $50, not more than $1000. Look at the license options on the MathWorks website for individual license for noncommercial use.

The Rosetta Stone of deep learning is ONNX (Open Neural Network Exchange), which allows model's to be transferred (I think) between environments such as PyTorch, MXNet, Core ML, Caffe2, TensorFlow, Microsoft Cognitive Toolkit, and MATLAB - I think. Check it out for yourself by searching for the command

exportONNXNetwork

References:

1) Open Neural Network Exchange https://github.com/onnx

2) ONNX. https://onnx.ai

3) MATLAB Documentation https://www.mathworks.com/help/deeplearning/ref/exportonnxnetwork.html

From Mathworks webpage description, see the attached figure to see a summary of supported deep learning frameworks.

Answer 4

[Ashiqur Rahman]

Try this one may be. A MATLAB implementation of the TensorFlow Neural Network Playground

[Royi Avital]

It's not using TensorFlow.

Answer 5

[Alam Noor]

https://www.mathworks.com/videos/importing-tensorflow-keras-models-in-matlab-1530199279961.html

find here