Learning to relate terms in a multiple agent environment

Author(s):  
P. Brazdil ◽  
S. Muggleton
Keyword(s):  
1984 ◽  
Vol 6 (1) ◽  
pp. 10-19
Author(s):  
Giulio J. D'Angio

Major advances have been made in the understanding and management of the malignant diseases of childhood. More than 50% of children with cancer can now be expected to survive five or more years; a few decades ago, most of these patients died within 1 year. These good results have been obtained through the use of combined-modality therapy; that is, the conjoined use of surgery, radiation therapy, and multiple-agent chemotherapy. Wilms' tumor provides a spectacular example (Fig 1). Although achieving higher cure rates, combined-modality treatment is often rigorous, and has its associated early and late complications. The goals of modern pediatric oncology reflect both of these facts. Higher cure rates continue to be sought, but there is a growing recognition that not all patients need maximum treatment. Therapy can now be modulated according to well-defined prognostic factors for most of the malignant conditions. In that way, the most aggressive therapies are reserved for those at highest risk, while those with a good prognosis can be managed less intensively. The objectives of modern management, then, are to cure most patients while at the same time minimizing, as much as possible, the associated deleterious late consequences of successful treatment. wilms' tumor and neuroblastoma serve as good examples to demonstrate the above points.


Author(s):  
Gregor Mehlmann ◽  
Markus Häring ◽  
René Bühling ◽  
Michael Wißner ◽  
Elisabeth André

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.


2012 ◽  
pp. 553-564
Author(s):  
Wei-Shuo Lo ◽  
Tzung-Pei Hong

The fashion industry is experiencing rapid changes in many areas, including the supply chain. Typical quick response (QR) systems have been broadly used in the fashion industry to enable agile supply chain management (SCM). However, the original functions of QR systems cannot completely address the challenge of issuing early warnings to prevent customer loss. This article merges the typical MIS system development procedure with that of an e-SCM multiple-agent decision support system to confront this problem. The system has three levels: data mining, ontology, and decision support. These levels are interlinked in handling different databases. Different agents execute different tasks at each level to achieve integration and communication in a supply chain with less human intervention. The proposed framework emphasizes transparent connections among businesses and assists in information sharing, thereby preventing customer loss.


Sign in / Sign up

Export Citation Format

Share Document