AR-Neurons, AI Development Environment

Beschreibung

The AR-Neurons Development Environment is build on top of the Trust-Smalltalk blockchain framework, to provide neuronal network artificial intelligence systems for the need of Augmented Reality. These are mainly object, language and pattern recognition for the user interface and the mapping of the World.

But also many other applications like virtual robotics are possible. As with the KaraSpace privacy principles, the user can inspect the open source system and the running data even while it is operating. This gives a perfect piece of mind what the system is actually doing with the data.

The AR-Neurons system has a binding to TensorFlow, the state of the art, and open source machine learning system, making the execution extremely fast on specially designed hardware. It keeps the system updated to the latest developments in neuronal network deep learning advances.

The system has the Roassel visualisation engine available to visualize and understand the complex behavior of the cascaded neuronal networks.

Modern deep learning systems consist of many individual neuronal networks that perform slightly different tasks, and need to be cascaded to perform the complete object recognition.

The linking of these networks will become increasingly complex, to perform more sophisticated tasks.

While the individual neuronal networks have no means to perform secret tasks, the code to link these networks has the potential to grab any data of interest and forward it to anyone who is interested. For the user to maintain privacy in an extremely monitoring environment like AR, these algorithms must stay transparent and open source at the users disposal.

So the delivery of neuronal network services and the development and visualisation tool is one system that further needs to be secured on the blockchain, so that it can not be tampered with, after being inspected and approved by the programing community.

 

Further information about TensorFlow at:  tensorflow.org

TensorFlow on GitHub: https://github.com/tensorflow

Pharo bindin to TensorFlow on GitHub: https://github.com/PolyMathOrg/libtensorflow-pharo-bindings