GBT Technologies announced that is now working on Graph Neural Network (GNN) based autonomous machine algorithms development, as part of the research to enhance its Avant! AI.
AI systems lend themselves to practical applications. GBT’s Avant! AI is a computer hardware and software system that is able to perform tasks that normally require human intelligence; it is based on machine learning technology and is capable of being trained with a data set.
GBT is a development-stage company which considers itself a native IoT creator, developing Internet of Things (IoT) and Artificial Intelligence (AI) enabled mobile technology platforms. Gopher has a portfolio of Intellectual Property that, when commercialized, will include smart microchips, mobile and security applications and protocols, and supporting cloud software. Gopher’s system envisions the creation of a global mesh network.
The core of the system will be its advanced microchip technology that can be installed in any mobile or fixed device worldwide. GBT envisions this system as a low-cost, secure, private mesh network between any enabled devices, providing shared processing, advanced mobile database management/sharing and enhanced mobile features as an alternative to traditional carrier services.
GBT is now developing a set of algorithms within Avant! that are leveraging neural nets to process a huge amount of data, specifically for autonomous machines applications. The new technology is based on a Graph Neural Network (GNN) – a type of Neural Network which directly operates on the Graph structure, modeling the dependencies between nodes in a graph and enabling advanced algorithms in a supervised manner.
The data will be analyzed in multiple layers of a neural network to produce probabilities of a certain outcome. Avant! then will take the data processed from its deep learning and make ongoing decisions about responses and actions. The system’s advantage is the capability to enable efficient data analysis without any manual human assistance, automatically dig deeper into insights based on what it has learned, and making conclusions in real time.
“We develop advanced AI technology to solve problems that only humans could have solved in the past” stated Danny Rittman, GBT’s CTO. “Through our development projects, we are seeking to implement our deep learning technology specifically for conditions including autonomous cars. Avant! already includes advanced pattern recognition, sequence learning, and neural network graphs, and now our goal is to expand into supervised, dynamic GNN based algorithms to sharpen its learning curve and decision making.
Avant!’s new class of neural networks – the GNNs – will capture the dependence of vast amounts of nodes, each node representing a condition/situation, and will build a dependency tree in order to create reasoning models. The outcome of this development is a sophisticated AI system which is a perfect match for safe autonomous vehicles.
Besides the complex task of getting constant inputs about the vehicle and its surroundings conditions, analyzing it and making decisions in real time, we also target Avant! to become personalized to the user, learning his or her habits and life style, advising about anything from what type of coffee to buy or keeping them on schedule for an appointment.”