A5063464075- Advanced Graph Neural Networks
- Topic Modeling
- Maritime Transport Emissions and Efficiency
- Direction-of-Arrival Estimation Techniques
- Underwater Acoustics Research
- Natural Language Processing Techniques
- Ship Hydrodynamics and Maneuverability
- Speech and Audio Processing
- Maritime Navigation and Safety
- Blind Source Separation Techniques
- Structural Health Monitoring Techniques
- Advanced SAR Imaging Techniques
- Ocean Waves and Remote Sensing
- Text and Document Classification Technologies
- Data Quality and Management
Xidian University
2023-2024
Xuzhou University of Technology
2024
Wuhan University of Technology
2022-2023
Named entity recognition (NER) is an important research direction in natural language processing (NLP). Traditional machine learning algorithms NER have problems such as low accuracy, highly dependent feature design, poor domain adaptability, and inability to handle the different contexts of multiple meanings term recognizing Chinese entities. Based on these problems, this paper adopts a method based BERT-CRF model NER. The BERT preprocessing generates word vectors that represent contextual...
Information retrieval-based question answering (IRQA) and knowledge-based (KBQA) are the main forms of (QA) systems. The answer generated by IRQA system is extracted from relevant text but has a certain degree randomness, while KBQA retrieves structured data, its accuracy relatively high. In field policy regulations such as household registration, QA requires precise rigorous answers. Therefore, we design based on registration knowledge graph, aiming to provide accurate answers for...
This paper utilizes the BERTbased-BiLSTM-CRF models to complete Chinese named entity recognition tasks, including finetuned and unfinetuned BERT models. Use pre-training model BERT(Bidirectional Encoder Representations from Transformers), a BiLSTM(Bi-directional Long Short-Term Memory) network CRF(Conditional Random Field) perform NER(Named Entity Recognition) on Chinese. Tested people-daily-ner-pretreatment corpus, compared with other models, BERTbased can effectively identify information,...
Accurate large-scale regional wave height prediction is important for the safety of ocean sailing. A multi-step model (ConvGRU-RMWP) based on ConvGRU designed two problems difficult spatial feature resolution and low accuracy in navigation prediction. For prediction, a multi-input multi-output strategy used, direction period are used as exogenous variables, which combined with historical data to expand sample space. features, convolutional gated recurrent neural network an Encoder-Forecaster...
Mathematically speaking, the direction of arrival (DOA) estimation methods with data-driven deep learning generally exhibit enhanced robustness under unknown clutter and noise scenarios. However, it is worthily noticed that existing mostly suffer from signal model mismatch, due to relying on real-valued operations ignoring actual received signals are complex values. To this regard, a novel DOA method sparse prior based complex-valued convolutional neural network (CV -CNN) proposed....
In response to the increasingly complex navigational conditions and in order ensure safety of ship navigation, improved particle swarm optimisation algorithm is proposed optimise weights fuzzy assessment method establish an environmental risk evaluation model for ocean-going passenger ships. As traditional easy fall into local optimum, it give perturbation when particles are stable state jump out optimum. The optimized applied assess under analysis factors affecting ship's navigation. scores...
In traditional multiple input output (MIMO) mmWave radar, missing estimation is a common problem in coherent source scenarios due to low angle resolution elevation and azimuth dimensions. To address this, we propose novel two-dimensional (2-D) super-resolution method based on compressed sensing. Our approach involves spatial virtual transformation of the nonuniform sparse array <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$M$</tex> emitters...
Domain knowledge graph, as a branch of focuses both the coverage and usage in specific field, so it has high requirements for depth accuracy this field. At present, there are still problems rapid construction application domain graphs. Especially before construction, complex scenarios involve wide range dimensions, is difficult business experts to comprehensively construct conceptual data model short period time. In response problem, paper proposes graph method based on review documents,...