A multi-agent knowledge integration process for enterprise management innovation from the perspective of neural network

2022 ◽  
Vol 59 (2) ◽  
pp. 102873
Author(s):  
Zongke Bao ◽  
Chenguang Wang
Author(s):  
Sebastian Rudolph ◽  
Madalina Croitoru
Keyword(s):  

Author(s):  
Evgen Yuriiovych Sakhno ◽  
Nataliia Viktorivna Moroz ◽  
Serhii Ivanovych Ponomarenko

Urgency of the research. The study of the methods of economic and financial evaluation of the effectiveness of development projects is an underdeveloped area of economic knowledge, due to the impossibility of predicting the obstacles associated with investments in the future, and the implementation of such projects is most often associated with risk and uncertainty. Target setting. Therefore, in this case, it is proposed to use fuzzy logic theory, which defines a modern approach to describe business processes that present uncertainty and inaccuracy of the source information. Actual scientific researches and issues analysis. The question of using the theory of fuzzy logic in the management of development projects is highlighted in the scholarly works of Ukrainian and foreign scholars such as Asai K, Borisov A. N., Gordienko I. V., Semenenko M. V., Mityushkin Yu. I., Mokin B. I. and others. Uninvestigated parts of general matters defining. Known studies have shown that classical control methods work quite effectively at fully deterministic control objects and environments, and for systems with incomplete information and high complexity, fuzzy analysis methods that are optimal to be adapted to a modern project management system for constructing an integrated neural network are optimal. The research objective. The task is to use the fuzzy models to move on to the development of modern management technology with the use of artificial neural networks to integrate the enterprise management system and development projects. The statement of basic materials. The transition from traditional control systems to systems with fuzzy logic occurs using fuzzy variables. Let's consider the process of neural network modeling in the integration of enterprise management systems and development projects for the construction of a single integrated enterprise management system. Conclusion. In this paper we propose a methodology for the implementation of investment projects for the implementation of information systems based on fuzzy-plural approach, which allows to take into account qualitative aspects that do not have an exact numerical evaluation.


Author(s):  
Yanlin Han ◽  
Piotr Gmytrasiewicz

This paper introduces the IPOMDP-net, a neural network architecture for multi-agent planning under partial observability. It embeds an interactive partially observable Markov decision process (I-POMDP) model and a QMDP planning algorithm that solves the model in a neural network architecture. The IPOMDP-net is fully differentiable and allows for end-to-end training. In the learning phase, we train an IPOMDP-net on various fixed and randomly generated environments in a reinforcement learning setting, assuming observable reinforcements and unknown (randomly initialized) model functions. In the planning phase, we test the trained network on new, unseen variants of the environments under the planning setting, using the trained model to plan without reinforcements. Empirical results show that our model-based IPOMDP-net outperforms the other state-of-the-art modelfree network and generalizes better to larger, unseen environments. Our approach provides a general neural computing architecture for multi-agent planning using I-POMDPs. It suggests that, in a multi-agent setting, having a model of other agents benefits our decision-making, resulting in a policy of higher quality and better generalizability.


2021 ◽  
Vol 9 ◽  
Author(s):  
Albert Scrieciu ◽  
Alessandro Pagano ◽  
Virginia Rosa Coletta ◽  
Umberto Fratino ◽  
Raffaele Giordano

There is a growing interest worldwide on the potential of nature-based solutions (NBSs) as measures for dealing with water-related risks while producing multiple co-benefits that can contribute to several societal challenges and many of the sustainable development goals. However, several barriers still hamper their wider implementation, such as mainly the lack of stakeholders’ engagement and the limited integration of stakeholders’ knowledge throughout the phases of NBS design and implementation. This is a crucial aspect to guarantee that the multidimensional implications of NBSs are adequately understood and considered by decision-makers. Innovative methods and tools for improving NBS design and supporting decision-makers in overcoming the main barriers to implementation, ultimately enhancing their effectiveness, are therefore needed. The present work proposes a combined approach based on the integration of fuzzy cognitive maps, hydraulic modeling, and participatory Bayesian belief networks aiming to facilitate the stakeholders’ engagement and the knowledge integration process in NBS design and assessment. The approach was developed and implemented within the NAIAD project in the Lower Danube demo site, specifically oriented to support the process of the Potelu Wetland restoration. First, fuzzy cognitive maps are adopted as a problem structuring method for eliciting stakeholders’ risk perception and problem understanding, and for constructing a causal model describing the system as a whole, with specific attention to the expected role of the NBS in reducing flood risk and addressing the key local challenges. Second, hydraulic modeling is used to analyze the effect of extreme floods starting from the retrospective analysis of a specific event and to model the potential benefits of risk reduction measures. Last, a Bayesian belief network is used to support the model integration process and a scenario analysis with a user-friendly tool. The whole process can be replicated in other areas and is particularly suitable to support an active engagement of stakeholders (both institutional and not) in the process of NBS design and assessment.


2020 ◽  
Author(s):  
Douglas Meneghetti ◽  
Reinaldo Bianchi

This work proposes a neural network architecture that learns policies for multiple agent classes in a heterogeneous multi-agent reinforcement setting. The proposed network uses directed labeled graph representations for states, encodes feature vectors of different sizes for different entity classes, uses relational graph convolution layers to model different communication channels between entity types and learns distinct policies for different agent classes, sharing parameters wherever possible. Results have shown that specializing the communication channels between entity classes is a promising step to achieve higher performance in environments composed of heterogeneous entities.


Sign in / Sign up

Export Citation Format

Share Document