Automotive radars have faced to various jamming signals with different types and parameters. One counter measure is to identify the fine-grained modulation types of these jamming signals, which can provide supplementary information for intercepting. Most existing signal modulation recognition methods attempt to establish a machine-learning mechanism by training with a large number of annotated samples, which is hardly applicable in real-world electronic reconnaissance scenarios where only a few samples can be intercepted in advance. Few-Shot Learning aims to learn from base classes with many samples and infuse this knowledge to support classes with only a few samples, thus realizing model generalization. In this paper, inspired by the fact that the energy distribution of the signal of interest is always concentrated in time-frequency images, a novel few-shot learning framework based on foreground segmentation is proposed, which benefits from its powerful filtering to remove the noise and clutter from background. The experimental results show that the proposed method can achieve excellent performance for fine-grained signal modulation recognition even with only one support sample and is robust under different signal-to-noise-ratio conditions.

The multi-functional radars (MFRs) can switch among a variety of fine-grained working modes, which often have flexible modulation types and programmable parameters. In an electromagnetic reconnaissance system, the process of identifying different working modes in pulse sequences guarantees the subsequent intention analysis and assists in devising a jamming strategy. Most of the existing working mode recognition methods attempt to establish a machine-learning mechanism by training a model using a large number of annotated samples. However, this is hardly applicable in the real-world scenarios where only a few samples can be intercepted in advance. As the labeled signal samples are expensive, one direction is to augment the dataset by generating either samples or signal features in an embedding space. In this work, inspired by the fact that different modalities of the same working modes are generated from a set of fixed parameters, a few-shot learning framework based on compound alignments is proposed. Three branches, which take observations of long windows, observations of short windows, and coded semantic attributes as inputs, are aligned in both the latent variable space and the reconstruction space to learn a shared embedded space. A simple soft-max classifier is trained by sampling sufficient instances in the learned space to realize the final identification process. The experimental results and analysis show that the proposed method achieves excellent performance for fine-grained working modes recognition even for small number of observed pulses. In addition, the proposed method is robust under different non-ideal conditions, such as noise contamination, incomplete sequences, and spurious pulses.

Multi-functional radars (MFRs) can generate a variety of working modes for different tasks based on flexible modulation types and programmable parameters. The neural network has been widely used to recognize of these fine-grained MFR modes. However, it requires a large number of samples with expert annotation in advance, which is hardly available in practical applications. Therefore, few-shot learning (FSL) method is used to learn “general information” and transfer it into new tasks where only a small number of labeled samples are provided. In order to improve its effect, unlabeled samples are utilized to provide “manifolds information” that describes the distribution among data. This paper proposes a framework of coding refined prototypical random walk network (C-RPRWN) combining these two kinds of information. The whole framework is divided into three modules: preprocessing module, embedding module and refined prototypical random walk (RPRW) module, which are respectively used to enhance the signal expression to adapt to non-ideal situations, extract distinguishable features to compute prototypes, and utilize “manifolds information” for better classification. The experimental results and analysis show that the proposed method achieves excellent performance for MFR fine-grained modes recognition even under the condition of a small number of samples. In addition, the robustness of the proposed method is verified under the influence of different non-ideal factors.

Semi-supervised incremental learning is the task of classifying data streams with partially labeled data when annotation information is difficult to obtain. Besides the sequential learning manner and lack of label information, multiple novel classes and concept drift may emerge from incremental learning. Most previous studies have only considered these problems in part. To tackle challenges involved in semi-supervised incremental learning, an Adaptive Matrix Sketching and Clustering (AMaSC) method is proposed in this paper, which cohesively and adaptively classifies known classes, identifies multiple novel classes, and updates the learning model. Experiments were conducted to evaluate this method on three benchmark datasets, containing various data types, including network attack analysis, the geospatial information of forests, and images of handwritten numbers. Results validated the effectiveness of our proposed method and its superiority over many previous studies.

Multi-function radars (MFRs) can perform multiple missions simultaneously by optimizing transmission signals using programmable parameters, which is a great challenge for reconnaissance and identification. In particular, this recognition becomes more challenging when there is no prior information on radiation sources and when the available labeled signal data is insufficient. The received signal is also often a mixture of signals from targets, noise, unknown radiation sources, or unknown working modes. This paper proposes a Multi-alignment Taskadaptive (MaTa) method to simultaneously complete the detection of unknown signals and the classification of target mode signals with limited samples. The proposed method utilizes generative model to map the observed signal sample and its semantic descriptions of the working mode to the same latent variables space through multi-alignment. Each working mode generates a prototype using a small number of projections in this space to support classification. This paper additionally generates negative prototypes without unknown signal sample provided to meet the requirement of dynamic adjusting the detection boundary in different tasks for unknown samples. The proposed method shows the best experimental results compared with baselines, which achieves a 96.73% accuracy for 5-categories providing one sample per class under few-shot open-set learning condition.

Zero-shot learning (ZSL) is the task of recognizing samples from their related classes which have never been seen during model training. ZSL is generally realized through learning a common embedding space where both high dimensional visual features and some pre-defined semantics can be mapped. However, this kind of solutions usually suffers from domain shift. In addition, the limitation and subjectivity of manual semantic information can also affect the classification results. To address these challenges, this paper proposes a novel end-to-end deep learning model called Cross-Layer Autoencoder (CLAE), which integrates different ways of semantic mapping and maintains reconstruction information. Besides, a regularized loss function is used to preserve local class manifolds. Extensive experiments for both traditional and generalized ZSL tasks are conducted on several benchmark datasets, and high effectiveness of the proposed method and its superiority over many previous researches are demonstrated.