Search

Your search keyword '"Zhou, Changjiang"' showing total 566 results
Start Over Author "Zhou, Changjiang"
566 results on '"Zhou, Changjiang"'

2. CorpusBrain++: A Continual Generative Pre-Training Framework for Knowledge-Intensive Language Tasks

Author

Guo, Jiafeng, Zhou, Changjiang, Zhang, Ruqing, Chen, Jiangui, de Rijke, Maarten, Fan, Yixing, and Cheng, Xueqi

Subjects
Computer Science - Information Retrieval, Computer Science - Computation and Language
Abstract

Knowledge-intensive language tasks (KILTs) typically require retrieving relevant documents from trustworthy corpora, e.g., Wikipedia, to produce specific answers. Very recently, a pre-trained generative retrieval model for KILTs, named CorpusBrain, was proposed and reached new state-of-the-art retrieval performance. However, most existing research on KILTs, including CorpusBrain, has predominantly focused on a static document collection, overlooking the dynamic nature of real-world scenarios, where new documents are continuously being incorporated into the source corpus. To address this gap, it is crucial to explore the capability of retrieval models to effectively handle the dynamic retrieval scenario inherent in KILTs. In this work, we first introduce the continual document learning (CDL) task for KILTs and build a novel benchmark dataset named KILT++ based on the original KILT dataset for evaluation. Then, we conduct a comprehensive study over the use of pre-trained CorpusBrain on KILT++. Unlike the promising results in the stationary scenario, CorpusBrain is prone to catastrophic forgetting in the dynamic scenario, hence hampering the retrieval performance. To alleviate this issue, we propose CorpusBrain++, a continual generative pre-training framework. Empirical results demonstrate the significant effectiveness and remarkable efficiency of CorpusBrain++ in comparison to both traditional and generative IR methods., Comment: Submitted to ACM Transactions on Information Systems

Published
2024

4. End-to-End Entity Detection with Proposer and Regressor

Author

Wen, Xueru, Zhou, Changjiang, Tang, Haotian, Liang, Luguang, Jiang, Yu, and Qi, Hong

Subjects
Computer Science - Computation and Language
Abstract

Named entity recognition is a traditional task in natural language processing. In particular, nested entity recognition receives extensive attention for the widespread existence of the nesting scenario. The latest research migrates the well-established paradigm of set prediction in object detection to cope with entity nesting. However, the manual creation of query vectors, which fail to adapt to the rich semantic information in the context, limits these approaches. An end-to-end entity detection approach with proposer and regressor is presented in this paper to tackle the issues. First, the proposer utilizes the feature pyramid network to generate high-quality entity proposals. Then, the regressor refines the proposals for generating the final prediction. The model adopts encoder-only architecture and thus obtains the advantages of the richness of query semantics, high precision of entity localization, and easiness of model training. Moreover, we introduce the novel spatially modulated attention and progressive refinement for further improvement. Extensive experiments demonstrate that our model achieves advanced performance in flat and nested NER, achieving a new state-of-the-art F1 score of 80.74 on the GENIA dataset and 72.38 on the WeiboNER dataset., Comment: Accepted to the Neural Processing Letters (2023)

Published
2022
Full Text
View/download PDF

5. Type-supervised sequence labeling based on the heterogeneous star graph for named entity recognition

Author

Wen, Xueru, Zhou, Changjiang, Tang, Haotian, Liang, Luguang, Jiang, Yu, and Qi, Hong

Subjects
Computer Science - Computation and Language
Abstract

Named entity recognition is a fundamental task in natural language processing, identifying the span and category of entities in unstructured texts. The traditional sequence labeling methodology ignores the nested entities, i.e. entities included in other entity mentions. Many approaches attempt to address this scenario, most of which rely on complex structures or have high computation complexity. The representation learning of the heterogeneous star graph containing text nodes and type nodes is investigated in this paper. In addition, we revise the graph attention mechanism into a hybrid form to address its unreasonableness in specific topologies. The model performs the type-supervised sequence labeling after updating nodes in the graph. The annotation scheme is an extension of the single-layer sequence labeling and is able to cope with the vast majority of nested entities. Extensive experiments on public NER datasets reveal the effectiveness of our model in extracting both flat and nested entities. The method achieved state-of-the-art performance on both flat and nested datasets. The significant improvement in accuracy reflects the superiority of the multi-layer labeling strategy.

Published
2022

20. Phase-weighted slant stacking for surface wave dispersion measurement

Author

Cheng, Feng, Xia, Jianghai, Zhang, Kai, Zhou, Changjiang, and Ajo-Franklin, Jonathan B

Subjects
Seismic interferometry, Seismic noise, Surface waves and free oscillations, Geology, Geophysics, Geomatic Engineering, Geochemistry & Geophysics
Abstract

Surface wave retrieval from ambient noise records using seismic interferometry techniques has been widely used for multiscale shear wave velocity (Vs) imaging. One key step during Vs imaging is the generation of dispersion spectra and the extraction of a reliable dispersion curve from the retrieved surface waves. However, the sparse array geometry usually affects the ability for high-frequency (>1 Hz) seismic signals' acquisition. Dispersion measurements are degraded by array response due to sparse sampling and often present smeared dispersion spectra with sidelobe artefacts. Previous studies usually focus on interferograms' domain (e.g. cross-correlation function) and attempt to enhance coherent signals before dispersion measurement. We propose an alternative technique to explicitly deblur dispersion spectra through use of a phase-weighted slant-stacking algorithm. Numerical examples demonstrate the strength of the proposed technique to attenuate array responses as well as incoherent noise. Three different field examples prove the flexibility and superiority of the proposed technique: The first data set consists of ambient noise records acquired using a nodal seismometer array; the second data set utilizes distributed acoustic sensing (DAS) and a marine fibre-optic cable to acquire a similar ambient noise data set; the last data set is a vibrator-based active-source surface wave data. The enhanced dispersion measurements provide cleaner and higher-resolution spectra without distortions which will assist both human interpreters as well as ML algorithms in efficiently picking curves for subsequent Vs inversion.

Published
2021

25. EXPERIMENT AND SIMULATION ON MECHANICAL PROPERTIES OF OSTEOPOROTIC MOUSE BONE (MT)

Author

ZHAO Qing, YIN ZiYang, ZHOU Wen, REN GuanLin, KUANG ChunMei, HU Cong, and ZHOU ChangJiang

Subjects
Mouse tibia, Osteoporosis, Mechanical properties, Finite element simulation, Mechanical engineering and machinery, TJ1-1570, Materials of engineering and construction. Mechanics of materials, TA401-492
Abstract

The osteoporosis model of mouse tibia was established based on castration, and the effect of osteoporosis on the mechanical properties of tibia was investigated. The two-dimensional sections of mouse tibia were obtained by Micro-CT. The three-dimensional digital model of mouse tibia was obtained by reconstruction technology, and trabecular bone shape and density were analyzed. The effect of osteoporosis on the mechanical properties of tibia was explored by testing and finite element simulation. The results showed that osteoporosis could damage the trabecular structure and decrease the mechanical properties of tibia. Compared with healthy tibia, the maximum load, bending stiffness and elastic modulus of tibia decreased by 20%, 30% and 25%, respectively, and the simulation results were in good agreement with the experiment. This study provides an important basis for the clinical diagnosis and treatment of osteoporosis.

Published
2023
Full Text
View/download PDF

47. High Frequency Ultrasound Transducer Based on Sm-Doped Pb(Mg1/3Nb2/3)O3-0.28PbTiO3 Ceramic for Intravascular Ultrasound Imaging.

Author

Weiyan, Ding, Chen, Xingfei, Zhang, Yongcheng, Li, Xiaobing, Sun, Fenglong, Yang, Zhaoping, Tang, Xi, Zhou, Changjiang, Wang, Feifei, and Zhao, Xiangyong

Subjects
ULTRASONIC imaging, PIEZOELECTRIC materials, TRANSDUCERS, INTRAVASCULAR ultrasonography, ACOUSTIC imaging
Abstract

Sm-doped Pb(Mg1/3Nb2/3)O3-0.28PbTiO3 (PMN-0.28PT) ceramic has been reported to exhibit very large piezoelectric response (d 33~1300 pC/N) that can be comparable with PMN-0.30PT single crystal. Based on the Sm-doped PMN-0.28PT ceramics, a high frequency ultrasound transducer with the center frequency above 30 MHz has been designed and fabricated for intravascular ultrasound imaging, and the performance of the transducer was investigated via ultrasound pulse-echo tests. Further, for a porcine vessel wall, the 2D and 3D ultrasound images were constructed using signal acquisition and processing from the fabricated high-frequency transducer. The obtained details of the vessel wall by the IVUS transducer indicate that Sm-doped PMN-0.28PT ceramic is a promising candidate for high frequency transducers. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results