You can have your ensemble and run it too - Deep Ensembles Spread Over Time
Meding, Isak ; Bodin, Alexander ; Tonderski, Adam LU
; Johnander, Joakim
; Petersson, Christoffer
and Svensson, Lennart
(2023)
2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023
In Proceedings - 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023
p.4022-4031
- Abstract
Ensembles of independently trained deep neural networks yield uncertainty estimates that rival Bayesian networks in performance. They also offer sizable improvements in terms of predictive performance over single models. However, deep ensembles are not commonly used in environments with limited computational budget - such as autonomous driving - since the complexity grows linearly with the number of ensemble members. An important observation that can be made for robotics applications, such as autonomous driving, is that data is typically sequential. For instance, when an object is to be recognized, an autonomous vehicle typically observes a sequence of images, rather than a single image. This raises the question, could the deep ensemble... (More)
Ensembles of independently trained deep neural networks yield uncertainty estimates that rival Bayesian networks in performance. They also offer sizable improvements in terms of predictive performance over single models. However, deep ensembles are not commonly used in environments with limited computational budget - such as autonomous driving - since the complexity grows linearly with the number of ensemble members. An important observation that can be made for robotics applications, such as autonomous driving, is that data is typically sequential. For instance, when an object is to be recognized, an autonomous vehicle typically observes a sequence of images, rather than a single image. This raises the question, could the deep ensemble be spread over time?In this work, we propose and analyze Deep Ensembles Spread Over Time (DESOT). The idea is to apply only a single ensemble member to each data point in the sequence, and fuse the predictions over a sequence of data points. We implement and experiment with DESOT for traffic sign classification, where sequences of tracked image patches are to be classified. We find that DESOT obtains the benefits of deep ensembles, in terms of predictive and uncertainty estimation performance, while avoiding the added computational cost. Moreover, DESOT is simple to implement and does not require sequences during training. Finally, we find that DESOT, like deep ensembles, outperform single models for out-of-distribution detection.
(Less)
- author
- Meding, Isak
; Bodin, Alexander
; Tonderski, Adam
LU
; Johnander, Joakim
; Petersson, Christoffer
and Svensson, Lennart
- organization
- publishing date
- 2023
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- ensemble, out of distribution detection, traffic sign recognition, uncertainty estimation
- host publication
- Proceedings - 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023
- series title
- Proceedings - 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023
- pages
- 10 pages
- publisher
- IEEE - Institute of Electrical and Electronics Engineers Inc.
- conference name
- 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023
- conference location
- Paris, France
- conference dates
- 2023-10-02 - 2023-10-06
- external identifiers
-
- scopus:85182919373
- ISBN
- 9798350307443
- DOI
- 10.1109/ICCVW60793.2023.00434
- language
- English
- LU publication?
- yes
- id
- 9bf0dbfe-e6ab-4b88-81de-42478b0d290c
- date added to LUP
- 2024-02-16 14:50:51
- date last changed
- 2024-02-16 14:53:07
@inproceedings{9bf0dbfe-e6ab-4b88-81de-42478b0d290c,
abstract = {{<p>Ensembles of independently trained deep neural networks yield uncertainty estimates that rival Bayesian networks in performance. They also offer sizable improvements in terms of predictive performance over single models. However, deep ensembles are not commonly used in environments with limited computational budget - such as autonomous driving - since the complexity grows linearly with the number of ensemble members. An important observation that can be made for robotics applications, such as autonomous driving, is that data is typically sequential. For instance, when an object is to be recognized, an autonomous vehicle typically observes a sequence of images, rather than a single image. This raises the question, could the deep ensemble be spread over time?In this work, we propose and analyze Deep Ensembles Spread Over Time (DESOT). The idea is to apply only a single ensemble member to each data point in the sequence, and fuse the predictions over a sequence of data points. We implement and experiment with DESOT for traffic sign classification, where sequences of tracked image patches are to be classified. We find that DESOT obtains the benefits of deep ensembles, in terms of predictive and uncertainty estimation performance, while avoiding the added computational cost. Moreover, DESOT is simple to implement and does not require sequences during training. Finally, we find that DESOT, like deep ensembles, outperform single models for out-of-distribution detection.</p>}},
author = {{Meding, Isak and Bodin, Alexander and Tonderski, Adam and Johnander, Joakim and Petersson, Christoffer and Svensson, Lennart}},
booktitle = {{Proceedings - 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023}},
isbn = {{9798350307443}},
keywords = {{ensemble; out of distribution detection; traffic sign recognition; uncertainty estimation}},
language = {{eng}},
pages = {{4022--4031}},
publisher = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}},
series = {{Proceedings - 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023}},
title = {{You can have your ensemble and run it too - Deep Ensembles Spread Over Time}},
url = {{http://dx.doi.org/10.1109/ICCVW60793.2023.00434}},
doi = {{10.1109/ICCVW60793.2023.00434}},
year = {{2023}},
}