Artificial Neural Network (ANN) Visualization

[Quant3D]

I wrote an interactive neural network visualization web program where people can explore the inner workings of neural networks.

Here is the description page (where I mention and link to BabylonJS in the FAQ):  miabellaAI.net

This is the browser-based application:  ann.miabellaAI.net

 

 

[jerome]

Impressive

[satguru]

Impressive indeed!

what do you think about this?

https://tenso.rs/

worth exploring?

[Quant3D]

@satguru: Thanks for mentioning TensorFire (tenso.rs). Definitely worth checking out.

I've done some work with GPU-accelerated computing in the past. I think their claim of a possible 100x speed-up is about right, provided that a problem can be divided into pieces (i.e., the solution doesn't depend on sequential calculations or other dependencies between the data points).

[Quant3D]

A couple months ago I posted this topic about a neural network visualization I wrote (ann.miabellaAI.net).

I've greatly expanded the interaction between BabylonJS and HTML/CSS. As long as your display elements are in front of your WebGL scene, CSS allows for a wide variety of dynamic effects.

Here's a tip: there can be issues with rendering quality when zooming in on HTML text and canvas graphics. If an element is going to be zoomed in and out, one idea is to start by making it oversized and then shrinking accordingly.

That is what I do here. To expand or shrink the node graphs, you press the "x" key or click the main mouse button.