Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition
Spiking neural networks have significant potential utility in robotics due to their high energy efficiency on specialized hardware, but proof-of-concept implementations have not yet typically achieved competitive performance or capability with conventional approaches. In this paper, we tackle one of...
Published in: | 2023 IEEE International Conference on Robotics and Automation (ICRA) |
---|---|
Main Authors: | , , |
Format: | Book Part |
Language: | unknown |
Published: |
Institute of Electrical and Electronics Engineers Inc.
2023
|
Subjects: | |
Online Access: | https://eprints.qut.edu.au/243068/ |
id |
ftqueensland:oai:eprints.qut.edu.au:243068 |
---|---|
record_format |
openpolar |
spelling |
ftqueensland:oai:eprints.qut.edu.au:243068 2024-04-28T08:29:09+00:00 Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition Hussaini, Somayeh Milford, Michael Fischer, Tobias 2023 application/pdf https://eprints.qut.edu.au/243068/ unknown Institute of Electrical and Electronics Engineers Inc. https://eprints.qut.edu.au/243068/1/144266682.pdf doi:10.1109/ICRA48891.2023.10160749 Hussaini, Somayeh, Milford, Michael, & Fischer, Tobias (2023) Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition. In Proceedings of the 2023 IEEE International Conference on Robotics and Automation (ICRA). Institute of Electrical and Electronics Engineers Inc., United States of America, pp. 4200-4207. http://purl.org/au-research/grants/arc/FL210100156 https://eprints.qut.edu.au/243068/ Centre for Robotics; Faculty of Engineering; School of Electrical Engineering & Robotics free_to_read http://creativecommons.org/licenses/by-nc/4.0/ 2023 IEEE. © 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Proceedings of the 2023 IEEE International Conference on Robotics and Automation (ICRA) Chapter in Book, Report or Conference volume 2023 ftqueensland https://doi.org/10.1109/ICRA48891.2023.10160749 2024-04-10T00:23:49Z Spiking neural networks have significant potential utility in robotics due to their high energy efficiency on specialized hardware, but proof-of-concept implementations have not yet typically achieved competitive performance or capability with conventional approaches. In this paper, we tackle one of the key practical challenges of scalability by introducing a novel modular ensemble network approach, where compact, localized spiking networks each learn and are solely responsible for recognizing places in a local region of the environment only. This modular approach creates a highly scalable system. However, it comes with a high-performance cost where a lack of global regularization at deployment time leads to hyperactive neurons that erroneously respond to places outside their learned region. Our second contribution introduces a regularization approach that detects and removes these problematic hyperactive neurons during the initial environmental learning phase. We evaluate this new scalable modular system on benchmark localization datasets Nordland and Oxford RobotCar, with comparisons to standard techniques NetVLAD, DenseVLAD, and SAD, and a previous spiking neural network system. Our system substantially outperforms the previous SNN system on its small dataset, but also maintains performance on 27 times larger benchmark datasets where the operation of the previous system is computationally infeasible, and performs competitively with the conventional localization systems. Book Part Nordland Nordland Nordland Queensland University of Technology: QUT ePrints 2023 IEEE International Conference on Robotics and Automation (ICRA) 4200 4207 |
institution |
Open Polar |
collection |
Queensland University of Technology: QUT ePrints |
op_collection_id |
ftqueensland |
language |
unknown |
description |
Spiking neural networks have significant potential utility in robotics due to their high energy efficiency on specialized hardware, but proof-of-concept implementations have not yet typically achieved competitive performance or capability with conventional approaches. In this paper, we tackle one of the key practical challenges of scalability by introducing a novel modular ensemble network approach, where compact, localized spiking networks each learn and are solely responsible for recognizing places in a local region of the environment only. This modular approach creates a highly scalable system. However, it comes with a high-performance cost where a lack of global regularization at deployment time leads to hyperactive neurons that erroneously respond to places outside their learned region. Our second contribution introduces a regularization approach that detects and removes these problematic hyperactive neurons during the initial environmental learning phase. We evaluate this new scalable modular system on benchmark localization datasets Nordland and Oxford RobotCar, with comparisons to standard techniques NetVLAD, DenseVLAD, and SAD, and a previous spiking neural network system. Our system substantially outperforms the previous SNN system on its small dataset, but also maintains performance on 27 times larger benchmark datasets where the operation of the previous system is computationally infeasible, and performs competitively with the conventional localization systems. |
format |
Book Part |
author |
Hussaini, Somayeh Milford, Michael Fischer, Tobias |
spellingShingle |
Hussaini, Somayeh Milford, Michael Fischer, Tobias Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition |
author_facet |
Hussaini, Somayeh Milford, Michael Fischer, Tobias |
author_sort |
Hussaini, Somayeh |
title |
Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition |
title_short |
Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition |
title_full |
Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition |
title_fullStr |
Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition |
title_full_unstemmed |
Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition |
title_sort |
ensembles of compact, region-specific and regularized spiking neural networks for scalable place recognition |
publisher |
Institute of Electrical and Electronics Engineers Inc. |
publishDate |
2023 |
url |
https://eprints.qut.edu.au/243068/ |
genre |
Nordland Nordland Nordland |
genre_facet |
Nordland Nordland Nordland |
op_source |
Proceedings of the 2023 IEEE International Conference on Robotics and Automation (ICRA) |
op_relation |
https://eprints.qut.edu.au/243068/1/144266682.pdf doi:10.1109/ICRA48891.2023.10160749 Hussaini, Somayeh, Milford, Michael, & Fischer, Tobias (2023) Ensembles of Compact, Region-specific and Regularized Spiking Neural Networks for Scalable Place Recognition. In Proceedings of the 2023 IEEE International Conference on Robotics and Automation (ICRA). Institute of Electrical and Electronics Engineers Inc., United States of America, pp. 4200-4207. http://purl.org/au-research/grants/arc/FL210100156 https://eprints.qut.edu.au/243068/ Centre for Robotics; Faculty of Engineering; School of Electrical Engineering & Robotics |
op_rights |
free_to_read http://creativecommons.org/licenses/by-nc/4.0/ 2023 IEEE. © 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. |
op_doi |
https://doi.org/10.1109/ICRA48891.2023.10160749 |
container_title |
2023 IEEE International Conference on Robotics and Automation (ICRA) |
container_start_page |
4200 |
op_container_end_page |
4207 |
_version_ |
1797587484640542720 |