Adaptation and contextualization of deep neural network models

Dimitrios Kollias, Miao Yu, Athanasios Tagaris, Georgios Leontidis, Andreas Stafylopatis, Stefanos Kollias

Research output: Chapter in Book/Report/Conference proceedingPublished conference contribution

16 Citations (Scopus)


The ability of Deep Neural Networks (DNNs) to provide very high accuracy in classification and recognition problems makes them the major tool for developments in such problems. It is, however, known that DNNs are currently used in a 'black box' manner, lacking transparency and interpretability of their decision-making process. Moreover, DNNs should use prior information on data classes, or object categories, so as to provide efficient classification of new data, or objects, without forgetting their previous knowledge. In this paper, we propose a novel class of systems that are able to adapt and contextualize the structure of trained DNNs, providing ways for handling the above-mentioned problems. A hierarchical and distributed system memory is generated and used for this purpose. The main memory is composed of the trained DNN architecture for classification/prediction, i.e., its structure and weights, as well as of an extracted - equivalent - Clustered Representation Set (CRS) generated by the DNN during training at its final - before the output - hidden layer. The latter includes centroids - 'points of attraction' - which link the extracted representation to a specific area in the existing system memory. Drift detection, occurring, for example, in personalized data analysis, can be accomplished by comparing the distances of new data from the centroids, taking into account the intra-cluster distances. Moreover, using the generated CRS, the system is able to contextualize its decision-making process, when new data become available. A new public medical database on Parkinson's disease is used as testbed to illustrate the capabilities of the proposed architecture.

Original languageEnglish
Title of host publication2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Number of pages8
ISBN (Electronic)9781538627259
Publication statusPublished - 2 Feb 2018
Event2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Honolulu, United States
Duration: 27 Nov 20171 Dec 2017

Publication series

Name2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017 - Proceedings


Conference2017 IEEE Symposium Series on Computational Intelligence, SSCI 2017
Country/TerritoryUnited States


  • adaptation
  • classification
  • clustered representation sets
  • contextualization
  • Deep neural networks
  • Parkinson's disease
  • prediction


Dive into the research topics of 'Adaptation and contextualization of deep neural network models'. Together they form a unique fingerprint.

Cite this