嫩BBB槡BBBB槡BBBB,四川少妇BBW搡BBBB槡BBBB,四川少妇BBB凸凸凸BBB,四川少妇搡BBW搡BBBB,擦老太BBB擦BBB擦BBB擦

2022

• Record 1 of
  
  Title:The Earth 2.0 space mission analysis and spacecraft design
  
  Author(s):Chen, Wen(1); Chen, Kun(1); Yang, Yingquan(1); Han, Xingbo(1); Bi, Xingzi(1); He, Tao(1); Duan, Xuliang(1); Huang, Jiangjiang(1); Liang, Hong(1); Zhang, Kuoxiang(1); Wang, Haoyu(1); Liu, Liu(1); He, Junwang(1); Qin, Genjian(1); Li, Jinsong(1); Wang, Tian(1); Ge, Jian(2); Zhang, Hui(2); Zhang, Yongshuai(2); Zhou, Dan(2); Zhang, Congcong(2); Tang, Zhenghong(2); Yu, Yong(2); Zang, Weicheng(3); Mao, Shude(3); Chen, Yonghe(4); Liu, Xiaohua(4); Song, Zongxi(5); Gao, Wei(5); Zhang, Hongfei(6); Wang, Jian(6)
  Source: Proceedings of SPIE - The International Society for Optical Engineering  Volume: 12180  Issue:   DOI: 10.1117/12.2629697  Published: 2022  
  Abstract:The Earth 2.0 (ET) mission is a Chinese next-generation space mission to detect thousands of Earth-sized terrestrial planets, including habitable Earth-like planets orbiting solar type stars (Earth 2.0s), cold low-mass planets, and free-floating planets. To meet the scientific goals, the ET spacecraft will carry six 30 cm diameter transit telescopes with each field of view of 500 square degrees, and one 35 cm diameter microlensing telescope with a field of view of 4 square degrees, monitor ~1.2M FGKM dwarfs in the original Kepler field and its neighboring fields continuously while monitoring over 30M stars in the Galactic bulge direction. The high precision transit observations require high photometry precision and pointing stability, which is the key drive for the ET spacecraft design. In this paper, details of the overall mission modeling and analysis will be presented. The spacecraft orbit, pointing strategy, stability requirements are presented, as well as the space-ground communication analysis. The ET spacecraft adopts an ultra-high photometry precision & high stable platform, largely inherited from other space science missions. The preliminary design of spacecraft which meets mission requirements is introduced, including the spacecraft overall configuration, observation modes, avionics architecture and development plan, which pays great attention to the pointing stability and huge volume science telemetry download. ? 2022 SPIE.
  Accession Number: 20230413449799
• Record 2 of
  
  Title:ET White Paper: To Find the First Earth 2.0
  
  Author(s):Ge, Jian(1); Zhang, Hui(1); Zang, Weicheng(2); Deng, Hongping(1); Mao, Shude(2,17); Xie, Ji-Wei(3); Liu, Hui-Gen(3); Zhou, Ji-Lin(3); Willis, Kevin(20); Huang, Chelsea(26); Howell, Steve B.(41,42); Feng, Fabo(5); Zhu, Jiapeng(1); Yao, Xinyu(1); Liu, Beibei(8); Aizawa, Masataka(5); Zhu, Wei(2); Li, Ya-Ping(1); Ma, Bo(4); Ye, Quanzhi(11,12); Yu, Jie(6); Xiang, Maosheng(7,17); Yu, Cong(4); Liu, Shangfei(4); Yang, Ming(3); Wang, Mu-Tian(3); Shi, Xian(1); Fang, Tong(1); Zong, Weikai(28); Liu, Jinzhong(13); Zhang, Yu(13); Zhang, Liyun(16); El-Badry, Kareem(36); Shen, Rongfeng(4); Tam, Pak-Hin Thomas(4); Hu, Zhecheng(4); Yang, Yanlv(4); Zou, Yuan-Chuan(14); Wu, Jia-Li(14); Lei, Wei-Hua(14); Wei, Jun-Jie(15); Wu, Xue-Feng(15); Sun, Tian-Rui(15); Wang, Fa-Yin(3); Zhang, Bin-Bin(3); Xu, Dong(17); Yang, Yuan-Pei(18); Li, Wen-Xiong(19); Xiang, Dan-Feng(2); Wang, Xiaofeng(2); Wang, Tinggui(9,10); Zhang, Bing(43); Jia, Peng(40); Yuan, Haibo(28); Zhang, Jinghua(17); Wang, Sharon Xuesong(2); Gan, Tianjun(2); Wang, Wei(14); Zhao, Yinan(24,25); Liu, Yujuan(14); Chen, Yonghe(21); Wei, Chuanxin(21); Kang, Yanwu(21); Yang, Baoyu(21); Qi, Chao(21); Liu, Xiaohua(21); Zhang, Quan(21); Zhu, Yuji(21); Zhou, Dan(1); Zhang, Congcong(1); Yu, Yong(1); Zhang, Yongshuai(1); Li, Yan(1,63,64,65,66); Tang, Zhenghong(1); Wang, Chaoyan(1); Wang, Fengtao(22); Li, Wei(22); Cheng, Pengfei(22); Shen, Chao(22); Li, Baopeng(22); Pan, Yue(22); Yang, Sen(22); Gao, Wei(22); Song, Zongxi(22); Wang, Jian(9); Zhang, Hongfei(9); Chen, Cheng(9); Wang, Hui(9); Zhang, Jun(9); Wang, Zhiyue(9); Zeng, Feng(9); Zheng, Zhenhao(9); Zhu, Jie(9); Guo, Yingfan(9); Zhang, Yihao(9); Li, Yudong(44); Wen, Lin(44); Feng, Jie(44); Chen, Wen(23); Chen, Kun(23); Han, Xingbo(23); Yang, Yingquan(23); Wang, Haoyu(23); Duan, Xuliang(23); Huang, Jiangjiang(23); Liang, Hong(23); Bi, Shaolan(28); Gai, Ning(30); Ge, Zhishuai(46); Guo, Zhao(29); Huang, Yang(18); Li, Gang(39); Li, Haining(17); Li, Tanda(28); Lu, Yuxi Lucy(37,38); Rix, Hans-Walter(7); Shi, Jianrong(17); Song, Fen(31); Tang, Yanke(30); Ting, Yuan-Sen(26,27); Wu, Tao(63,64,65,66); Wu, Yaqian(17); Yang, Taozhi(47); Yin, Qing-Zhu(45); Gould, Andrew(7,32); Lee, Chung-Uk(33); Dong, Subo(34); Yee, Jennifer C.(34); Shvartzvald, Yossi(35); Yang, Hongjing(2); Kuang, Renkun(2); Zhang, Jiyuan(2); Liao, Shilong(1); Qi, Zhaoxiang(1); Yang, Jun(44); Zhang, Ruisheng(3); Jiang, Chen(6); Ou, Jian-Wen(48); Li, Yaguang(49,54); Beck, Paul(50); Bedding, Timothy R.(49,54); Campante, Tiago L.(51,52); Chaplin, William J.(53,54,55); Christensen-Dalsgaard, J?rgen(54); García, Rafael A.(56); Gaulme, Patrick(6); Gizon, Laurent(6,57,58); Hekker, Saskia(59,60); Huber, Daniel(61); Khanna, Shourya(62); Mathur, Savita(67,68); Miglio, Andrea(53,70,71); Mosser, Beno?t(72); Ong, J.M. Joel(61,73)
  Source: arXiv  Volume:   Issue:   DOI: 10.48550/arXiv.2206.06693  Published: June 14, 2022  
  Abstract:The ET mission is a wide-field and ultra-high-precision photometric survey mission being developed in China. This mission is designed to measure, for the first time, the occurrence rate and the orbital distributions of Earth-sized planets. ET consists of seven 30 cm telescopes to be launched to the Earth-Sun's L2 point. Six of these are transit telescopes with a FOV of 500 square degrees. Staring in the direction that encompasses the original Kepler field for four continuous years, this monitoring will yield tens of thousands of transiting planets, including the elusive Earth twins orbiting solar-type stars. The seventh is a 30 cm microlensing telescope that will monitor an area of 4 square degrees toward the galactic bulge. Combined with simultaneous ground-based KMTNet observations, it will measure masses of hundreds of long-period and free-floating planets. Together, the transit and the microlensing telescopes will revolutionize our understanding of terrestrial planets across a large swath of orbital distances and free space. In addition, the survey data will also facilitate studies in the fields of asteroseismology, Galactic archaeology, time-domain sciences, and black holes in binaries. ? 2022, CC BY-NC-ND.
  Accession Number: 20220183176
• Record 3 of
  
  Title:Effective half-wavelength pitch optical phased array design for aliasing-free 2D beam steering
  
  Author(s):Lei, Yufang(1,2); Zhang, Lingxuan(1,2); Xue, Yulong(1,2); Ren, Yangming(1,2); Zhang, Qihao(1,2); Zhang, Wenfu(1,2); Sun, Xiaochen(1,2)
  Source: Applied Optics  Volume: 61  Issue: 32  DOI: 10.1364/AO.474504  Published: November 10, 2022  
  Abstract:We present a method to design an optical phased array (OPA) simultaneously realizing both narrow beam width and aliasing-free 2D beam steering without the need to arrange the antennas at actual half-wavelength pitch. The method realizes an effective half-wavelength pitch in one direction formed by location projection of the antennas. The distances between the antennas in the other direction can be sufficiently large to form an effective large aperture realizing narrow beam width without needing a long grating. The presented method is proven by both theory and numerical simulations to achieve an equivalent grating-lobe-free far field of an ordinary half-wavelength pitch design. One design example exhibits 180? steering with a minimal beam width of 0.4? * 0.032? and a sidelobe suppression ratio of >13 dB. Journal ? 2022 Optica Publishing Group.
  Accession Number: 20224713152145
• Record 4 of
  
  Title:Dynamic synopsis and storage algorithm based on infrared surveillance video
  
  Author(s):Li, Xuemei(1); Qiu, Shi(2); Song, Yang(3)
  Source: Infrared Physics and Technology  Volume: 124  Issue:   DOI: 10.1016/j.infrared.2022.104213  Published: August 2022  
  Abstract:Infrared surveillance video is difficult to watch quickly and store efficiently, a surveillance video synopsis and storage algorithm is proposed based on dynamic. On the basis of extracting moving targets, the constraints of time and space is broken to build an energy functional based on filling density to quickly display the video content on the premise of ensuring the monitoring video information. The Tube structure is formed by the moving target information, and the mapping relationship between the original video and the stored video is established. Image similarity from time and space dimensions is fully utilized to realize the storage of surveillance video. The space ratio between the stored information and the original video is less than 0.2. ? 2022 Elsevier B.V.
  Accession Number: 20222212185955
• Record 5 of
  
  Title:Fabrication and Spectroscopic Properties of Heavily Pr3+ Doped Selenide Chalcogenide Glass and Fiber for Mid-infrared Fiber Laser
  
  Author(s):Xu, Chen-Yu(1,2); Cui, Jian(1,2); Xu, Yan-Tao(1); Xiao, Xu-Sheng(1); Cui, Xiao-Xia(1); Guo, Hai-Tao(1,2)
  Source: Faguang Xuebao/Chinese Journal of Luminescence  Volume: 43  Issue: 6  DOI: 10.37188/CJL.20220088  Published: June 2022  
  Abstract:In order to develop a high gain medium for fiber lasers operating at 3-5 μm waveband，0-0. 4%（in weight）Pr3+ ions doped Ge12As20.8Ga4Se63.2 selenide chalcogenide glasses were prepared and the 0. 2%（in weight）Pr3+ ions doped one was successfully drawn into step-index double-cladding fiber with the lowest loss of 2. 95 dB/m@6. 58 μm by a multistage rod-in-tube method. The electron-probe measure microanalysis（EPMA），X-ray diffraction （XRD），differential scanning calorimeter（DSC），field emission transmission electron microscope（FE-TEM），trans? mission and mid-infrared fluorescence spectra were carried out to analyze the dispersion of Pr3+ ions in glass，the im? purity contents，thermal and optical changes caused by the Pr3+ ions’introduction. By analyzing the absorption and emission measurements of the serial glasses with the Judd-Ofelt theory，the Judd-Ofelt strength parameters，transi? tion probabilities，exited state lifetime，branching ratios，and emission cross-sections were also calculated. This sel? enide chalcogenide glass has high Pr3+ ions’solubility and emission characteristic，good thermal stability and fiber forming performance，indicating that it has potential to be used as mid-infrared laser working medium. ? 2022 Chines Academy of Sciences. All rights reserved.
  Accession Number: 20223212553301
• Record 6 of
  
  Title:Two-dimensional single-lobe Si photonic optical phased array with minimal antennas using a non-uniform large spacing array design
  
  Author(s):Xue, Yulong(1,2); Zhang, Qihao(1); Ren, Yangming(1,2); Lei, Yufang(1,2); Sun, Xiaochen(1,2); Zhang, Lingxuan(1)
  Source: Applied Optics  Volume: 61  Issue: 24  DOI: 10.1364/AO.463542  Published: August 20, 2022  
  Abstract:We report a two-dimensional Si photonic optical phased array (OPA) optimized for a large optical aperture with a minimal number of antennas while maintaining single-lobe far field. The OPA chip has an optical aperture of ～200 μm by 150 μm comprising a 9 × 9 antenna array. The two-dimensional spacings between these antennas are much larger than the wavelength and are highly non-uniform optimized by the genetic deep learning algorithm. The phase of each antenna is independently tunable by a thermo-optical phase shifter. The experimental results validate the design and exhibit a 0.39? × 0.41? beamwidth within the 3 dB steering range of 14? × 11? limited by the numerical aperture of the far-field camera system. The method can be easily extended to a larger aperture for narrower beamwidth and wider steering range. ? 2022 Optica Publishing Group.
  Accession Number: 20223712737101
• Record 7 of
  
  Title:Thermal Management Technologies Used for High Heat Flux Automobiles and Aircraft: A Review
  
  Author(s):Lv, Yi-Gao(1); Zhang, Gao-Peng(2); Wang, Qiu-Wang(1); Chu, Wen-Xiao(1)
  Source: Energies  Volume: 15  Issue: 21  DOI: 10.3390/en15218316  Published: November 2022  
  Abstract:In recent years, global automotive industries are going through a significant revolution from traditional internal combustion engine vehicles (ICEVs) to electric vehicles (EVs) for CO2 emission reduction. Very similarly, the aviation industry is developing towards more electric aircraft (MEA) in response to the reduction in global CO2 emission. To promote this technology revolution and performance advancement, plenty of electronic devices with high heat flux are implemented on board automobiles and aircraft. To cope with the thermal challenges of electronics, in addition to developing wide bandgap (WBG) semiconductors with satisfactory electric and thermal performance, providing proper thermal management solutions may be a much more cost-effective way at present. This paper provides an overview of the thermal management technologies for electronics used in automobiles and aircraft. Meanwhile, the active methods include forced air cooling, indirect contact cold plate cooling, direct contact baseplate cooling, jet impingement, spray cooling, and so on. The passive methods include the use of various heat pipes and PCMs. The features, thermal performance, and development tendency of these active and passive thermal management technologies are reviewed in detail. Moreover, the environmental influences introduced by vibrations, shock, acceleration, and so on, on the thermal performance and reliability of the TMS are specially emphasized and discussed in detail, which are usually neglected in normal operating conditions. Eventually, the possible future directions are discussed, aiming to serve as a reference guide for engineers and promote the advancement of the next-generation electronics TMS in automobile and aircraft applications. ? 2022 by the authors.
  Accession Number: 20224613126037
• Record 8 of
  
  Title:A Unified Perspective of Multi-level Cross-Modal Similarity for Cross-Modal Retrieval
  
  Author(s):Huang, Yingying(1); Wang, Quan(2); Zhang, Yipeng(1); Hu, Bingliang(3)
  Source: 2022 5th International Conference on Information Communication and Signal Processing, ICICSP 2022  Volume:   Issue:   DOI: 10.1109/ICICSP55539.2022.10050678  Published: 2022  
  Abstract:Cross-modal retrieval is an intelligent understanding task between cross-modal data, and it comes with challenges to measure the similarity between cross-modal data. Existing methods mainly learned a common space by feature-wise or label-based supervised learning. Still, feature-wise methods only focused on the interactions between pairs of cross-modal data and label-based supervised learning relied excessively on classification accuracy. In the same space, these methods cannot capture more comprehensive interaction between cross-mode data, that is, given a query, this query and the retrieved data exist one-to-many correspondence, and the similarity between the pair-wise data is the largest. Therefore, a unified perspective of multi-level cross-modal similarity (MCMS) is proposed for cross-modal retrieval. Core ideas of MCMS are as follows: 1) The local similarity between cross-modal data is integrated to enrich the fine-grained cross-modal information. 2) The similarity between common feature vector and label is designed to obtain one-to-many correspondences between cross-modal data. In addition, Normalize Discounted Cumulative Gain (NDCG) as the evaluation metric is first used to comprehensively evaluate the results of cross-modal retrieval. Extensive experiments demonstrate that MCMS has better performance in cross-modal retrieval tasks. ? 2022 IEEE.
  Accession Number: 20231113742249
• Record 9 of
  
  Title:Design and Ground Verification for Multispectral Camera on the Mars Tianwen-1 Rover
  
  Author(s):Yang, Jian-Feng(1); Liu, Da-Wei(2); Xue, Bin(1); Lyu, Juan(1); Liu, Jian-Jun(2); Li, Fu(1); Ren, Xin(2); Ge, Wei(1); Liu, Bin(2); Ma, Xiao-Long(1); Lyu, Bao-Gang(1); Ruan, Ping(1); Qiao, Wei-Dong(1); Lu, Di(1)
  Source: Space Science Reviews  Volume: 218  Issue: 3  DOI: 10.1007/s11214-022-00886-3  Published: April 2022  
  Abstract:As part of China’s first Mars exploration mission ‘Tianwen-1’, the Zhurong rover has successfully touched down on the surface of southern Utopia Planitia on May 15th 2021 and has been conducting surface operations for several months. A?multispectral camera (MSCam), as an important payload onboard the Zhurong rover, aims to acquire multispectral images to investigate the morphological characteristics and mineralogic properties of the Martian surface. In this study, a?detailed optimization design for the MSCam was carried out to achieve the abovementioned scientific objectives. The MSCam can perform multispectral imaging without chromatic aberration by utilizing eight narrow bandwidth filters made of glass of different thicknesses. Clear images of observation targets at different distances can be obtained by utilizing the six focal plane compensation lenses of varying thicknesses through the rotation of wheels. Calibration experiments, key specification tests and ground verification tests were also conducted in this study. Our results show that the pixel resolution of the MSCam can reach 0.146 mrad, the system static modulation transfer function (MTF) of the MSCam is better than 0.25@525?nm, and the signal-to-noise ratio (SNR) is higher than 40?dB, all of which allow clear imaging and accurate multispectral data acquisition of the targets. The high-resolution images obtained by the MSCam will provide detailed geological context for the data interpretation of other payloads on the rover, such as the Mars surface composition detector (MarSCoDe). The mineralogy information of the targets (e.g., fresh rock, dune) indicated by the MSCam multispectral data will also help to constrain the surface material composition of Mars. ? 2022, The Author(s), under exclusive licence to Springer Nature B.V.
  Accession Number: 20221611980797
• Record 10 of
  
  Title:Ship Detection in Remote Sensing Image Based on Dense RFB and LSTM
  
  Author(s):Zhang, Tao(1); Yang, XiaoGang(1); Lu, XiaoQiang(2); Lu, RuiTao(1); Zhang, ShengXiu(1)
  Source: National Remote Sensing Bulletin  Volume: 26  Issue: 9  DOI: 10.11834/jrs.20211042  Published: September 2022  
  Abstract:Deep learning method had get great progress in remote sensing ship target detection, however there are still two main shortcomings as follows. One is that remote sensing image targets have multi-scale and multidirectional characteristics, especially for ship targets which are arbitrarily densely arranged, while existing detection networks lack of interactions between high-level and low-level features and ignore the context semantic information, which leads to poor detection results. The other is that the background of remote sensing images is complex and easily affected by factors such as light and clouds, resulting in the imbalance of positive and negative samples for target detection. In order to solve the problems above, a multi-scale ship target detection algorithm based on Dense RFB and LSTM is proposed in this paper. Firstly, a Dense RFB feature enhance module (Dense RFB-FE) is designed, which adopts feature multiplexing and expanded convolution to simulate the human eye point of view mechanism to increase the feature experience without increasing the amount of calculation, enhancing the ability to extract feature of shallow network details. Secondly, a deep multi-scale feature pyramid fusion module (MFPF) is designed, drawing on the ideas of FPN and LSTM, using deconvolution and residual structure to fuse deep multi-scale features, filtering invalid feature information, effectively to extract deep semantic information and enhance the expressive ability of the network feature layer. Finally, a new loss function is designed, the focus classification loss function is added to effectively solve the problem of imbalance of positive and negative sample, improving the accuracy of ship target detection. Experiments on optical remote sensing image dataset show that the average detection accuracy of the proposed algorithm for ship targets reaches 81.98%, and the detection speed reaches 29.6fps, which reduces the false detection rate and missed detection rate of target detection to a certain extent. In addition, for ship targets that are blurred, occluded, and partially cropped, the detection effect of the algorithm in this paper is also better than that of the original classic algorithm, which shows that by fusing the semantic information of the feature layer and the detailed positioning information, the generalization ability and characterization of the feature can be improved, which improves the accuracy of ship target detection in remote sensing images. In the future, the algorithm will be further optimized for the problems of multi-scale and dense arrangement of ship targets in remote sensing images. The rotating boxes will be used to accurately position the ship to reduce the interference of complex backgrounds. At the same time, the remote sensing image ship target datasets will be expanded to improve the ship target detection capability of the optical remote sensing image. ? 2022 National Remote Sensing Bulletin. All rights reserved.
  Accession Number: 20224713139256
• Record 11 of
  
  Title:Optical Neuromorphic Processor at 11 TeraOPs/s based on Kerr Soliton Crystal Micro-combs
  
  Author(s):Tan, Mengxi(1); Xu, Xingyuan(2); Wu, Jiayang(1); Boes, Andreas(3); Corcoran, Bill(2); Nguyen, Thach G.(3); Chu, Sai T.(4); Little, Brent E.(5); Hicks, Damien G.(1,6); Morandotti, Roberto(7); Mitchell, Arnan(3); Moss, David J.(1)
  Source: 2022 Optical Fiber Communications Conference and Exhibition, OFC 2022 - Proceedings  Volume:   Issue:   DOI:   Published: 2022  
  Abstract:We demonstrate a universal optical vector convolutional accelerator operating at 11 Tera-OPS, generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously. We use the same hardware to form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 88% accuracy. Our approach is scalable and trainable for applications to unmanned vehicle and real-time video recognition. ? 2022 OSA.
  Accession Number: 20221812050726
• Record 12 of
  
  Title:Retrieving Water Quality Parameters from Noisy-Label Data Based on Instance Selection
  
  Author(s):Liu, Yuyang(1,2); Liu, Jiacheng(1,2); Zhao, Yubo(1); Wang, Xueji(1); Song, Shuyao(1,2); Liu, Hong(1); Yu, Tao(1,2)
  Source: Remote Sensing  Volume: 14  Issue: 19  DOI: 10.3390/rs14194742  Published: October 2022  
  Abstract:As an important part of the "air–ground" integrated water quality monitoring system, the inversion of water quality from unmanned airborne hyperspectral image has attracted more and more attention. Meanwhile, unmanned aerial vehicles (UAVs) have the characteristics of small size, flexibility and quick response, and can complete the task of water environment detection in a large area, thus avoiding the difficulty in obtaining satellite data and the limitation of single-point monitoring by ground stations. Most researchers use UAV for water quality monitoring, they take water samples back to library or directly use portable sensors for measurement while flying drones at the same time. Due to the UAV speed and route planning, the actual sampling time and the UAV passing time cannot be guaranteed to be completely synchronized, and there will be a difference of a few minutes. For water quality parameters such as chromaticity (chroma), chlorophyll-a (chl-a), chemical oxygen demand (COD), etc., the changes in a few minutes are small and negligible. However, for the turbidity, especially in flowing water body, this value of it will change within a certain range. This phenomenon will lead to noise error in the measured suspended matter or turbidity, which will affect the performance of regression model and retrieval accuracy. In this study, to solve the quality problem of label data in a flowing water body, an unmanned airborne hyperspectral water quality retrieval experiment was carried out in the Xiao River in Xi’an, China, which verified the rationality and effectiveness of label denoising analysis of different water quality parameters. To identify noisy label instances efficiently, we proposed an instance selection scheme. Furthermore, considering the limitation of the dataset samples and the characteristic of regression task, we build a 1DCNN model combining a self attention mechanism (SAM) and the network achieves the best retrieving performance on turbidity and chroma data. The experiment results show that, for flowing water body, the noisy-label instance selection method can improve retrieval performance slightly on the COD parameter, but improve greatly on turbidity and chroma data. ? 2022 by the authors.
  Accession Number: 20224212985351