The utilization of deep learning in the field of industrial automation is hindered by two factors: The amount and diversity of training data needed as well as the need to continuously retrain as the use case changes over time. Both problems can be addressed by industrial deep transfer learning allowing for the performant, continuous and potentially distributed training on small, dispersed datasets. As a specific example, a dual memory algorithm for computer vision problems is developed and evaluated. It shows the potential for state-of-the-art performance while being trained only on fractions of the complete ImageNet dataset at multiple locations at once.
%0 Journal Article
%1 maschler2021industrial
%A Maschler, Benjamin
%A Kamm, Simon
%A Weyrich, Michael
%D 2021
%J at - Automatisierungstechnik, vol. 69, no. 3, pp. 211-220, 03.2021
%K 2021ias ias
%R 10.1515/auto-2020-0119
%T Deep industrial transfer learning at runtime for image recognition
%U https://www.ias.uni-stuttgart.de/dokumente/publikationen/2021_Deep_industrial_transfer_learning_at_runtime_for_image_recognition.pdf
%X The utilization of deep learning in the field of industrial automation is hindered by two factors: The amount and diversity of training data needed as well as the need to continuously retrain as the use case changes over time. Both problems can be addressed by industrial deep transfer learning allowing for the performant, continuous and potentially distributed training on small, dispersed datasets. As a specific example, a dual memory algorithm for computer vision problems is developed and evaluated. It shows the potential for state-of-the-art performance while being trained only on fractions of the complete ImageNet dataset at multiple locations at once.
@article{maschler2021industrial,
abstract = {The utilization of deep learning in the field of industrial automation is hindered by two factors: The amount and diversity of training data needed as well as the need to continuously retrain as the use case changes over time. Both problems can be addressed by industrial deep transfer learning allowing for the performant, continuous and potentially distributed training on small, dispersed datasets. As a specific example, a dual memory algorithm for computer vision problems is developed and evaluated. It shows the potential for state-of-the-art performance while being trained only on fractions of the complete ImageNet dataset at multiple locations at once.},
added-at = {2021-03-08T16:12:46.000+0100},
author = {Maschler, Benjamin and Kamm, Simon and Weyrich, Michael},
biburl = {https://puma.ub.uni-stuttgart.de/bibtex/2fe94fd09f9ac201c097c28e9c1429be7/taylansngerli},
doi = {10.1515/auto-2020-0119},
interhash = {24872906be9a1725bb657b572d3954df},
intrahash = {fe94fd09f9ac201c097c28e9c1429be7},
issn = {2196-677X},
journal = {at - Automatisierungstechnik, vol. 69, no. 3, pp. 211-220, 03.2021 },
keywords = {2021ias ias},
timestamp = {2021-04-22T11:52:32.000+0200},
title = {Deep industrial transfer learning at runtime for image recognition},
url = {https://www.ias.uni-stuttgart.de/dokumente/publikationen/2021_Deep_industrial_transfer_learning_at_runtime_for_image_recognition.pdf},
year = 2021
}
