Home // International Journal On Advances in Software, volume 11, numbers 3 and 4, 2018 // View article

A Comprehensive Workplace Environment based on a Deep Learning Architecture for Cognitive Systems

Authors:
Thorsten Gressling
Veronika Thurner

Keywords: Cognitive system; Multilayer architecture; Technical work place; Machine learning; Ensemble averaging; Synthesized training data; Context sensitive; Neural network

Abstract:
Many technical work places, such as laboratories or test beds, are the setting for well-defined processes requiring both high precision and extensive documentation, to ensure accuracy and support accountability that often is required by law, science, or both. In this type of scenario, it is desirable to delegate certain routine tasks, such as documentation or preparatory next steps, to some sort of automated assistant, in order to increase precision and reduce the required amount of manual labor in one fell swoop. At the same time, this automated assistant should be able to interact adequately with the human worker, to ensure that the human worker receives exactly the kind of support that is required in a certain context. To achieve this, we introduce a multilayer architecture for cognitive systems that structures the system’s computation and reasoning across well-defined levels of abstraction, from mass signal processing up to organizationwide, intention-driven reasoning. By partitioning the architecture into well-defined, distinct layers, we reduce complexity and thus facilitate both the implementation and the training of the cognitive system. Each layer comprises a building block that adheres to a specific structural pattern, consisting of storage, processing units and components that are used for training. We incorporate ensemble methods to allow for a modular expansion of a specific layer, thus making it possible to introduce pretrained functional blocks into the system. In addition, we provide strategies for generating synthetical data that support the training of the neural network parts within the processing layers. On this basis, we outline the functional modules of a cognitive system supporting the execution of partially manual processes in technical work places. Finally, a prototypical implementation serves as a proof of concept for this multilayer architecture.

Pages: 358 to 367

Copyright: Copyright (c) to authors, 2018. Used with permission.

Publication date: December 30, 2018

Published in: journal

ISSN: 1942-2628