Patent Issued for Systems and methods for graph-based AI training (USPTO 12043280).

Předmět:	INFORMATION technology DATA structures KNOWLEDGE graphs ARTIFICIAL intelligence DATABASES
Zdroj:	Blood Weekly; 8/15/2024, p811-811, 1p
Abstrakt:	A patent has been issued to dRISK Inc. for systems and methods for graph-based AI training. The patent describes using graphs to represent data and encode information within nodes and edges. By encoding a data set within the structure of a graph, AI systems can interact with the entire space of information and learn more complex behaviors. The patent also discusses the challenges of training AI systems with large and complex data sets, and proposes using graphs to generate a manifold that captures the complexity of the required behavior. The AI model can be trained by traversing the manifold. The patent provides detailed claims and methods for implementing the graph interface system. [Extracted from the article]
Databáze:	Complementary Index
Externí odkaz:	Zobrazit plný text záznamu