A5020274002- Wireless Signal Modulation Classification
- Topic Modeling
- Domain Adaptation and Few-Shot Learning
- Text and Document Classification Technologies
- Full-Duplex Wireless Communications
- Image Retrieval and Classification Techniques
- Recommender Systems and Techniques
- Hydrocarbon exploration and reservoir analysis
- Mineral Processing and Grinding
- Natural Language Processing Techniques
- Digital Media Forensic Detection
- Nuclear Physics and Applications
- Radar Systems and Signal Processing
- Geophysical Methods and Applications
National University of Defense Technology
2021-2022
Specific emitter identification (SEI) refers to distinguishing emitters using individual features extracted from wireless signals. The current SEI methods have proven be accurate in tackling large labeled data sets at a high signal-to-noise ratio (SNR). However, their performance declines dramatically the presence of small samples and significant noise environment. To address this issue, we propose complex self-supervised learning scheme fully exploit unlabeled samples, comprised pretext...
Specific emitter identification (SEI) is extracting the features of received radio signals and determining individuals that generate signals. Although deep learning-based methods have been effectively applied for SEI, their performance declines dramatically with smaller number labeled training samples in presence significant noise. To address this issue, we propose an improved Bootstrap Your Own Late (BYOL) self-supervised learning scheme to fully exploit unlabeled samples, which comprises...
Specific Emitter Identification (SEI) is the approach to identify emitter individuals using received wireless signals. Despite fact that deep learning has been successfully applied in SEI, performance still unsatisfying when receiver changes. In this paper, we introduce a domain adaptation method, namely Deep Adversarial Neural Network (DANN), for cross-receiver SEI. Furthermore, separated batch normalization (SepBN) proposed improve performance. Results of experiments real data show...
The intelligent recognition of rock sample lithology plays an important role in mineral resources exploration. According to the image, a depth learning model is established. In order solve problem gradient disappearance caused by excessive neural network, residual structure introduced, ResNet built, and comparison based self-supervised classification algorithm established, which does not depend on any label value. Using encoder network extract features calculate reconstruction error pixel...
Abstract Emitter identification technology can distinguish the types of radiation sources and identify identity emitter. It has broad application prospects in both military civilian fields. The article mainly reviews source feature extraction methods for individual recent years, discusses advantages disadvantages manually extracted features based on deep learning. technical difficulties are summarized with respect to environment, number sources, performance algorithms, etc. Finally, points...
Among various recommender techniques, collaborative filtering (CF) is the most successful one. And a key problem in CF how to represent users and items. Previous works usually user (an item) as vector of latent factors (aka. \textit{embedding}) then model interactions between items based on representations. Despite its effectiveness, we argue that it's insufficient yield satisfactory embeddings for filtering. Inspired by idea SVD++ represents themselves their interacted items, propose...
Relation extraction is a fundamental task in natural language processing, aiming at extracting relational triples from plain text. However, there are fewer instances the manually constructed dataset to meet learning needs of relation models. Distant supervision approach has attracted interest numerous researchers due its ability construct large datasets low cost. Nevertheless, certain problems with distant overly strong assumptions. In this paper, we introduce three main supervised...