Open AccessThe Neural Modules Network with Collective Relearning for the Recognition of Diseases: Fault- Tolerant Structures and Reliability Assessment
Author(s) -
Iraj Elyasi Komari,
Mykola Fedorenko,
Vyacheslav Kharchenko,
Yevhenia Yehorova,
Nikolaos G. Bardis,
Liudmyla Lutai
Publication year - 2020
Publication title -
international journal of circuits, systems and signal processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.156
H-Index - 13
ISSN - 1998-4464
DOI - 10.46300/9106.2020.14.102
Subject(s) - redundancy (engineering) , artificial neural network , computer science , generalization , reliability (semiconductor) , reliability block diagram , artificial intelligence , architecture , block diagram , block (permutation group theory) , fault tolerance , data mining , machine learning , distributed computing , reliability engineering , engineering , fault tree analysis , mathematics , geography , operating system , mathematical analysis , power (physics) , physics , archaeology , geometry , electrical engineering , quantum mechanics
The article presents the architecture of multi-level information-analytical system (IAS) based on the neural modules network (NMN). This network consists of neural modules which are placed at the three levels (local, region and nation geographically distributed medical centers). Procedures of learning and collectiverelearning of neural modules consider region particularities and are based on analysis, generalization and exchange of experience related to diagnosis of diseases. These procedures provide modification and filtering parameters used as input for the further learning of local and regional neural modules.A few fault-tolerant structures of NMN-based IAS are researched taking into account different options of server and communication redundancy. Reliability block diagrams for redundant IAS structures are developed and formulas for calculation of probability of upstate are analyzed.