Search

Your search keyword '"Jiang, Jinrong"' showing total 50 results
Start Over Author "Jiang, Jinrong"
50 results on '"Jiang, Jinrong"'

1. Unveiling and Controlling Anomalous Attention Distribution in Transformers

Author

Yan, Ruiqing, Du, Xingbo, Deng, Haoyu, Zheng, Linghan, Sun, Qiuzhuang, Hu, Jifang, Shao, Yuhang, Jiang, Penghao, Jiang, Jinrong, and Zhao, Lian

Subjects
Computer Science - Machine Learning, Computer Science - Artificial Intelligence
Abstract

With the advent of large models based on the Transformer architecture, researchers have observed an anomalous phenomenon in the Attention mechanism--there is a very high attention on the first element, which is prevalent across Transformer-based models. It is crucial to understand it for the development of techniques focusing on attention distribution, such as Key-Value (KV) Cache compression and infinite extrapolation; however, the latent cause leaves to be unknown. In this paper, we analyze such a phenomenon from the perspective of waiver phenomenon, which involves reducing the internal values of certain elements in the sequence, allowing them to absorb excess attention without affecting their contribution to information. In specific models, due to differences in positional encoding and attention patterns, we have found that the selection of waiver elements by the model can be categorized into two methods: positional-encoding-based and feature-distribution-within-elements-based.

Published
2024

2. High-Performance Computing Technology Contributes to the Construction of the Earth System Numerical Simulation Device

Author

Zhou, Guangqing, Zhao, Lian, Jiang, Jinrong, Chinese Academy of Sciences, Ministry of Education of the PRC, Ministry of Science and Technology of the PRC, China Association for Science and Technology, Chinese Academy of Social Sciences, Chinese Academy of Engineering, National Natural Science Foundation of China, and Chinese Academy of Agricultural Sciences

Published
2024
Full Text
View/download PDF

4. CAINNFlow: Convolutional block Attention modules and Invertible Neural Networks Flow for anomaly detection and localization tasks

Author

Yan, Ruiqing, Zhang, Fan, Huang, Mengyuan, Liu, Wu, Hu, Dongyu, Li, Jinfeng, Liu, Qiang, Jiang, Jinrong, Guo, Qianjin, and Zheng, Linghan

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence
Abstract

Detection of object anomalies is crucial in industrial processes, but unsupervised anomaly detection and localization is particularly important due to the difficulty of obtaining a large number of defective samples and the unpredictable types of anomalies in real life. Among the existing unsupervised anomaly detection and localization methods, the NF-based scheme has achieved better results. However, the two subnets (complex functions) $s_{i}(u_{i})$ and $t_{i}(u_{i})$ in NF are usually multilayer perceptrons, which need to squeeze the input visual features from 2D flattening to 1D, destroying the spatial location relationship in the feature map and losing the spatial structure information. In order to retain and effectively extract spatial structure information, we design in this study a complex function model with alternating CBAM embedded in a stacked $3\times3$ full convolution, which is able to retain and effectively extract spatial structure information in the normalized flow model. Extensive experimental results on the MVTec AD dataset show that CAINNFlow achieves advanced levels of accuracy and inference efficiency based on CNN and Transformer backbone networks as feature extractors, and CAINNFlow achieves a pixel-level AUC of $98.64\%$ for anomaly detection in MVTec AD.

Published
2022

7. Portably parallel construction of a CI wave function from a matrix-product state using the Charm++ framework

Author

Wang, Ting, Ma, Yingjin, Zhao, Lian, and Jiang, Jinrong

Subjects
Physics - Computational Physics, Condensed Matter - Strongly Correlated Electrons
Abstract

The constructions of configuration interaction (CI) expansions from a matrix-product state (MPS) involves numerous matrix operations and the skillful sampling of important configurations when in a huge Hilbert space. In this work, we present an efficient procedure for constructing CI expansions from MPS using the Charm++ parallel programming framework, upon which automatic load balancing and object migration facilities can be employed. This procedure was employed in the MPS-to-CI utility (Moritz et al., J. Chem. Phys. 2007, 126, 224109), sampling-reconstructed complete active space algorithm (SR-CAS, Boguslawski et al., J. Chem. Phys. 2011, 134, 224101) and entanglement-driven genetic algorithm (EDGA, Luo et al., J. Chem. Theory Comput. 2017, 13, 4699-4710). It enhances productivity and allows the sampling programs evolve to their population-expansion versions (e.g., EDGA with population expansion [PE-EDGA]). Examples of 1,2-dioxetanone and firefly dioxetanone anion (FDO-) molecules demonstrated that 1) the procedure could be flexibly employed among various multi-core architectures; 2) the parallel efficiencies could be persistently improved simply by increasing the proportion of asynchronous executions; 3) PE-EDGA could construct a CAS-type CI wave function from a huge Hilbert space, with 0.9952 CI completeness and 96.7% correlation energy via ~1.66x10^6 configurations (only 0.0000028% of the total configurations) of a bi-radical state of FDO- molecule using the full valence active space within a few hours.

Published
2020

12. The Super-large Ensemble Experiments of CAS FGOALS-g3

Author

Lin, Pengfei, Zhao, Bowen, Wei, Jilin, Liu, Hailong, Zhang, Wenxia, Chen, Xiaolong, Jiang, Jie, Ding, Mengrong, Man, Wenmin, Jiang, Jinrong, Zhang, Xu, Ding, Yuewen, Bai, Wenrong, Jin, Chenyang, Yu, Zipeng, Li, Yiwen, Zheng, Weipeng, and Zhou, Tianjun

Published
2022
Full Text
View/download PDF

14. DLENSO: A Deep Learning ENSO Forecasting Model

Author

He, Dandan, Lin, Pengfei, Liu, Hailong, Ding, Lei, Jiang, Jinrong, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Woeginger, Gerhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Nayak, Abhaya C., editor, and Sharma, Alok, editor

Published
2019
Full Text
View/download PDF

26. Typhoon storm surge ensemble forecast based on GPU technique

Author

Changkuan Zhang, Jianxi Dong, Xiaojiang Song, Yuzhu Wang, Fujiang Yu, Jiang Jinrong, and Qiuxing Liu

Subjects
South china, 010504 meteorology & atmospheric sciences, Meteorology, Ensemble forecasting, Fortran, Storm surge, Aquatic Science, 010502 geochemistry & geophysics, Oceanography, Track (rail transport), 01 natural sciences, Physics::Geophysics, Typhoon, Environmental science, Surge, computer, Physics::Atmospheric and Oceanic Physics, 0105 earth and related environmental sciences, computer.programming_language, Parametric statistics
Abstract

The accuracy of typhoon forecasts plays an important role in the prediction of storm surges. The uncertainty of a typhoon’s intensity and track means it is necessary to use an ensemble model to predict typhoon storm surges. A hydrodynamic model, which is operational at the National Marine Environmental Forecasting Center, is applied to conduct surge simulations for South China coastal areas using the best track data with parametric wind and pressure models. The results agree well with tidal gauge observations. To improve the calculation efficiency, the hydrodynamic model is modified using CUDA Fortran. The calculation results are almost the same as those from the original model, but the calculation time is reduced by more than 99%. A total of 150 typhoon cases are generated by combining 50 typhoon tracks from the European Centre for Medium-Range Weather Forecasts with three possible typhoon intensity forecasts. The surge ensembles are computed by the improved hydrodynamic model. Based on the simulated storm surges for the different typhoon cases, ensemble and probability forecast products can be provided. The mean ensemble results and probability forecast products are shown to agree well with the observed storm surge caused by Typhoon Mangkhut. The improved model is highly suitable for ensemble numerical forecasts, providing better forecast products for decision-making, and can be easily implemented to run on regular workstations.

Published
2020

29. The GPU version of LASG/IAP Climate System Ocean Model version 3 (LICOM3) under the heterogeneous-compute interface for portability (HIP) framework and its large-scale application

Author

Wang, Pengfei, primary, Jiang, Jinrong, additional, Lin, Pengfei, additional, Ding, Mengrong, additional, Wei, Junlin, additional, Zhang, Feng, additional, Zhao, Lian, additional, Li, Yiwen, additional, Yu, Zipeng, additional, Zheng, Weipeng, additional, Yu, Yongqiang, additional, Chi, Xuebin, additional, and Liu, Hailong, additional

Published
2021
Full Text
View/download PDF

33. Description and Climate Simulation Performance of CAS‐ESM Version 2

Author

Zhang, He, primary, Zhang, Minghua, additional, Jin, Jiangbo, additional, Fei, Kece, additional, Ji, Duoying, additional, Wu, Chenglai, additional, Zhu, Jiawen, additional, He, Juanxiong, additional, Chai, Zhaoyang, additional, Xie, Jinbo, additional, Dong, Xiao, additional, Zhang, Dongling, additional, Bi, Xunqiang, additional, Cao, Hang, additional, Chen, Huansheng, additional, Chen, Kangjun, additional, Chen, Xueshun, additional, Gao, Xin, additional, Hao, Huiqun, additional, Jiang, Jinrong, additional, Kong, Xianghui, additional, Li, Shigang, additional, Li, Yangchun, additional, Lin, Pengfei, additional, Lin, Zhaohui, additional, Liu, Hailong, additional, Liu, Xiaohong, additional, Shi, Ying, additional, Song, Mirong, additional, Wang, Huijun, additional, Wang, Tianyi, additional, Wang, Xiaocong, additional, Wang, Zifa, additional, Wei, Ying, additional, Wu, Baodong, additional, Xie, Zhenghui, additional, Xu, Yongfu, additional, Yu, Yongqiang, additional, Yuan, Liang, additional, Zeng, Qingcun, additional, Zeng, Xiaodong, additional, Zhao, Shuwen, additional, Zhou, Guangqing, additional, and Zhu, Jiang, additional

Published
2020
Full Text
View/download PDF

44. The Relationship of Financial Development, Urbanization and Urban-Rural Income Gap: An Empirical Research Based on Provincial Panel Data in China

Author

Chen Shaowei and Jiang Jinrong

Subjects
Economic growth, Comprehensive income, financial development, the urban-rural income gap, lcsh:Social Sciences, lcsh:H, Empirical research, Income distribution, Scale (social sciences), Urbanization, Development economics, Economics, the urbanization, Financial market efficiency, Panel data, Diversity (business)
Abstract

Financial development and the urbanization are important influence factors of the urban-rural income gap, and financial development can be measured by three main indexes which are financial scale, financial activities and financial efficiency. The financial development scale of China has obviously widened the urban-rural income gap. But rural financial activities have obvious effect on increasing farmers’ income, and the improvement of financial efficiency is helpful for narrowing the urban-rural income gap. The Kuznets effect between economic development and the urban-rural income gap has regional diversity. Besides that, the improvement of urbanization is also helpful for shorting the urban-rural income gap.

Published
2014

47. Application of Grid Nesting Techniques on Cuace-Dust

Author

Cai Chang-qing, Jiang Jinrong, and Chi Xuebin

Subjects
Atmosphere (unit), Meteorology, Numerical analysis, Key (cryptography), Mesoscale meteorology, Weather forecasting, Environmental science, Nesting (computing), Storm, Grid, computer.software_genre, computer
Abstract

This paper introduces the grid nesting techniques and their application on the sand-dust storm forecasting model CUACE-Dust (CMA unified Atmosphere Chemistry Environment-Dust), which is coupled with PSU/NCAR mesoscale model. Based on CUACE-Dust, three numerical experiments are designed to show the advantage of the grid nesting techniques. In the key area that attracts attention, the nesting techniques help us get a result as good as that of the finer grid for the total area, but its time cost has been decreased to about 1/6

Published
2012

48. A coupled experiment with LICOM2 as the ocean component of CESM1.

Author

Zhang, Minghua, Lin, Pengfei, Liu, Hailong, Song, Mirong, Song, Yi, Wang, Fuchang, Xue, Wei, Li, Huimin, and Jiang, Jinrong

Abstract

In the present study, the LASG/IAP Climate system Ocean Model version 2 (LICOM2) was implemented to replace the original ocean component in the Community Earth System Model version 1.0.4 (CESM1) to form a new coupled model referred to as CESM1+LICOM2. The simulation results from a 300-yr preindustrial experiment by using this model were evaluated against both observations and the Flexible Global Ocean-Atmosphere-Land System Model with grid-atmospheric model version 2 (FGOALS-g2). It was found that CESM1+LICOM2 simulates well the mean features of the ocean, sea ice, and atmosphere, relative to models used in the Coupled Model Intercomparison Experiment (CMIP5), when compared with observations. The spatial distribution of SST bias in CESM1+LICOM2 is similar to that in the Community Climate System Model version 4 (CCSM4). The simulated climate variabilities, such as ENSO and Pacific decadal oscillation, are also reasonably simulated when compared with observations. The successful implementation of LICOM2 in the CESM1 framework greatly enhances the capability of LICOM2 in conducting high-resolution simulations and model tuning. Compared with FGOALS-g2, the simulations of both SST and Atlantic meridional overturning circulation are significantly improved in CESM1+LICOM2. The former can be mainly attributed to the atmospheric model, and the latter to the improvement in the parameterization of diapycnal mixing. The study provides a base to further improve the present version of LICOM and its functionalities in the coupled model FGOALS at both low and high resolutions. [ABSTRACT FROM AUTHOR]

Published
2016
Full Text
View/download PDF

50. Portably parallel construction of a configuration-interaction wave function from a matrix-product state using the Charm++ framework.

Author

Wang T, Ma Y, Zhao L, and Jiang J

Abstract

The construction of configuration-interaction (CI) expansions from a matrix product state (MPS) involves numerous matrix operations and the skillful sampling of important configurations in a large Hilbert space. In this work, we present an efficient procedure for constructing CI expansions from MPS employing the parallel object-oriented Charm++ programming framework, upon which automatic load-balancing and object migrating facilities can be employed. This procedure was employed in the MPS-to-CI utility (Moritz et al., J. Chem. Phys. 2007, 126, 224109), the sampling-reconstructed complete active-space algorithm (SR-CAS, Boguslawski et al., J. Chem. Phys. 2011, 134, 224101), and the entanglement-driven genetic algorithm (EDGA, Luo et al., J. Chem. Theory Comput. 2017, 13, 4699). It enhances productivity and allows the sampling programs to evolve to their population-expansion versions, for example, EDGA with population expansion (PE-EDGA). Further, examples of 1,2-dioxetanone and firefly dioxetanone anion (FDO - ) molecules demonstrated the following: (a) parallel efficiencies can be persistently improved by simply by increasing the proportions of the asynchronous executions and (b) a sampled CAS-type CI wave function of a bi-radical-state FDO - molecule utilizing the full valence (30e,26o) active space can be constructed within a few hours with using thousands of cores., (© 2020 Wiley Periodicals LLC.)

Published
2020
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results