A Feature Optimization Approach Based on Inter-Class and Intra-Class Distance for Ship Type Classification

Abstract

Deep learning based methods have achieved state-of-the-art results on the task of ship type classification. However, most existing ship type classification algorithms take time–frequency (TF) features as input, the underlying discriminative information of these features has not been explored thoroughly. This paper proposes a novel feature optimization method which is designed to minimize an objective function aimed at increasing inter-class and reducing intra-class feature distance for ship type classification. The objective function we design is able to learn a center for each class and make samples from the same class closer to the corresponding center. This ensures that the features maximize underlying discriminative information involved in the data, particularly for some targets that usually confused by the conventional manual designed feature. Results on the dataset from a real environment show that the proposed feature optimization approach outperforms traditional TF features.

Underwater Acoustic Target Recognition with a Residual Network and the Optimized Feature Extraction Method

“…To address the face verification problem, center loss [22] is first exploited to compensate for softmax loss in 2016. For the UATR task, we improve generalization by leveraging such a simple but useful cost function, since it learns a center for the features of each class and meanwhile tries to pull the deep features of the same class close to the corresponding center [5]. Basically, for classification task {(x i , y i )} N i=1 consisting of N samples x i and their corresponding labels y i ∈ {1, 2, .…”

“…As for the CRNN network, the first layer has 64 filters of the size of (5,5) and the regularization method is L2 with the lambda of 0.01. The activation function is ReLU.…”

“…Hence, developing a robust recognizing system to replace humans' work of identifying ship-radiated noise is of great importance. From a technical perspective, efforts are consistently made to improve the classification accuracy in the aspects of feature extraction and classifier training [5].…”

“…As the results obtained by typical methods based on machine learning are not very high, the methods based on deep learning models are worthy of researching. Li et al [5] introduce a feature optimization approach with Deep Neural Networks (DNN) and optimizing loss function and achieve an accuracy of 84%. Yang et al [17] propose a so-called competitive Deep Belief Nets (cDBN) for UATR.…”

Underwater Target Signal Classification Using the Hybrid Routing Neural Network

“…The deep neural network (DNN) uses modified time–frequency characteristics as input, which more clearly expresses the outstanding center feature of ship samples. There are impressive classification results in the ship signal dataset, and it is due to the fact that the identification anchors based on the objective function are gained in the space distance [ 13 ]. The two dimensions of a convolutional neural network (CNN) excavates the ship signal characteristics in the spatial-temporal spectrum domain, which can weaken spectral fluctuation and prevent local minima [ 14 ].…”