Advancing Neuro-Symbolic AI with Abstract Urban Simulation

View a PDF file from the paper entitled Logicity: AI Neuro-Symbolic AI introduces an abstract urban simulation, written by Bowen Lee and 12 other authors

PDF view

a summary:Recent years have witnessed the rapid development of nervous intelligence systems (NESY), which integrates symbolic thinking into deep nerve networks. However, most of the current criteria for Nesy AI fail to provide long thinking tasks on the horizon with multiple complex agents. Moreover, it is usually restricted with fixed and simple logical rules on limited entities, making them far from the complications in the real world. To treat these decisive gaps, we offer Logicity, the first simulation dependent on the logic of the first -to -custom -ancient order (for) an urban -like environment with multiple dynamic factors. Logicity models diversity of urban elements using semantic and spatial concepts, such as ISAMBULANCE (X) and ISCLOSE (X, Y). These concepts are used to determine the rules for that governing the behavior of various agents. Since concepts and rules are abstracts, they can be applied globally to cities that have any agents, which facilitates the creation of various scenarios. Moreover, the main feature of Logicity is to support the user’s formal abstraction, allowing the complications of customized simulations for logical thinking. To explore different aspects of NESY AI, Logicity offers two tasks, one of which includes making a long -term serial decision, and the other focuses on visual thinking from one step, changing the behavior of difficulty and agent. Our intense evaluation reveals the NESY action feature in abstract thinking. Moreover, we highlight the great challenges of dealing with the most complex abstracts in multi -agents of long agents or under unbalanced high -dimensional data. Thanks to its flexible design, the various features and the newly uploaded challenges, we believe that Logicity represents a pivotal step forward in progressing the next generation of Nesy AI. All software instructions and data are open on our website: URL https