Deep learning methods for underwater target feature extraction and recognition
Open accessAbstract
The classification and recognition technology of underwater acoustic signal were always an important research content in the field of underwater acoustic signal processing. Currently, wavelet transform, Hilbert-Huang transform, and Mel frequency cepstral coefficients are used as a method of underwater acoustic signal feature extraction. In this paper, a method for feature extraction and identification of underwater noise data based on CNN and ELM is proposed. An automatic feature extraction method of underwater acoustic signals is proposed using depth convolution network. An underwater target recognition classifier is based on extreme learning machine. Although convolution neural networks can execute both feature extraction and classification, their function mainly relies on a full connection layer, which is trained by gradient descent-based; the generalization ability is limited and suboptimal, so an extreme learning machine (ELM) was used in classification stage. Firstly, CNN learns deep and robust features, followed by the removing of the fully connected layers. Then ELM fed with the CNN features is used as the classifier to conduct an excellent classification. Experiments on the actual data set of civil ships obtained 93.04% recognition rate; compared to the traditional Mel frequency cepstral coefficients and Hilbert-Huang feature, recognition rate greatly improved.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The CNN+ELM model represented an improvement of at least 6.79 percentage points compared with other approaches that used classical features as inputs and fully connected layers as classifiers.
“representing an improvement of, at least, 6.79 percentage points when compared with other approaches that used classical features as inputs and fully connected layers as classifiers.”
From A Survey of Underwater Acoustic Data Classification Methods Using Deep Learning for Shoreline Surveillance (2022) · cites this paper as [91]
Section 5.2 reports that CNN-extracted features classified by ELM achieved 93.04% recognition and performed 5%–10% better than traditional MFCC and Hilbert–Huang features. It does not establish a minimum improvement of 6.79 percentage points. It also says the traditional-feature comparisons in Table 4 used ELM classifiers, not fully connected layers. (LLM, conf=0.88)
Experimental — accuracy not yet measuredA particular UATC-DenseNet architecture (six convolutional filters with kernel size 1 × 15), using original sound data without preprocessing layers or filters, outperformed other classical methods such as CNN-extreme learning machine, ResNet18, and SqueezeNet.
“The results reported show that a particular model architecture (with six convolutional filters with a kernel size of 1 × 15), using the original sound data as input without preprocessing layers or filters, outperformed other classical methods, such as CNN-extreme learning machine [91], ResNet18 [58] and SqueezeNet [97].”
From A Survey of Underwater Acoustic Data Classification Methods Using Deep Learning for Shoreline Surveillance (2022) · cites this paper as [91]
Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
Experimental — accuracy not yet measuredThe features generated by the one-dimensional convolutional layers acted as band-pass filters, similarly to the ADCNN model.
“The final analysis showed that the features generated by the one-dimensional convolutional layers acted as band-pass filters, similarly to the ADCNN model.”
From A Survey of Underwater Acoustic Data Classification Methods Using Deep Learning for Shoreline Surveillance (2022) · cites this paper as [91]
Section 5.2 (Experiment Analysis; page unavailable) says the learned kernels in the first convolutional layer act as band-pass filters. Section 3.2 describes the input as a one-dimensional sound signal. The paper does not make the comparison with an ADCNN model, and it identifies the kernels—not the generated features—as the filters. (LLM, conf=0.85)
Experimental — accuracy not yet measuredThe CNN+ELM model achieved a 93.04% recognition rate on a dataset of civil ships.
“Results presented in [91] show that the resulting CNN+ELM model was able to achieve a 93.04% recognition rate on a dataset of civil ships”
From A Survey of Underwater Acoustic Data Classification Methods Using Deep Learning for Shoreline Surveillance (2022) · cites this paper as [91]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Backpropagation Applied to Handwritten Zip Code Recognition
LeCun, Y., Boser, B., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W., & Jackel, L. D. (1989) · Neural Computation · doi:10.1162/neco.1989.1.4.541 · Cited by 11,891 · see citing works
Hinton, G., Deng, L., Yu, D., Dahl, G., Mohamed, A.-R., Jaitly, N., Senior, A., Vanhoucke, V., Nguyen, P., Sainath, T., et al. (2012) · IEEE Signal Processing Magazine · doi:10.1109/msp.2012.2205597 · Cited by 10,260 · see citing works
Learning a better representation of speech soundwaves using restricted boltzmann machines
Jaitly, N., & Hinton, G. (2011) · 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2011.5947700 · Cited by 231 · cited 1× · see citing works
“In 2012, four major international scientific research institutions summed up the progress made by deep learning in sound recognition task [15] and pointed out in the paper that the experiments show the effect of deep neural network is better than the one of traditional Gaussian mixed model.”
Background
Flagged first, then by citations
Research works printed but not linked: 15
- [1]LeCun Y. Bengio Y. Hinton G. Deep learning Nature 2015 521 7553 436 444 10.1038/nature14539 26017442 · DOI · Paper page
- [2]Mohamed A.-R. Dahl G. E. Hinton G. Acoustic modeling using deep belief networks IEEE Transactions on Audio, Speech and Language Processing 2012 20 1 14 22 10.1109/tasl.2011.2109382 2-s2.0-84055211743 · DOI
- [3]Chen J. Wang Y. Wang D. A feature study for classification-based speech separation at low signal-to-noise ratios IEEE/ACM Transactions on Audio Speech and Language Processing 2014 22 12 1993 2002 2-s2.0-84921769616 10.1109/TASLP.2014.2359159 PMC4293540 25599083 · DOI
- [4]Jiang Y. Wang D. L. Liu R. S. Feng Z. M. Binaural classification for reverberant speech segregation using deep neural networks IEEE/ACM Transactions on Audio Speech and Language Processing 2014 22 12 2112 2121 2-s2.0-84921817110 10.1109/TASLP.2014.2361023 · DOI
- [5]Cortes C. Vapnik V. Support-vector networks Machine Learning 1995 20 3 273 297 2-s2.0-34249753618 10.1007/BF00994018 Zbl0831.68098 · DOI · Paper page
- [6]Peterson L. E. K-nearest neighbor Scholarpedia 2009 4 2, article 1883 10.4249/scholarpedia.1883 · DOI
- [7]Huang N. E. Shen S. P. Hilbert-Huang Transform and Its Applications 2005 Singapore World Scientific MR2188500 Zbl1087.62500
- [9]Zeng Q.-J. Wang F. Huang G.-J. Technique of passive sonar target recognition based on continuous spectrum feature extraction Journal of Shanghai Jiaotong University 2002 36 3 382 386 2-s2.0-0036520195
- [12]Li X.-K. Xie L. Qin Y. Underwater target feature extraction using Hilbert-Huang transform Journal of Harbin Engineering University 2009 30 5 542 546 10.3969/j.issn.1006-7043.2009.05.014 2-s2.0-67649451980 · DOI
- [13]Hinton G. E. Salakhutdinov R. R. Reducing the dimensionality of data with neural networks American Association for the Advancement of Science: Science 2006 313 5786 504 507 MR2242509 10.1126/science.1127647 Zbl1226.68083 2-s2.0-33746600649 16873662 · DOI · Paper page
- [14]Lee H. Yan L. Pham P. Ng A. Y. Unsupervised feature learning for audio classification using convolutional deep belief networks 9 Proceedings of the 23rd Annual Conference on Neural Information Processing Systems (NIPS '09) December 2009 1096 1104 2-s2.0-84863380535
- [17]Palaz D. Collobert R. Magimai-Doss M. Estimating phoneme class conditional probabilities from raw speech signal using convolutional neural networks Proceedings of the 14th Annual Conference of the International Speech Communication Association, INTERSPEECH 2013 August 2013 1766 1770 2-s2.0-84906273908
- [19]Abdel-Hamid O. Mohamed A.-R. Jiang H. Deng L. Penn G. Yu D. Convolutional neural networks for speech recognition IEEE Transactions on Audio, Speech and Language Processing 2014 22 10 1533 1545 2-s2.0-84911473441 10.1109/TASLP.2014.2339736 · DOI
- [20]Huang G. B. Zhu Q. Y. Siew C. K. Extreme learning machine: a new learning scheme of feedforward neural networks 2 Proceedings of the IEEE International Joint Conference on Neural Networks July 2004 985 990 10.1109/IJCNN.2004.1380068 2-s2.0-10944272650 · DOI
- [21]Huang G. Huang G.-B. Song S. You K. Trends in extreme learning machines: A review Neural Networks 2015 61 32 48 2-s2.0-84908682236 10.1016/j.neunet.2014.10.001 Zbl1325.68190 25462632 · DOI
Cited by
Sorted by citations
Domingos, L. C. F., Santos, P. E., Skelton, P. S. M., Brinkworth, R. S. A., & Sammut, K. (2022) · Sensors · doi:10.3390/s22062181 · Cited by 93 · see citing works
“Results presented in [91] show that the resulting CNN+ELM model was able to achieve a 93.”
BackgroundMethodDeep convolution stack for waveform in underwater acoustic target recognition
Tian, S., Chen, D., Wang, H., & Liu, J. (2021) · Scientific Reports · doi:10.1038/s41598-021-88799-z · Cited by 86 · see citing works
Ship Type Classification by Convolutional Neural Networks with Auditory-Like Mechanisms
Shen, S., Yang, H., Yao, X., Li, J., Xu, G., & Sheng, M. (2020) · Sensors · doi:10.3390/s20010253 · Cited by 57 · see citing works
Domingos, L. C. F., Santos, P. E., Skelton, P. S. M., Brinkworth, R. S. A., & Sammut, K. (2022) · IEEE Access · doi:10.1109/access.2022.3220265 · Cited by 37 · see citing works
- Authors
- Hu, G., Wang, K., Peng, Y., Qiu, M., Shi, J., & Liu, L.
- Affiliations
- Harbin Engineering University (CN)
- 760 Research Institute of China Shipbuilding Industry, Liaoning, Anshan, China
- Corresponding author
- Wang, Kejun
- Venue
- Computational Intelligence and Neuroscience
- Publisher
- Wiley
- Published
- 2018
- Volume
- 2018
- Pages
- 1-10
- ISSN
- 1687-5265 (ISSN-L)
- 1687-5273
- PMC
- PMC5892262
- OpenAlex
- W2794574871
- Cited by
- 151
- References
- 3
- Citations by year
- 2026: 9 · 2025: 14 · 2024: 29 · 2023: 20 · 2022: 25 · 2021: 23 · 2020: 18 · 2019: 11 · 2018: 2
- FWCI
- 8.68
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- Computational Intelligence and Neuroscience (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Underwater Measurement and Control Foundation of Key Laboratory (9140C260505150C26115)
- Topics
- Physical Sciences › Earth and Planetary Sciences › Oceanography › Underwater Acoustics Research
- Physical Sciences › Computer Science › Signal Processing › Music and Audio Processing
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and ELM
- Keywords
- underwater acoustic signals, underwater target recognition, feature extraction, convolutional neural networks, extreme learning machine, Mel-frequency cepstral coefficients, Hilbert-Huang transform, wavelet transform
- MeSH terms
- Machine Learning, Acoustics, Pattern Recognition, Automated, Signal Processing, Computer-Assisted, Water, Neural Networks, Computer
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
