Paper Info
Title
Label Noise Modeling and Correction via Loss Curve Fitting for SAR ATR
Abstract
The success of deep learning in synthetic aperture radar (SAR) automatic target recognition (ATR) relies on a large number of labeled samples; however, there are often wrong (noisy) labels in a large-scale dataset. In this article, we propose a loss curve-fitting-based method, which can identify the noisy labels and train the classification network effectively. We propose to model label noise by unsupervised clustering via fitting loss curve to identify whether the sample's label is clean or noisy. Then, we train the network using augmented samples with clean labels to correct noisy labels further. The experiments on the moving and stationary target acquisition and recognition (MSTAR) dataset prove that our proposed method can deal with the situation when training a network with different ratios of noisy labels and correct noisy labels effectively. When the noise ratio is small (40%) in the training dataset, our method can correct 97.9% of noisy labels and train the classification network with 98.8% classification accuracy. While the noise ratio is large (80%), our method can correct 78.1% of noisy labels and train the classification network with 79.6% classification accuracy.
Year
DOI
Venue
2022
10.1109/TGRS.2021.3121397
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
Keywords
DocType
Volume
Noise measurement, Training, Synthetic aperture radar, Radar polarimetry, Deep learning, Convolutional neural networks, Prototypes, Deep learning, label noise correction, label noise modeling, noisy labels, synthetic aperture radar (SAR)
Journal
60
ISSN
Citations 
PageRank 
0196-2892
0
0.34
References 
Authors
0
9
Name
Order
Citations
PageRank
Chen Wang139.53
Jun Shi2111.52
Yuanyuan Zhou305.07
Li Liang41417.68
Xiaqing Yang504.73
Tianwen Zhang6115.94
Shunjun Wei7317.63
Xiaoling Zhang8512.90
Chongben Tao900.34