A Multi-Agent Motion Prediction and Tracking Method Based on Non-Cooperative Equilibrium

A Multi-Agent Motion Prediction and Tracking method based on non-cooperative equilibrium (MPT-NCE) is proposed according to the fact that some multi-agent intelligent evolution methods, like the MADDPG, lack adaptability facing unfamiliar environments, and are unable to achieve multi-agent motion prediction and tracking, although they own advantages in multi-agent intelligence. Featured by a performance discrimination module using the time difference function together with a random mutation module applying predictive learning, the MPT-NCE is capable of improving the prediction and tracking ability of the agents in the intelligent game confrontation. Two groups of multi-agent prediction and tracking experiments are conducted and the results show that compared with the MADDPG method, in the aspect of prediction ability, the MPT-NCE achieves a prediction rate at more than 90%, which is 23.52% higher and increases the whole evolution efficiency by 16.89%; in the aspect of tracking ability, the MPT-NCE promotes the convergent speed by 11.76% while facilitating the target tracking by 25.85%. The proposed MPT-NCE method shows impressive environmental adaptability and prediction and tracking ability.

Human-Agent Team Game Analysis Framework: Case Studies

This research details the development of a human-agent team (HAT) analysis framework specifically aimed at video games. The framework identifies different dimensions of interest related to humans and software agents working together. Video games have a variety of user-tested interaction paradigms that may offer useful insights into HAT dynamics, but it can be difficult for researchers to know which games are relevant to their research without a systematic method of characterizing HAT relationships. The framework was developed based on previous literature and gameplay analysis. This paper offers three case studies, applying the framework to the games Madden 21, Call to Arms, and Civilization V. Possible trends related to agent intelligence, team structures, and interdependence are discussed.

Arena: A General Evaluation Platform and Building Toolkit for Multi-Agent Intelligence

Learning agents that are not only capable of taking tests, but also innovating is becoming a hot topic in AI. One of the most promising paths towards this vision is multi-agent learning, where agents act as the environment for each other, and improving each agent means proposing new problems for others. However, existing evaluation platforms are either not compatible with multi-agent settings, or limited to a specific game. That is, there is not yet a general evaluation platform for research on multi-agent intelligence. To this end, we introduce Arena, a general evaluation platform for multi-agent intelligence with 35 games of diverse logics and representations. Furthermore, multi-agent intelligence is still at the stage where many problems remain unexplored. Therefore, we provide a building toolkit for researchers to easily invent and build novel multi-agent problems from the provided game set based on a GUI-configurable social tree and five basic multi-agent reward schemes. Finally, we provide Python implementations of five state-of-the-art deep multi-agent reinforcement learning baselines. Along with the baseline implementations, we release a set of 100 best agents/teams that we can train with different training schemes for each game, as the base for evaluating agents with population performance. As such, the research community can perform comparisons under a stable and uniform standard. All the implementations and accompanied tutorials have been open-sourced for the community at https://sites.google.com/view/arena-unity/.

Multiversal SpaceTime (MSpaceTime) Not Neural Network as Source of Intelligence in Generalized Quantum Mechanics, Extended General Relativity, Darwin Dynamics for Artificial Super Intelligence Synthesis

From Synthesis perspective, whether Logic Synthesis, Physical Synthesis, Chemical Synthesis, or Biological Synthesis, Physical Geometry such as Universal Geometry and Quantum Geometry, and Biological Geometry like Conformal Geometry supported by Tensors and Manifolds, are the outcome of physical laws and biological laws in modeling non-linear physical and biological dynamics as opposed to traditional partial differential/difference equation way. We discover that Multiversal SpaceTime instead of Neural Network, governing physical and biological world at macroscopic and microscopic level, is the ultimate source of intelligence. With that we propose Multiversal Synthesis-based Artificial Design Automation (ADA), a bio-physical inspired model based on Multiverse in Darwin Dynamics, Generalized Quantum Mechanics, and Extended General Relativity, for Artificial Super Intelligence (ASI) implementation. Based on Schrodinger Equation of Quantum Mechanics, we generalize the 4-Dimensional Hilbert Space based Discrete Quantum SpaceTime to N-Dimensional (1 ≪ N < M, with M is limited by Planck Length) Hilbert Space based Discrete MSpaceTime as part of MSpaceTime, in modeling both Micro-Environment Intelligence and Micro-Agent Intelligence of ASI; likewise based on Einstein Equations of General Relativity, we make a T-Symmetry extension first, and then extend the 4-Dimensional Pseudo-Riemannian Manifold based Continuous Curved SpaceTime as part of MSpaceTime to N-Dimensional (1 ≪ N < ∞) Pseudo-Riemannian Manifold based Continuous MSpaceTime extension, in modeling both Macro-Environment Intelligence and Macro-Agent Intelligence of ASI. Our discovery only solves the black box puzzle of AI, but also paves the way in achieving ASI through ADA. Of course, our Multiverse Endeavor will never stop from there.

Improving the Cognitive Agent Intelligence by Deep Knowledge Classification

In this paper, a new idea is developed for improving the agent intelligence. In fact with the presented convolutional neural network (CNN) approach for knowledge classification, the agent will be able to manage its knowledge. This new concept allows the agent to select only the actionable rule class, instead of trying to infer its whole rule base exhaustively. In addition, through this research, we developed a comparative study between the proposed CNN approach and the classical classification approaches. As foreseeable the deep learning method outperforms the others in term of classification accuracy.

THE HYBRID AGENT MODEL OF BEHAVIORAL TESTING

Operation testing and diagnostic tests, applied for distributed information systems, inherit and employ the properties of distribution, autonomy, goal formation and cooperation, natural for the multi-agent systems. This paper presents the behavioral diagnostics agent model, based on the evolutionary organization of component tests in the automata network environment. The model can be used to construct a multi-agent diagnostics system. A hybrid agent model provides a combination of reactive operation testing and deliberative diagnostic tests, based on the deterministic and evolutionary methods of synthesis of behavioral tests. An agent model consists of the component models of allocation environment, functioning goals and strategies, operations of observation, enforcement strategy and adaptation, initial component models, goals and strategies for ensuring the autonomy. Agent intelligence is based on a locally-exhaustive deterministic and pseudorandom targeted evolutionary synthesis of behavioral tests, providing and accumulating the results. Cooperation of the agents involves their deterministic and evolutionary interactions under the conditions of test feasibility and portability.

Intelligent Information Retrieval System Based on Muti-Agent Model

The intelligent information retrieval model discussed in this paper is constructed by multi-agent. Currently, BDI cognitive theory is accepted widely by the scholars of this field, but the related researches mainly focus on the theoretical derivation and presentation of symbols, lack of model facing practical application. In this article, a dynamic rule-based reasoning model is proposed. The model based on BDI theory is an expression of agent intelligence. The basic logical reasoning of the BDI theory is extended in this article. The author not only introduces several functions to study the dynamic changes of agents mental state, but also give a detailed description of how to use the theory of production rules to express BDI-based reasoning. This article also studies and designs the agent communication mechanism in MAS. Finally, the intelligent information retrieval system is designed and implemented with the idea of AOP.