Auditory Inspired Convolutional Neural Networks for Ship Type Classification with Raw Hydrophone Data

Abstract

Detecting and classifying ships based on radiated noise provide practical guidelines for the reduction of underwater noise footprint of shipping. In this paper, the detection and classification are implemented by auditory inspired convolutional neural networks trained from raw underwater acoustic signal. The proposed model includes three parts. The first part is performed by a multi-scale 1D time convolutional layer initialized by auditory filter banks. Signals are decomposed into frequency components by convolution operation. In the second part, the decomposed signals are converted into frequency domain by permute layer and energy pooling layer to form frequency distribution in auditory cortex. Then, 2D frequency convolutional layers are applied to discover spectro-temporal patterns, as well as preserve locality and reduce spectral variations in ship noise. In the third part, the whole model is optimized with an objective function of classification to obtain appropriate auditory filters and feature representations that are correlative with ship categories. The optimization reflects the plasticity of auditory system. Experiments on five ship types and background noise show that the proposed approach achieved an overall classification accuracy of 79.2%, which improved by 6% compared to conventional approaches. Auditory filter banks were adaptive in shape to improve accuracy of classification.

Ship Type Classification by Convolutional Neural Networks with Auditory-Like Mechanisms

“…Biophysical based models [3,4] are limited to early auditory stages for extracting auditory features. Auditory features designed from perceptual evidence always focus on the properties of signal description rather than the classification purpose [6]. These features do not utilize the plastic mechanism and representation at various auditory stages to improve the recognition performance.…”

“…Deep learning has made it possible to model the original signal as well as to predict targets in a whole model [6,14], to which the auditory system is thought to be adapted. The time convolutional layer in an auditory inspired convolutional neural network (CNN) [6] provided a new way for modeling underwater acoustic signals. However, it did not have enough depth to build an appropriate model to match the expanding acquired dataset.…”

“…The construction of the deep network refers to inception and residual network [15] in the field of machine vision, in which some ideas are also found in the auditory pathway. Thus, the frequency convolutional layers in Reference [6] are improved to increase the depth of the network. At the final stage, the substitution of the fully connected layer by global average pooling at target layer greatly reduces the parameters.…”

Deep Neural Networks for ECG-Based Pulse Detection during Out-of-Hospital Cardiac Arrest

“…Previously a machine learning technique was proposed using a random forest (RF) classifier based on hand-crafted features [ 34 ]. Alternatively, deep neural networks (DNN) have shown superior performance in classification problems with large datasets in many fields [ 35 , 36 , 37 , 38 ]. DNN solutions have no need of feature engineering as the signals are directly fed to the network which does the exploratory data analysis.…”

Information Theory Applications in Signal Processing

“…In [ 25 ], Ballesteros et al propose variable length codes, based on Collatz conjecture, for transforming images into non-intelligible audio, aiming in this way at providing privacy to image transmissions through an encryption scheme. Finally, in the field of acoustic signal processing, Shen et al [ 26 ] develop an auditory inspired convolutional neural network that simulates the processing procedure of the human auditory system in discriminating ships of different classes from raw hydrophone data.…”