Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet

Shelly Vishwakarma; Wenda Li; Raviraj Adve; Kevin Chetty

doi:10.1049/icp.2022.2314

Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet

Shelly Vishwakarma, Wenda Li, Raviraj Adve, Kevin Chetty

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

4 Citations (Scopus)

Abstract

This work introduces an attention mechanism that can be integrated into any standard convolution neural network (CNN) to improve model sensitivity and prediction accuracy with minimal computational overhead. We introduce the attention mechanism in a lightweight network-Alexnet and evaluate its classification performance for human micro-Doppler signatures. We show that the Alexnet model trained with an attention module can implicitly learn to highlight the salient regions in the radar signatures whilst suppressing the irrelevant background regions and consistently improve the network predictions by more than 4% in most cases. We further provide network visualizations through class activation mapping, providing better insights into how the predictions are made.

Original language	English
Title of host publication	IET Conference Proceedings
Publisher	Institution of Engineering and Technology
Pages	190 - 195
Number of pages	6
ISBN (Electronic)	9781839537776
DOIs	https://doi.org/10.1049/icp.2022.2314
Publication status	Published - 2 Mar 2023
Event	International Conference on Radar Systems 2022 - Edinburgh, United Kingdom Duration: 24 Oct 2022 → 27 Oct 2022 https://www.aconf.org/conf_181833.2022_International_Radar_Conference.html

Publication series

Name	IET Conference Proceedings
Publisher	The Institution of Engineering and Technology
Number	17
Volume	2022
ISSN (Print)	2732-4494

Conference

Conference	International Conference on Radar Systems 2022
Abbreviated title	RADAR 2022
Country/Territory	United Kingdom
City	Edinburgh
Period	24/10/22 → 27/10/22
Internet address	https://www.aconf.org/conf_181833.2022_International_Radar_Conference.html

Keywords

convolutional neural nets
doppler radar
learning (artificial intelligence)
radar computing
radar signal processing
Alexnet model
attention enhanced Alexnet
attention mechanism
attention module
classification performance
lightweight network-Alexnet
minimal computational overhead
model sensitivity
network predictions
network visualizations
prediction accuracy
radar microDoppler signatures
radar signatures
salient features
salient regions
standard convolution neural network

Access to Document

10.1049/icp.2022.2314

Cite this

@inproceedings{f6d7055e508a45afbab04f0c46b088d3,

title = "Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet",

abstract = "This work introduces an attention mechanism that can be integrated into any standard convolution neural network (CNN) to improve model sensitivity and prediction accuracy with minimal computational overhead. We introduce the attention mechanism in a lightweight network-Alexnet and evaluate its classification performance for human micro-Doppler signatures. We show that the Alexnet model trained with an attention module can implicitly learn to highlight the salient regions in the radar signatures whilst suppressing the irrelevant background regions and consistently improve the network predictions by more than 4\% in most cases. We further provide network visualizations through class activation mapping, providing better insights into how the predictions are made.",

keywords = "convolutional neural nets, doppler radar, learning (artificial intelligence), radar computing, radar signal processing, Alexnet model, attention enhanced Alexnet, attention mechanism, attention module, classification performance, lightweight network-Alexnet, minimal computational overhead, model sensitivity, network predictions, network visualizations, prediction accuracy, radar microDoppler signatures, radar signatures, salient features, salient regions, standard convolution neural network",

author = "Shelly Vishwakarma and Wenda Li and Raviraj Adve and Kevin Chetty",

year = "2023",

month = mar,

day = "2",

doi = "10.1049/icp.2022.2314",

language = "English",

series = "IET Conference Proceedings",

publisher = "Institution of Engineering and Technology",

number = "17",

pages = "190 -- 195",

booktitle = "IET Conference Proceedings",

address = "United Kingdom",

note = "International Conference on Radar Systems 2022, RADAR 2022 ; Conference date: 24-10-2022 Through 27-10-2022",

url = "https://www.aconf.org/conf\_181833.2022\_International\_Radar\_Conference.html",

}

Vishwakarma, S, Li, W, Adve, R & Chetty, K 2023, Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet. in IET Conference Proceedings. IET Conference Proceedings, no. 17, vol. 2022, Institution of Engineering and Technology, pp. 190 - 195, International Conference on Radar Systems 2022, Edinburgh, United Kingdom, 24/10/22. https://doi.org/10.1049/icp.2022.2314

Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet. / Vishwakarma, Shelly; Li, Wenda; Adve, Raviraj et al.
IET Conference Proceedings. Institution of Engineering and Technology, 2023. p. 190 - 195 (IET Conference Proceedings; Vol. 2022, No. 17).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet

AU - Vishwakarma, Shelly

AU - Li, Wenda

AU - Adve, Raviraj

AU - Chetty, Kevin

PY - 2023/3/2

Y1 - 2023/3/2

N2 - This work introduces an attention mechanism that can be integrated into any standard convolution neural network (CNN) to improve model sensitivity and prediction accuracy with minimal computational overhead. We introduce the attention mechanism in a lightweight network-Alexnet and evaluate its classification performance for human micro-Doppler signatures. We show that the Alexnet model trained with an attention module can implicitly learn to highlight the salient regions in the radar signatures whilst suppressing the irrelevant background regions and consistently improve the network predictions by more than 4% in most cases. We further provide network visualizations through class activation mapping, providing better insights into how the predictions are made.

AB - This work introduces an attention mechanism that can be integrated into any standard convolution neural network (CNN) to improve model sensitivity and prediction accuracy with minimal computational overhead. We introduce the attention mechanism in a lightweight network-Alexnet and evaluate its classification performance for human micro-Doppler signatures. We show that the Alexnet model trained with an attention module can implicitly learn to highlight the salient regions in the radar signatures whilst suppressing the irrelevant background regions and consistently improve the network predictions by more than 4% in most cases. We further provide network visualizations through class activation mapping, providing better insights into how the predictions are made.

KW - convolutional neural nets

KW - doppler radar

KW - learning (artificial intelligence)

KW - radar computing

KW - radar signal processing

KW - Alexnet model

KW - attention enhanced Alexnet

KW - attention mechanism

KW - attention module

KW - classification performance

KW - lightweight network-Alexnet

KW - minimal computational overhead

KW - model sensitivity

KW - network predictions

KW - network visualizations

KW - prediction accuracy

KW - radar microDoppler signatures

KW - radar signatures

KW - salient features

KW - salient regions

KW - standard convolution neural network

UR - https://www.scopus.com/pages/publications/85170518158

U2 - 10.1049/icp.2022.2314

DO - 10.1049/icp.2022.2314

M3 - Conference contribution

T3 - IET Conference Proceedings

SP - 190

EP - 195

BT - IET Conference Proceedings

PB - Institution of Engineering and Technology

T2 - International Conference on Radar Systems 2022

Y2 - 24 October 2022 through 27 October 2022

ER -

Learning salient features in radar micro-Doppler signatures using Attention Enhanced Alexnet

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this