Visualize a neural network activation using JavaFX

Sometimes you want to visualize what is going on inside your neural network to identify what you could improve. This is not an easy task and may require patience. To help on this task I am sharing today a small JavaFX application that will go through all the activated layers and show the activated output and the ones that were not updated. The results are interesting, see some screenshots I took using the Brazilian coin model, which was trained on a deep pre-trained Resnet50 neural network:

At the early layers we can see some sort of the shadow of the coin

In the middle we can see some features that were activated

Close to the end only some outputs were activated

It is confusing when we check these huge third party neural networks. It is more interesting when we inspect some neural network we created ourselves. Remember when we used a trained MNIST model in a JavaFX application? These are the layers after an image classification:

This was possible using the activate method in the layer of a pre-trained model. First you select the model, set its input and then select some image. The Layer.activate method should return an INDArray with the values for the last entered data. We go through this array, normalize its values and build a visual representation using JavaFX Canvas. This is all possible thanks to the great DeepLearning4J APIs. The API behind DL4j, ND4J, is awesome!

The application code is on my github. It is a maven project, what you have to do is import it on an IDE and the App class or configure the maven exec plugin to execute the App class. If you are interested in testing this application but having trouble to run it please let me know in the comments.

It does seen to be very stable and there are some bugs, I mean, it seems that the results are not reproducible and I guess I am doing something wrong, but I will try to update the code when I find what is wrong. There's also space for improvements. like adding colors or find a way to mix all the channels values, but I am pretty satisfied with this result so far. Feel Free to send me pull requests and give me feedback, I will be happy to know if someone at least tried to run it on your machine.

Comentários

Postagens mais visitadas deste blog

Dancing lights with Arduino - The idea

I have been having fun with Arduino these days! In this article I am going to show how did I use an electret mic with Arduino to create a Dancing Lights circuit. Dancing Lights I used to be an eletronician before starting the IT college. I had my own electronics maintenance office to fix television, radios, etc. In my free time I used to create electronic projects to sell and I made a few "reais" selling a version of Dancing lights, but it was too limited: it simply animated lamps using a relay in the output of a 4017 CMOS IC. The circuit was a decimal counter controlled by a 555. 4017 decimal counter. Source in the image When I met Arduino a few years ago, I was skeptical because I said: I can do this with IC, why should I use a microcontroller. I thought that Arduino was for kids. But now my pride is gone and I am having a lot of fun with Arduino :-) The implementation of Dancing Lights with Arduino uses an electret mic to capture the sound and light leds...

Simplest JavaFX ComboBox autocomplete

Based on this Brazilian community post , I've created a sample Combobox auto complete. What it basically does is: When user type with the combobox selected, it will work on a temporary string to store the typed text; Each key typed leads to the combobox to be showed and updated If backspace is type, we update the filter Each key typed shows the combo box items, when the combobox is hidden, the filter is cleaned and the tooltip is hidden: The class code and a sample application is below. I also added the source to my personal github , sent me PR to improve it and there are a lot of things to improve, like space and accents support.

Genetic algorithms with Java

One of the most fascinating topics in computer science world is Artificial Intelligence . A subset of Artificial intelligence are the algorithms that were created inspired in the nature. In this group, we have Genetic Algorithms (GA). Genetic Algorithms To find out more about this topic I recommend the following MIT lecture and the Nature of Code book and videos created by Daniel Shiffman. Genetic Algorithms using Java After I remembered the basics about it, I wanted to practice, so I tried my own implementation, but I would have to write a lot of code to do what certainly others already did. So I started looking for Genetic Algorithm libraries and found Jenetics , which is a modern library that uses Java 8 concepts and APIs, and there's also JGAP . I decided to use Jenetics because the User Guide was so clear and it has no other dependency, but Java 8. The only thing I missed for Jenetics are more small examples like the ones I will show i...

FXApps

Pesquisar este blog