检索结果-南通市图书馆

Multi-label active learning by model guided distribution matching

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2016年第5期10卷 845-855页

作者： Nengneng GAO sheng-jun huang Songcan CHEN College of Computer Science and Technology Nanjing University of Aeronautics and AstronauticsCollaborative Innovation Center of Novel Software Technology and Industrialization Nanjing 211106 China

Multi-label learning is an effective framework for learning with objects that have multiple semantic labels, and has been successfully applied into many real-world tasks, In contrast with traditional single-label learning, the cost of la- beling a multi-label example is rather high, thus it becomes an important task to train an effective multi-label learning model with as few labeled examples as possible. Active learning, which actively selects the most valuable data to query their labels, is the most important approach to reduce labeling cost. In this paper, we propose a novel approach MADM for batch mode multi-label active learning. On one hand, MADM exploits representativeness and diversity in both the feature and label space by matching the distribution between labeled and unlabeled data. On the other hand, it tends to query predicted positive instances, which are expected to be more informative than negative ones. Experiments on benchmark datasets demonstrate that the proposed approach can reduce the labeling cost significantly.

关键词： multi-label learning batch mode active learning distribution matching

A Deep Model for Partial Multi-label Image Classification with Curriculum-based Disambiguation

在线全文

学校读者我要写书评

暂无评论

Machine Intelligence Research 2024年第4期21卷 801-814页

作者： Feng Sun Ming-Kun Xie sheng-jun huang MIIT Key Laboratory of Pattern Analysis and Machine Intelligence College of Computer Science and TechnologyNanjing University of AeronauticsandAstronauticsNanjing211106China

In this paper,we study the partial multi-label(PML)image classification problem,where each image is annotated with a candidate label set consisting of multiple relevant labels and other noisy *** PML methods typically design a disambiguation strategy to filter out noisy labels by utilizing prior knowledge with extra assumptions,which unfortunately is unavailable in many real ***,because the objective function for disambiguation is usually elaborately designed on the whole training set,it can hardly be optimized in a deep model with stochastic gradient descent(SGD)on *** this paper,for the first time,we propose a deep model for PML to enhance the representation and discrimination *** the one hand,we propose a novel curriculum-based disambiguation strategy to progressively identify ground-truth labels by incorporating the varied difficulties of different *** the other hand,consistency regularization is introduced for model training to balance fitting identified easy labels and exploiting potential relevant *** experimental results on the commonly used benchmark datasets show that the proposed method significantlyoutperforms the SOTA methods.

关键词： Partial multi-label image classification curriculum-based disambiguation consistency regularization label difficulty candidatelabel set.

Relative difficulty distillation for semantic segmentation

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第9期67卷 126-145页

作者： Dong LIANG Yue SUN Yun DU Songcan CHEN sheng-jun huang MIIT Key Laboratory of Pattern Analysis and Machine Intelligence College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Shenzhen Research Institute Nanjing University of Aeronautics and Astronautics

Current knowledge distillation(KD) methods primarily focus on transferring various structured knowledge and designing corresponding optimization goals to encourage the student network to imitate the output of the teacher network. However, introducing too many additional optimization objectives may lead to unstable training, such as gradient conflicts. Moreover, these methods ignored the guidelines of relative learning difficulty between the teacher and student networks. Inspired by human cognitive science, in this paper, we redefine knowledge from a new perspective — the student and teacher networks' relative difficulty of samples, and propose a pixel-level KD paradigm for semantic segmentation named relative difficulty distillation(RDD). We propose a two-stage RDD framework: teacher-full evaluated RDD(TFE-RDD) and teacher-student evaluated RDD(TSE-RDD). RDD allows the teacher network to provide effective guidance on learning focus without additional optimization goals, thus avoiding adjusting learning weights for multiple losses. Extensive experimental evaluations using a general distillation loss function on popular datasets such as Cityscapes, Cam Vid, Pascal VOC, and ADE20k demonstrate the effectiveness of RDD against state-ofthe-art KD methods. Additionally, our research showcases that RDD can integrate with existing KD methods to improve their upper performance bound. Codes are available at https://***/sunyueue/***.

关键词： knowledge distillation semantic segmentation relative difficulty sample weighting prediction discrepancy

Incremental Multi-Label Learning with Active Queries

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2020年第2期35卷 234-246页

作者： sheng-jun huang Guo-Xiang Li Wen-Yu huang Shao-Yuan Li College of Computer Science and Technology Nanjing University of Aeronautics and AstronauticsNanjing 211106China Key Laboratory of Pattern Analysis and Machine Intelligence Ministry of Industry and Information Technology Nanjing University of Aeronautics and AstronauticsNanjing 211106China Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing University Nanjing 210023China

In multi-label learning,it is rather expensive to label instances since they are simultaneously associated with multiple ***,active learning,which reduces the labeling cost by actively querying the labels of the most valuable data,becomes particularly important for multi-label learning.A good multi-label active learning algorithm usually consists of two crucial elements:a reasonable criterion to evaluate the gain of querying the label for an instance,and an effective classification model,based on whose prediction the criterion can be accurately *** this paper,we first introduce an effective multi-label classification model by combining label ranking with threshold learning,which is incrementally trained to avoid retraining from scratch after every *** on this model,we then propose to exploit both uncertainty and diversity in the instance space as well as the label space,and actively query the instance-label pairs which can improve the classification model *** experiments on 20 datasets demonstrate the superiority of the proposed approach to state-of-the-art methods.

关键词： active learning multi-label learning uncertainty diversity

Film thickness effect on the performance of small molecular solar cell

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Optoelectronics Letters 2008年第5期4卷 321-323页

作者： LIN Hui YU jun-sheng huang Jiang JIANG Ya-dong State Key Laboratory of Electronic Thin Films and Integrated Devices School of Optoelectronic Information University of Electronic Science and Technology of China （UESTC） Chengdu 610054 China

An efficient organic photovoltaic （OPV） cell with an indium-tin-oxide/CuPc/C60/Ag structure has been investigated by changing the film thickness of organic layers. A high olin-circuit voltage （Yoc） of 0.5 V, a short-circuit current density （Jsc） of 5.81 mA/cm^2, and a high power conversion efficiency （ηp） of 1.2% were achieved at an optimum film thickness. The results demonstrate that material thickness is an important factor to cell optimization, especially for maximizing the absorption rate as will as reducing the cell resistance. Experimental results also indicate that the power conversion efficiency increases from 1.2% to 1.54% as a BCP exciton blocking layer of 10 nm is introduced.

关键词：油膜厚度分子太阳能电池有机层 OPV电池

在线全文

学校读者我要写书评

暂无评论

Preface

Journal of Computer Science & Technology 2021年第3期36卷 588-589页

作者： Zhang Min-Ling huang sheng-jun Long Mingsheng School of Computer Science and Engineering Southeast University Nanjing China School of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Nanjing China School of Software Tsinghua University Beijing China

It is our great pleasure to announce the publication of this special section in JCST:“Learning from Small Samples”.Machine learning has achieved great success in various *** the rapid growth of model size as in deep networks,the learning models become more and more complex,typically requiring a large scale of training samples with label ***,in real-world applications,labeled data is usually *** it could be rather expensive to collect more labeled data because the labeling process is time consuming and requires domain expertise.

关键词： And has consuming

Hypoxia-preconditioned bone marrow-derived mesenchymal stem cells protect neurons from cardiac arrest-induced pyroptosis

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Neural Regeneration Research 2025年第4期20卷 1103-1123页

作者： Xiahong Tang Nan Zheng Qingming Lin Yan You Zheng Gong Yangping Zhuang Jiali Wu Yu Wang Hanlin huang jun Ke Feng Chen shengli Clinical Medical College of Fujian Medical University Fujian Medical UniversityFuzhouFujian ProvinceChina Department of Emergency Fujian Provincial HospitalFuzhouFujian ProvinceChina Fujian Provincial Key Laboratory of Emergency Medicine FuzhouFujian ProvinceChina The Second Department of Intensive Care Unit Fujian Provincial Hospital South BranchFuzhouFujian ProvinceChina

Cardiac arrest can lead to severe neurological impairment as a result of inflammation,mitochondrial dysfunction,and post-cardiopulmonary resuscitation neurological *** preconditioning has been shown to improve migration and survival of bone marrow–derived mesenchymal stem cells and reduce pyroptosis after cardiac arrest,but the specific mechanisms by which hypoxia-preconditioned bone marrow–derived mesenchymal stem cells protect against brain injury after cardiac arrest are *** this end,we established an in vitro co-culture model of bone marrow–derived mesenchymal stem cells and oxygen–glucose deprived primary neurons and found that hypoxic preconditioning enhanced the protective effect of bone marrow stromal stem cells against neuronal pyroptosis,possibly through inhibition of the MAPK and nuclear factor κB ***,we transplanted hypoxia-preconditioned bone marrow–derived mesenchymal stem cells into the lateral ventricle after the return of spontaneous circulation in an 8-minute cardiac arrest rat model induced by *** results showed that hypoxia-preconditioned bone marrow–derived mesenchymal stem cells significantly reduced cardiac arrest–induced neuronal pyroptosis,oxidative stress,and mitochondrial damage,whereas knockdown of the liver isoform of phosphofructokinase in bone marrow–derived mesenchymal stem cells inhibited these *** conclude,hypoxia-preconditioned bone marrow–derived mesenchymal stem cells offer a promising therapeutic approach for neuronal injury following cardiac arrest,and their beneficial effects are potentially associated with increased expression of the liver isoform of phosphofructokinase following hypoxic preconditioning.

关键词： bone marrow–derived mesenchymal stem cells cardiac arrest cardiac resuscitation hypoxic preconditioning liver isoform of phosphofructokinase mitochondria neuroinflammation oxidative stress pyroptosis reactive oxygen species

锯齿形氮化铝纳米带负阻器件电子输运性质研究

在线全文

学校读者我要写书评

暂无评论

原子与分子物理学报 2025年第2期42卷 85-90页

作者：黄绛雪黄启俊何进王豪常胜武汉大学物理科学与技术学院武汉430072

低维纳米带材料因为其结构的特异性而呈现出新奇的物理性质.修饰纳米带的边缘能够调制其电子性质.本文运用基于密度泛函理论(DFT)的第一性原理计算方法,探究了锯齿形氮化铝纳米带(ZAlNNR)单边氟化、单边氯化以及单边氢化的电子结构和输... 详细信息

低维纳米带材料因为其结构的特异性而呈现出新奇的物理性质.修饰纳米带的边缘能够调制其电子性质.本文运用基于密度泛函理论(DFT)的第一性原理计算方法,探究了锯齿形氮化铝纳米带(ZAlNNR)单边氟化、单边氯化以及单边氢化的电子结构和输运特性.研究表明:锯齿型氮化铝纳米带可以通过以上手段实现能带结构的半导体-金属转换.计算电子输运特性,我们发现三种边缘修饰的器件均呈现出负微分电阻效应,其中AlN-F器件有最大峰谷电流比(PVCR),达到了1.78×10^(7),是硅烯纳米器件以及黑磷纳米器件的10^(6)倍.值得一提的是,相比于H-AlN-Cove纳米器件,AlN-F器件能够在更小的偏压范围内实现高PVCR.该结果为锯齿形氮化铝纳米带在低功耗纳米器件中的应用提供了广泛的前景.

关键词：锯齿形氮化铝纳米带电子输运负微分电阻效应第一性原理

维普期刊数据库博看期刊

在线全文

学校读者我要写书评

暂无评论

一种快速车标定位方法

四川大学学报（自然科学版） 2004年第6期41卷 1167-1171页

作者：庄永杨红雨游志胜李贵俊黄戈四川大学计算机学院图形图象研究所成都610064

车标定位是车标识别系统中至关重要的一步.作者提出了一种快速的车标定位算法,即在已知车牌位置的情况下,利用车标位置垂直方向能量集中的特点进行滤波处理以及在粗定位矩形框中利用模板匹配的方法得到车标的准确位置.实验结果表明,该... 详细信息

车标定位是车标识别系统中至关重要的一步.作者提出了一种快速的车标定位算法,即在已知车牌位置的情况下,利用车标位置垂直方向能量集中的特点进行滤波处理以及在粗定位矩形框中利用模板匹配的方法得到车标的准确位置.实验结果表明,该方法定位准确率在95%以上,平均速度在150ms至200ms之间.

关键词：车标定位车标识别能量特征模板匹配