Four data-driven methods—random forest (RF), support vector machine (SVM), feed-forward neural network (FNN), and convolutional neural network (CNN)—are applied to discriminate surface and underwater vessels in the ocean using low-frequency acoustic pressure data. Acoustic data are modeled considering a vertical line array by a Monte Carlo simulation using the underwater acoustic propagation model, KRAKEN, in the ocean environment of East Sea in Korea. The raw data are preprocessed and reorganized into the phone-space cross-spectral density matrix (pCSDM) and mode-space cross-spectral density matrix (mCSDM). Two additional matrices are generated using the absolute values of matrix elements in each CSDM. Each of these four matrices is used as input data for supervised machine learning. Binary classification is performed by using RF, SVM, FNN, and CNN, and the obtained results are compared. All machine-learning algorithms show an accuracy of >95% for three types of input data—the pCSDM, mCSDM, and mCSDM with the absolute matrix elements. The CNN is the best in terms of low percent error. In particular, the result using the complex pCSDM is encouraging because these data-driven methods inherently do not require environmental information. This work demonstrates the potential of machine learning to discriminate between surface and underwater vessels in the ocean.

Underwater acoustics is the study of all phenomena related to the occurrence, propagation, and reception of sound waves in the water medium. Because electromagnetic waves undergo a significant attenuation in water, sound waves, which have relatively low propagation loss and high propagation speed, are used for underwater communication and detection. In the field of underwater acoustics, studies are mainly conducted on underwater communications, underwater target detection, marine resources, and measurement and analysis of underwater sound sources. Most applications for underwater acoustics can be described as remote sensing. Remote sensing is employed when an object, condition, or phenomenon of interest cannot be directly observed and information about the target of interest is acquired indirectly using data. In underwater acoustics, this can be described simply as a sound navigation and ranging (sonar) system. Sonar systems can be broadly classified into passive and active systems. Passive sonar systems acquire information by using sensors to measure the acoustic energy (signal) emitted by the target of interest. In active sonar systems, the observer obtains information by directly emitting an acoustic pulse and gathering the returning signals that are reflected by the target. Machine learning, which is widely known today, was initially used in academia for developing artificial intelligence. Recently, the use of machine learning has become widespread owing to the introduction of high-speed parallel computing that uses graphics processing units (GPUs) and can perform reliable learning based on big data, as well as develop various machine learning techniques that can find optimal solutions. Machine learning has contributed to the evolution of acoustic signal processing and voice recognition, and it is also utilized in various ways in the field of underwater acoustics. It is used for traditional remote sensing, such as in detection/classification of underwater sound sources and targets and localization. In addition, it is being used in the field of acoustic signal processing for seabed classification and marine environment information extraction and is producing an abundance of scientific results. Data-driven machine learning divides the data into a training set and test set. The training set is used to create a model that is suitable for machine learning, and the model's accuracy is increased through a repetitive model update process in which the model is validated via the

In this article, we propose an active target detector by using a shallow neural network (NN) with novel features under small sonar data, where deep learning (DL) models are restricted. The features are contrived by considering differences between target and clutter signals in the temporal variation of amplitude and frequency when using an active ping; these are extracted from preprocessed spectrograms of beam time series. The classification ability of the features is demonstrated through comparison with that of prevalent timbral features in the training data set that reflect the human auditory system. Finally, the detection performance of the shallow NN trained using the proposed features is compared with those from the fine-tuned DL models. During the generalization test, the DL models miss a significant number of targets in the test data set—including target signals with patterns unseen during training—and the VGG16 used for detailed comparison has a detection probability of 0.69. The probability increases to 0.86 via the shallow NN, with remarkably enhanced target detection. This demonstrates the competitive performance and generalization ability of the shallow NN, using features that consider active sonar returns under a lack of data.

A physics-informed neural network (PINN) enables the estimation of horizontal modal wavenumbers using ocean pressure data measured at multiple ranges. Mode representations for the ocean acoustic pressure field are derived from the Hankel transform relationship between the depth-dependent Green's function in the horizontal wavenumber domain and the field in the range domain. We obtain wavenumbers by transforming the range samples to the wavenumber domain, and maintaining range coherence of the data is crucial for accurate wavenumber estimation. In the ocean environment, the sensitivity of phase variations in range often leads to degradation in range coherence. To address this, we propose using OceanPINN [Yoon, Park, Gerstoft, and Seong, J. Acoust. Soc. Am. 155(3), 2037–2049 (2024)] to manage spatially non-coherent data. OceanPINN is trained using the magnitude of the data and predicts phase-refined data. Modal wavenumber estimation methods are then applied to this refined data, where the enhanced range coherence results in improved accuracy. Additionally, sparse Bayesian learning, with its high-resolution capability, further improves the modal wavenumber estimation. The effectiveness of the proposed approach is validated through its application to both simulated and SWellEx-96 experimental data.