{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T09:33:36Z","timestamp":1767951216641,"version":"3.49.0"},"reference-count":32,"publisher":"MDPI AG","issue":"2","license":[{"start":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T00:00:00Z","timestamp":1767916800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"German Ministry of Education and Research","award":["033KI201"],"award-info":[{"award-number":["033KI201"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>The number of industrial processes in which smart devices have been employed rises every day; these devices can be found performing tasks related to the automation, digitization, or optimization of the process. Generally, for these tasks, the devices need to communicate with each other and with a central unit monitored by humans, which is where Industrial Internet of Things (IIoT) comes into play, allowing a network to be built between the devices. Communication might be enough for monitoring purposes, but the optimization and automation of the process are yet to be addressed. In this study, we use an object detection sensor as an initial test subject to explore the Artificial Neural Twin (ANT) as a distributed-process optimization tool in combination with Orthogonal Weight Modification (OWM), a continual learning (CL) method used to augment self-operating devices (i.e., microcontrollers used for machine-vision sensors) with the capacity to learn new tasks autonomously. Some of these devices lack the hardware capacity to run a CL algorithm, which also motivated the comparison of the Fisher matrix, NEig-OWM, and LoRA as matrix approximations to reduce the complexity of the operations between them. Among the compared matrices, we found the Fisher matrix to be the least expensive solution with a negligible reduction in the model\u2019s performance after CL, which makes it a viable solution for large AI models, while NEig-OWM is better suited for smaller models that require fewer hardware resources but more control over the CL algorithm.<\/jats:p>","DOI":"10.3390\/s26020425","type":"journal-article","created":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T08:09:06Z","timestamp":1767946146000},"page":"425","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Comparison of Input-Data Matrix Representations Used for Continual Learning with Orthogonal Weight Modification on Edge Devices"],"prefix":"10.3390","volume":"26","author":[{"ORCID":"https:\/\/orcid.org\/0009-0001-2653-3089","authenticated-orcid":false,"given":"Ronald","family":"Mendez","sequence":"first","affiliation":[{"name":"Fraunhofer IIS, Fraunhofer Institute for Integrated Circuits IIS, Division Development Center X-Ray Technology, Flugplatzstr. 75, 90768 F\u00fcrth, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9550-5284","authenticated-orcid":false,"given":"Andreas","family":"Maier","sequence":"additional","affiliation":[{"name":"Pattern Recognition Lab, Friedich-Alexander Universit\u00e4t Erlangen-N\u00fcrnberg, Schlossplatz 4, 91058 Erlangen, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6920-228X","authenticated-orcid":false,"given":"Johannes","family":"Emmert","sequence":"additional","affiliation":[{"name":"Fraunhofer IIS, Fraunhofer Institute for Integrated Circuits IIS, Division Development Center X-Ray Technology, Flugplatzstr. 75, 90768 F\u00fcrth, Germany"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1968","published-online":{"date-parts":[[2026,1,9]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"2985","DOI":"10.1109\/TII.2020.3023507","article-title":"Challenges and Opportunities in Securing the Industrial Internet of Things","volume":"17","author":"Serror","year":"2021","journal-title":"IEEE Trans. Ind. Inform."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"4724","DOI":"10.1109\/TII.2018.2852491","article-title":"Industrial Internet of Things: Challenges, Opportunities, and Directions","volume":"14","author":"Sisinni","year":"2018","journal-title":"IEEE Trans. Ind. Inform."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"106647","DOI":"10.1016\/j.neunet.2024.106647","article-title":"The Artificial Neural Twin\u2014Process optimization and continual learning in distributed process chains","volume":"180","author":"Emmert","year":"2024","journal-title":"Neural Netw."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1042","DOI":"10.1016\/j.procir.2024.01.141","article-title":"Open Data Sources for Post-Consumer Plastic Sorting: What We Have and What We Still Need","volume":"122","author":"Basedow","year":"2024","journal-title":"Procedia CIRP"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"54","DOI":"10.1016\/j.neunet.2019.01.012","article-title":"Continual lifelong learning with neural networks: A review","volume":"113","author":"Parisi","year":"2019","journal-title":"Neural Netw."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"364","DOI":"10.1038\/s42256-019-0080-x","article-title":"Continual learning of context-dependent processing in neural networks","volume":"1","author":"Guanxiong","year":"2019","journal-title":"Nat. Mach. Intell."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"2881","DOI":"10.1109\/TNSE.2022.3180632","article-title":"Computational Intelligence and Deep Learning for Next-Generation Edge-Enabled Industrial IoT","volume":"10","author":"Tang","year":"2023","journal-title":"IEEE Trans. Netw. Sci. Eng."},{"key":"ref_8","unstructured":"Wu, H., Judd, P., Zhang, X., Isaev, M., and Micikevicius, P. (2020). Integer Quantization for Deep Learning Inference: Principles and Empirical Evaluation. arXiv."},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"2900","DOI":"10.1109\/TPAMI.2023.3334614","article-title":"Structured Pruning for Deep Convolutional Neural Networks: A Survey","volume":"46","author":"He","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_10","first-page":"20366","article-title":"SparCL: Sparse Continual Learning on the Edge","volume":"Volume 35","author":"Koyejo","year":"2022","journal-title":"Advances in Neural Information Processing Systems, Proceedings of the 36th Conference on Neural Information Processing Systems, New Orleans, LA, USA, 28 November\u20139 December 2022"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Pellegrini, L., Graffieti, G., Lomonaco, V., and Maltoni, D. (2020, January 25\u201329). Latent Replay for Real-Time Continual Learning. Proceedings of the 2020 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS), Las Vegas, NV, USA.","DOI":"10.1109\/IROS45743.2020.9341460"},{"key":"ref_12","unstructured":"Wang, Z., Wu, F., Yu, F., Zhou, Y., Hu, J., and Min, G. (2024). Federated Continual Learning for Edge-AI: A Comprehensive Survey. arXiv."},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Maltoni, D., and Lomonaco, V. (2019). Continuous Learning in Single-Incremental-Task Scenarios. arXiv.","DOI":"10.1016\/j.neunet.2019.03.010"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"3521","DOI":"10.1073\/pnas.1611835114","article-title":"Overcoming catastrophic forgetting in neural networks","volume":"114","author":"Kirkpatrick","year":"2017","journal-title":"Proc. Natl. Acad. Sci. USA"},{"key":"ref_15","first-page":"3","article-title":"LoRA: Low-Rank Adaptation of Large Language Models","volume":"1","author":"Hu","year":"2021","journal-title":"ICLR"},{"key":"ref_16","unstructured":"Ding, C., Cao, X., Xie, J., Fan, L., Wang, S., and Lu, Z. (2024). LoRA-C: Parameter-Efficient Fine-Tuning of Robust CNN for IoT Devices. arXiv."},{"key":"ref_17","unstructured":"Mendez, R., Maier, A., and Emmert, J. (2025, January 26\u201327). Towards continual learning with the artificial neural twin applied to recycling processes. Proceedings of the OCM 2025-7th International Conference on Optical Characterization of Materials, Karlsruhe, Germany."},{"key":"ref_18","unstructured":"Jung, D., Lee, D., Hong, S., Jang, H., Bae, H., and Yoon, S. (2023, January 1\u20135). New Insights for the Stability-Plasticity Dilemma in Online Continual Learning. Proceedings of the Eleventh International Conference on Learning Representations, Kigali, Rwanda."},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Verma, T., Jin, L., Zhou, J., Huang, J., Tan, M., Choong, B.C.M., Tan, T.F., Gao, F., Xu, X., and Ting, D.S. (2023). Privacy-preserving continual learning methods for medical image classification: A comparative analysis. Front. Med., 10.","DOI":"10.3389\/fmed.2023.1227515"},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"3046","DOI":"10.1016\/j.sigpro.2010.05.009","article-title":"Theoretical convergence analysis of FxLMS algorithm","volume":"90","author":"Ardekani","year":"2010","journal-title":"Signal Process."},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"5362","DOI":"10.1109\/TPAMI.2024.3367329","article-title":"A Comprehensive Survey of Continual Learning: Theory, Method and Application","volume":"46","author":"Wang","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_22","first-page":"146:1","article-title":"New Insights and Perspectives on the Natural Gradient Method","volume":"21","author":"Martens","year":"2014","journal-title":"J. Mach. Learn. Res."},{"key":"ref_23","first-page":"5391","article-title":"An Efficient Fisher Matrix Approximation Method for Large-Scale Neural Network Optimization","volume":"45","author":"Yang","year":"2023","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"16196","DOI":"10.1109\/TNNLS.2023.3292359","article-title":"Overcoming Catastrophic Forgetting in Continual Learning by Exploring Eigenvalues of Hessian Matrix","volume":"35","author":"Kong","year":"2024","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"2212","DOI":"10.1080\/03081087.2016.1267104","article-title":"Literature survey on low rank approximation of matrices","volume":"65","author":"Kumar","year":"2017","journal-title":"Linear Multilinear Algebra"},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Ye, J. (2004, January 4\u20138). Generalized low rank approximations of matrices. Proceedings of the Twenty-First International Conference on Machine Learning, Banff, AB, Canada.","DOI":"10.1145\/1015330.1015347"},{"key":"ref_27","unstructured":"Lu, A., Yuan, H., Feng, T., and Sun, Y. (2025). Rethinking the Stability-Plasticity Trade-off in Continual Learning from an Architectural Perspective. arXiv."},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"127468","DOI":"10.1016\/j.eswa.2025.127468","article-title":"The continuous learning algorithm with null space and eigenface-based orthogonal weight modification","volume":"281","author":"Liao","year":"2025","journal-title":"Expert Syst. Appl."},{"key":"ref_29","first-page":"12","article-title":"PyTorch: An Imperative Style, High-Performance Deep Learning Library","volume":"32","author":"Paszke","year":"2019","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref_30","unstructured":"Juliani, A., Berges, V.P., Teng, E., Cohen, A., Harper, J., Elion, C., Goy, C., Gao, Y., Henry, H., and Mattar, M. (2020). Unity: A General Platform for Intelligent Agents. arXiv."},{"key":"ref_31","unstructured":"Simonyan, K., and Zisserman, A. (2015). Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"93","DOI":"10.1007\/s11760-014-0706-8","article-title":"Combination of high-level features with low-level features for detection of pedestrian","volume":"10","author":"Takarli","year":"2016","journal-title":"Signal Image Video Process."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/2\/425\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T08:11:29Z","timestamp":1767946289000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/2\/425"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,1,9]]},"references-count":32,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2026,1]]}},"alternative-id":["s26020425"],"URL":"https:\/\/doi.org\/10.3390\/s26020425","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,1,9]]}}}