Search

Your search keyword '"Lin, Huawei"' showing total 132 results
Start Over Author "Lin, Huawei"
132 results on '"Lin, Huawei"'

1. Token-wise Influential Training Data Retrieval for Large Language Models

Author

Lin, Huawei, Long, Jikai, Xu, Zhaozhuo, and Zhao, Weijie

Subjects
Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Cryptography and Security, Computer Science - Information Retrieval
Abstract

Given a Large Language Model (LLM) generation, how can we identify which training data led to this generation? In this paper, we proposed RapidIn, a scalable framework adapting to LLMs for estimating the influence of each training data. The proposed framework consists of two stages: caching and retrieval. First, we compress the gradient vectors by over 200,000x, allowing them to be cached on disk or in GPU/CPU memory. Then, given a generation, RapidIn efficiently traverses the cached gradients to estimate the influence within minutes, achieving over a 6,326x speedup. Moreover, RapidIn supports multi-GPU parallelization to substantially accelerate caching and retrieval. Our empirical result confirms the efficiency and effectiveness of RapidIn., Comment: Accepted to ACL 2024

Published
2024

2. Activation Template Matching Loss for Explainable Face Recognition

Author

Lin, Huawei, Liu, Haozhe, Li, Qiufu, and Shen, Linlin

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

Can we construct an explainable face recognition network able to learn a facial part-based feature like eyes, nose, mouth and so forth, without any manual annotation or additionalsion datasets? In this paper, we propose a generic Explainable Channel Loss (ECLoss) to construct an explainable face recognition network. The explainable network trained with ECLoss can easily learn the facial part-based representation on the target convolutional layer, where an individual channel can detect a certain face part. Our experiments on dozens of datasets show that ECLoss achieves superior explainability metrics, and at the same time improves the performance of face verification without face alignment. In addition, our visualization results also illustrate the effectiveness of the proposed ECLoss., Comment: 13 pages, 7 figures, 5 tables

Published
2022

3. Research on Gender-related Fingerprint Features

Author

Qi, Yong, Li, Yanping, Lin, Huawei, Chen, Jiashu, and Lei, Huaiguang

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Fingerprint is an important biological feature of human body, which contains abundant gender information. At present, the academic research of fingerprint gender characteristics is generally at the level of understanding, while the standardization research is quite limited. In this work, we propose a more robust method, Dense Dilated Convolution ResNet (DDC-ResNet) to extract valid gender information from fingerprints. By replacing the normal convolution operations with the atrous convolution in the backbone, prior knowledge is provided to keep the edge details and the global reception field can be extended. We explored the results in 3 ways: 1) The efficiency of the DDC-ResNet. 6 typical methods of automatic feature extraction coupling with 9 mainstream classifiers are evaluated in our dataset with fair implementation details. Experimental results demonstrate that the combination of our approach outperforms other combinations in terms of average accuracy and separate-gender accuracy. It reaches 96.5% for average and 0.9752 (males)/0.9548 (females) for separate-gender accuracy. 2) The effect of fingers. It is found that the best performance of classifying gender with separate fingers is achieved by the right ring finger. 3) The effect of specific features. Based on the observations of the concentrations of fingerprints visualized by our approach, it can be inferred that loops and whorls (level 1), bifurcations (level 2), as well as line shapes (level 3) are connected with gender. Finally, we will open source the dataset that contains 6000 fingerprint images, Comment: 11 pages, 6 figures, 4 tables

Published
2021

12. Characterization of the extracellular proteases from Bacillus inaquosorum strain E1‐8 and its application in the preparation of hydrolysates from plant and animal proteins with antioxidant, antifreeze and anti‐browning properties.

Author

Liu, Zhiyun, Lin, Huawei, Zhu, Xiaolong, Wu, Xueying, Wu, Chenxi, Obajemihi, Obafemi Ibitayo, Liu, Xinyi, Su, Wenrui, Liu, Guangchao, Li, Yang, Xu, Xingfeng, Yang, Jie, and Sun, Qingjie

Abstract

BACKGROUND RESULTS CONCLUSION Bacillus inaquosorum strains is widely recognized for their plant‐growth‐promoting and biocontrol capabilities, yet their roles in protease production remain unclear. The present study aimed to comprehensively assess the protease‐producing performance of B. inaquosorum strain E1‐8, at the same time as exploring the novel application of agricultural Bacillus proteases in the preparation of protein hydrolysates for fresh‐cut fruits preservation.First, genomic sequencing revealed the diversity of E1‐8 proteases, indicating 15 putative extracellular proteases. Subsequently, the fermentation conditions for E1‐8 protease production were optimized, with sweet potato powder and soybean meal identified as the most suitable carbon and nitrogen sources, respectively, resulting in a maximum protease activity of 321.48 U mL−1. Upon culturing the strain under these optimized conditions, only an S8 family serine protease and an M48 family metalloprotease were revealed by secretomic analysis and protease inhibitor assays. Additionally, the optimal protease conditions for generating protein hydrolysates from soy, pea, fish and porcine proteins were determined. The molecular weight of the hydrolysates primarily ranged from 2000 to 180 Da, with a total of 17 amino acids identified. The application of these hydrolysates demonstrated a 2,2‐diphenyl‐1‐picrylhydrazyl (i.e. DPPH) scavenging activity ranging from 58.64% to 84.12%, significantly reducing of the melting peaks and the freezing points. Furthermore, the browning index of apple slices stored at 4 °C decreased by 14.81% to 22.15% on the second day, and similar effects were observed in fresh‐cut banana stored at 4 °C for 7 days.The protein hydrolysates obtained exhibit remarkable antioxidant, antifreeze and anti‐browning properties for fresh‐cut fruits. © 2024 Society of Chemical Industry. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

19. Effects of exercise training under hypoxia versus normoxia on cognitive function in clinical and non-clinical populations: A systematic review and meta-analysis

Author

Lin, Yanting, primary, Yan, Jiamin, additional, Guo, Xiaoqin, additional, Lin, Huawei, additional, Ruan, Chendong, additional, Dai, Yaling, additional, Wang, Sinuo, additional, Cao, Yajun, additional, Xiang, Qing, additional, Yang, Minguang, additional, Liu, Weilin, additional, and Chen, Lidian, additional

Published
2023
Full Text
View/download PDF

27. Additional file 1 of Electroacupuncture protective effects after cerebral ischemia are mediated through miR-219a inhibition

Author

Dai, Yaling, Wang, Sinuo, Yang, Minguang, Zhuo, Peiyuan, Ding, Yanyi, Li, Xiaoling, Cao, Yajun, Guo, Xiaoqin, Lin, Huawei, Tao, Jing, Chen, Lidian, and Liu, Weilin

Abstract

Figure S1. EA could improve the cognitive function of MCAO rats A. Schematic representation of electroacupuncture points. B. Timeline and stages of the intervention design. C. Neurological deficit scores before treatment. D. Neurological deficit scores 24 hours after treatment. E. Neurological deficit scores 7 days after treatment. F. Neurological deficit scores 14 days after treatment. G. Results from the step-down avoidance test for each group before treatment. H. Results from the step-down avoidance test for each group 14 days after treatment. I. Escape latency in the Morris water maze test in each group. J. Morris water maze - visible platform task. K. Percentage of time spent in the target quadrant. *VS Sham group, P

Published
2023
Full Text
View/download PDF

28. Nanoscale-Thick a‑Ga2O3 Films on GaN for High-Performance Self-Powered Ultraviolet Photodetectors.

Author

Lin, Huawei, Su, Qin, Mao, Ruixi, Cheng, Qian, Zhu, Deliang, Han, Shun, Fang, Ming, Liu, Wenjun, Cao, Peijiang, Lu, Youming, and Pawar, Dnyandeo

Abstract

A super-high-performance self-powered ultraviolet photodetector constructed on a p-GaN/a-Ga2O3 p-n heterojunction was fabricated by growing nanoscale-thick films of n-type a-Ga2O3 on p-type GaN film using magnetron sputtering at room temperature. This device exhibited a self-powered effect and dual-band UV detection capability. The device showed exceptional photodetection performance with a responsivity of 5.65 A/W, Ilight/Idark ratio of 2.03 × 102, and a detectivity of 8.46 × 1012 cm Hz1/2W–1 to 254 nm light irradiation and a responsivity of 43.8 A/W, Ilight/Idark ratio of 1.53 × 103 and a detectivity of 6.56 × 1013 cmHz1/2W–1 under 365 nm UV light irradiation of 50 μW/cm2 light intensity at 0 V, respectively. This outstanding photodetection performance might be ascribed to the distanced separation between photogenerated charge carriers that occurs due to the built-in electric field across the depletion region of the p-n junction of the GaN/a-Ga2O3 thin film. This study shows the GaN/a-Ga2O3 thin-film-based p-n heterojunction UV photodetector with both self-powered performance and dual-band UV photodetection is very promising for practical applications. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

36. Shanghai cognitive intervention of mild cognitive impairment for delaying progress with longitudinal evaluation-a prospective, randomized controlled study (SIMPLE): rationale, design, and methodology

Author

Lin, Yiqi, Li, Binyin, Tang, Huidong, Xu, Qun, Wu, Yuncheng, Cheng, Qi, Li, Chunbo, Xiao, Shifu, Shen, Lu, Tang, Weiguo, Yu, Hui, He, Naying, Lin, Huawei, Yan, Fuhua, Cao, Wenwei, Yang, Shilin, Liu, Ye, Zhao, Wei, Lu, Dong, Jiao, Bin, Xiao, Xuewen, Zhou, Lin, and Chen, Shengdi

Published
2018
Full Text
View/download PDF

40. Enhanced Medial Prefrontal Cortex and Hippocampal Activity Improves Memory Generalization in APP/PS1 Mice: A Multimodal Animal MRI Study

Author

Liu, Weilin, primary, Li, Jianhong, additional, Li, Le, additional, Zhang, Yuhao, additional, Yang, Minguang, additional, Liang, Shengxiang, additional, Li, Long, additional, Dai, Yaling, additional, Chen, Lewen, additional, Jia, Weiwei, additional, He, Xiaojun, additional, Lin, Huawei, additional, and Tao, Jing, additional

Published
2022
Full Text
View/download PDF

43. Self‐decoupling coupled‐fed patch antennas for 5G multiple‐input multiple‐output applications.

Author

Ge, Lei, Lin, Huawei, Ji, Yuan, Tam, Kam‐Weng, and Zhao, Yue

Subjects
*ANTENNAS (Electronics), *LINEAR antenna arrays, *ANTENNA arrays, *MICROSTRIP antenna arrays, *ANTENNA design, *MIMO systems
Abstract

Self‐decoupling coupled‐fed patch antennas are designed in this letter. The self‐decoupling mechanism is based on the weak‐field region formed by the T‐shaped probe feed and the radiating patch of the antenna element itself. When another identical antenna element is placed in the weak‐field region, high isolation between the two elements can be achieved without adding extra decoupling structures. To validate the decoupling mechanism, a dual‐element antenna array is initially built and an isolation level of better than 30 dB is achieved over the impedance band of 3.3–3.66 GHz with an element spacing of 0.53λ0. Next, a four‐element multiple‐input multiple‐output (MIMO) linear antenna array is presented to further verify the decoupling method and good isolation performance is also obtained, which demonstrates that the proposed designs are good candidates for 5G MIMO applications. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

46. Longitudinal tracing of neurochemical metabolic disorders in working memory neural circuit and optogenetics modulation in rats with vascular cognitive impairment

Author

Lin, Huawei, primary, Jin, Tingting, additional, Chen, Lewen, additional, Dai, Yaling, additional, Jia, Weiwei, additional, He, Xiaojun, additional, Yang, Minguang, additional, Li, Jianhong, additional, Liang, Shengxiang, additional, Wu, Jinsong, additional, Huang, Jia, additional, Chen, Lidian, additional, Liu, Weilin, additional, and Tao, Jing, additional

Published
2021
Full Text
View/download PDF

50. Longitudinal tracing of neurochemical metabolic disorders in working memory neural circuit and optogenetics modulation in rats with vascular cognitive impairment

Author

Lin, Huawei, primary, Jin, Tingting, additional, Chen, Lewen, additional, Dai, Yaling, additional, Jia, Weiwei, additional, He, Xiaojun, additional, Yang, Minguang, additional, Li, Jianhong, additional, Liang, Shengxiang, additional, Wu, Jinsong, additional, Huang, Jia, additional, Chen, Lidian, additional, Liu, Weilin, additional, and Tao, Jing, additional

Published
2020
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results