基于标签关联的多标签演化超网络

针对多标签学习中如何有效挖掘利用高阶标签关联的问题,提出了一种基于标签关联的多标签演化超网络模型.该模型通过输入任意多标签学习方法的预测结果,利用超边表征挖掘高阶标签关联,并综合标签关联和特征信息作为最终的预测结果.与3种传统多标签学习方法在6个多标签数据集上的对比实验表明,本文提出模型不仅能够有效提升多个传统多标签学习方法的性能,而且能够提供具有良好可读性的学习结果.     

Graph regularized low-rank feature mapping for multi-label learning with application to image annotation

Automatic image annotation has emerged as a hot research topic in the last two decades due to its application in social images organization. Most studies treat image annotation as a typical multi-label classification problem, where the shortcoming of this approach lies in that in order to a learn reliable model for label prediction, it requires sufficient number of training images with accurate annotations. Being aware of this, we develop a novel graph regularized low-rank feature mapping for image annotation under semi-supervised multi-label learning framework. Specifically, the proposed method concatenate the prediction models for different tags into a matrix, and introduces the matrix trace norm to capture the correlations among different labels and control the model complexity. In addition, by using graph Laplacian regularization as a smooth operator, the proposed approach can explicitly take into account the local geometric structure on both labeled and unlabeled images. Moreover, considering the tags of labeled images tend to be missing or noisy, we introduce a supplementary ideal label matrix to automatically fill in the missing tags as well as correct noisy tags for given training images. Extensive experiments conducted on five different multi-label image datasets demonstrate the effectiveness of the proposed approach.     

Learning discriminative representations for multi-label image recognition

Multi-label recognition is a fundamental, and yet is a challenging task in computer vision. Recently, deep learning models have achieved great progress towards learning discriminative features from input images. However, conventional approaches are unable to model the inter-class discrepancies among features in multi-label images, since they are designed to work for image-level feature discrimination. In this paper, we propose a unified deep network to learn discriminative features for the multi-label task. Given a multi-label image, the proposed method first disentangles features corresponding to different classes. Then, it discriminates between these classes via increasing the inter-class distance while decreasing the intra-class differences in the output space. By regularizing the whole network with the proposed loss, the performance of applying the well-known ResNet-101 is improved significantly. Extensive experiments have been performed on COCO-2014, VOC2007 and VOC2012 datasets, which demonstrate that the proposed method outperforms state-of-the-art approaches by a significant margin of 3.5% on large-scale COCO dataset. Moreover, analysis of the discriminative feature learning approach shows that it can be plugged into various types of multi-label methods as a general module.     

一种适应于非完备标签数据和标签关联性的多标签分类方法

多标签分类已在很多领域得到了实际应用,所用标签大多具有很强的关联性,甚至存在非完备标签或部分标签遗失。然而,现有的多标签分类算法难以同时处理这两种情况。基于此,提出一种新的概率模型处理方法,实现同时对具有标签关联性和遗失标签情况进行多标签分类。该方法可以自动获知和掌握多标签的关联性。此外,通过整合遗失的标签信息,该方法能够提供一个自适应策略来处理遗失的标签。在完备标签和非完备标签的数据上进行实验,结果表明,与现有的多标签分类算法相比,提出的方法得到了较好的分类预测评价值。     

一种基于全局和局部标记相关性的多标记分类算法

多标记学习用于处理一个示例同时与多个类别标记相关的问题.在多标记学习中,标记相关性能够显著提升学习算法的性能.大多数现有的多标记学习算法在利用标记的相关性时,要么只使用被所有示例所共享的全局标记相关性,要么就使用局部标记相关性,它们认为不同簇中的示例应该存在不同的标记相关性.本文中,我们提出了一种同时利用全局和局部标记相关性的多标记学习算法,从而为学习进程提供更全面的标记信息.在计算全局和局部标记相关性时,我们使用了余弦相似性来获取不同标记之间的正相关性和负相关性,这样有助于我们进一步实现更可靠的多标记学习.我们在多种类型的数据集上进行了广泛的对比实验来验证所提算法的有效性.实验结果表明,该算法显著优于大多数对比算法,展现出其在多标记学习中的突出性能.     

哈希快速多标记学习算法

本文针对多标记学习耗时大、很难处理大规模数据的问题,提出了一种哈希快速多标记学习算法（HFMLL）,该算法将哈希算法与多标记学习算法结合,采用局部敏感哈希算法快速获得每个样本的近邻样本,并通过最小独立置换的MinHash算法快速找到每个标记的相关标记,根据其近邻样本及相关标记的信息,运用最大后验概率准则来预测新样本的标记集。实验表明HFMLL 算法在保持较高分类性能的情况下,算法速度明显优于目前的多标记算法,可以广泛应用于大规模的数据集。      

基于稀疏编码和集成学习的多示例多标记图像分类方法

 该文基于稀疏编码和集成学习提出了一种新的多示例多标记图像分类方法。首先,利用训练包中所有示例学习一个字典,根据该字典计算示例的稀疏编码系数;然后基于每个包中所有示例的稀疏编码系数计算包特征向量,从而将多示例多标记问题转化为多标记问题;最后利用多标记分类算法进行求解。为了提高分类器的泛化能力,对多个分类器进行集成。在多示例多标记图像数据集上的实验结果表明所提方法与其它方法相比有更好的性能。     

基于矩阵分解和注意力多任务学习的客服投诉工单分类

投诉工单自动分类是通信运营商客服数字化、智能化发展的要求。客服投诉工单的类别有多层,每一层有多个标签,层级之间有所关联,属于典型的层次多标签文本分类问题,现有解决方法大多数基于分类器同时处理所有的分类标签,或者对每一层级分别使用多个分类器进行处理,忽略了层次结构之间的依赖。提出了一种基于矩阵分解和注意力的多任务学习的方法(MF-AMLA),处理层次多标签文本分类任务。在通信运营商客服场景真实投诉工单分类数据下,与该场景常用的机器学习算法和深度学习算法的Top1F1值相比分别最大提高了21.1%和5.7%。已在某移动运营商客服系统上线,模型输出的正确率97%以上,对客服坐席单位时间的处理效率提升22.1%。     

基于迁移学习与权重支持向量机的图像多标签标注算法

为解决图像的多标签自动标注中标签不平衡性的问题,提出了一种基于迁移学习与权重支持向量机的图像自动标注方法。为了解决所选数据集规模较小无法训练出最优的卷积神经网络的问题,文中采用迁移学习的方法,将通过Imagenet数据集训练出的Alexnet的参数迁移到文中所用的卷积神经网络模型中,并对最后一层全连接层进行微调,利用多标签分类多合页损失函数构成多分类的支持向量机。最后,文中对低频标签进行权重排序以得到图像的多标签标注结果。在Corel-5k、Esp-Game和IAPR-TC12共3个数据集上进行了实验,权重支持向量机获得的平均召回率分别提升了10%、9%和6%,低频标签对其平均精确率均提升了12%。实验结果表明,基于迁移学习的权重支持向量机的图像多标签标注方法可在有效提高数据集的召回率的同时提升低频标签的平均精确度。     

非平衡化标签补全核极限学习机多标签学习

目前众多的研究者通常直接将标签置信度矩阵作为先验知识直接加入到分类模型中,并没有考虑未标注先验知识对标签集质量的影响.基于此,引入非平衡参数的方法,将先验知识获得的基础置信度矩阵进行非平衡化,从而提出一种非平衡化的标签补全的核极限学习机多标签学习算法（KELM-NeLC）：首先使用信息熵计算标签之间的相关关系得到标签置信度矩阵,然后利用非平衡参数方法对基础的标签置信度矩阵进行改进,构建出一个非平衡的标签补全矩阵,最后为了学习获得更加准确的标签置信度矩阵,将非平衡化的标签补全矩阵与核极限学习机进行联合学习,依此解决多标签分类问题.提出的算法在公开的多个基准多标签数据集中的实验结果表明,KELM-NeLC算法较其他对比的多标签学习算法有一定优势,使用统计假设检验进一步说明所提出算法的有效性.     

应用梯度提升树的小区域无线网络多标签流量预测

当前无线网络流量地理分布不均且可用网络资源有限,因而开展拆闲补忙工作极为必要。为合理投放无线网络资源以保证网络性能,提出了一种针对小区域范围的多标签流量预测算法。该算法结合历史流量信息,根据无线用户偏好特性建立多标签流量预测模型,并通过梯度提升树算法得到预测结果。仿真结果表明,相比于广泛应用的移动平均自回归(Autoregressive Integrated Moving Average, ARIMA)和神经网络预测方法,多标签预测模型在对小区域突发性流量的预测上具有很大的优越性。     

Multi-label text classification model based on semantic embedding

Text classification means to assign a document to one or more classes or categories according to content. Text classification provides convenience for users to obtain data. Because of the polysemy of text data, multi-label classification can handle text data more comprehensively. Multi-label text classification become the key problem in the data mining. To improve the performances of multi-label text classification, semantic analysis is embedded into the classification model to complete label correlation analysis, and the structure, objective function and optimization strategy of this model is designed. Then, the convolution neural network (CNN) model based on semantic embedding is introduced. In the end, Zhihu dataset is used for evaluation. The result shows that this model outperforms the related work in terms of recall and area under curve (AUC) metrics.     

一种基于多阶邻居的网络环境下多标签分类算法

随着标签分类应用的增长,社交网络环境下多标签分类已成为一个重要的数据挖掘研究领域.关系分类模型基于一阶邻居做标签分类,其性能优于传统的多标签分类器.但现有的关系分类模型也存在问题：第一,仅利用一阶邻居做分类,未能充分使用邻居信息.第二,网络数据通常包含大量不连通的孤立部分,其标签无法利用现有的关系分类模型分类.考虑基于共引规则为非孤立节点挖掘二阶邻居和基于节点特征向量相似度为孤立节点挖掘高阶邻居,本文提出一种新的基于多阶邻居的网络数据多标签分类算法,称为MORN算法.在多个真实数据集上将MORN与现有的关系分类模型作对比,实验表明,MORN算法能够学习到更多节点的标签且精度优于传统关系分类方法.     

基于非对称监督深度离散哈希的图像检索

哈希广泛应用于图像检索任务。针对现有深度监督哈希方法的局限性,该文提出了一种新的非对称监督深度离散哈希(ASDDH)方法来保持不同类别之间的语义结构,同时生成二进制码。首先利用深度网络提取图像特征,根据图像的语义标签来揭示每对图像之间的相似性。为了增强二进制码之间的相似性,并保证多标签语义保持,该文设计了一种非对称哈希方法,并利用多标签二进制码映射,使哈希码具有多标签语义信息。此外,引入二进制码的位平衡性对每个位进行平衡,鼓励所有训练样本中的–1和+1的数目近似。在两个常用数据集上的实验结果表明,该方法在图像检索方面的性能优于其他方法。     

Multi-output regression using a locally regularised orthogonalleast-squares algorithm

The paper considers data modelling using multi-output regression models. A locally regularised orthogonal least-squares (LROLS) algorithm is proposed for constructing sparse multi-output regression models that generalise well. By associating each regressor in the regression model with an individual regularisation parameter, the ability of the multi-output orthogonal least-squares (OLS) model selection to produce a parsimonious model with a good generalisation performance is greatly enhanced     

射频识别的技术特征分析

简要阐述了射频识别技术的组成原理,分析了其技术特征,主要包括工作频率、工作方式、射频标签存储容量、数据传输速率、读写距离、多标签识别能力和安全性能等,提出了今后进一步研究的方向。     

直接线性图嵌入算法及其在人脸识别中的应用

图嵌入算法使用无向有权图来描述数据集的流形结构,目前许多流形学习算法都可统一到这个框架下。线性图嵌入算法(LGE)在高维小样本应用中往往会遇到的奇异值问题,因此需把数据集预先投影到PCA子空间,往往会丢失了一些有用的信息。本文提出了一种直接的线性图嵌入算法(DLGE),可直接从原始数据集中提取特征。此外DLGE算法相对于基于迭代的正交化算法,在最小二乘意义下对截断的征向量进行正交化处理,计算简便有效。在多个人脸数据库库上的仿真结果表明,相对于传统算法,DLGE算法具有更强的人脸表征能力,更好的分类性能,且更加鲁棒。     

Efficient eigenspace-based array signal processing using multipleshift-invariant subarrays

This paper deals with the construction of eigensubspaces for adaptive array signal processing. An efficient technique for extracting the eigensubspaces spanned by the data vector received by an N-element adaptive array is presented. We first decompose the original array into several subarrays with multiple shift invariances and find the eigensubspaces corresponding to each of the subarrays. By solving a least-squares (LS) or total least-squares (TLS) problem, the signal and noise subspaces corresponding to the original array can be found from the eigensubspaces spanned by the subarray data vectors. Hence, there is no need to perform the eigenvalue decomposition of the N×N correlation matrix of the received data vector. The proposed technique significantly reduces the required computational complexity as compared to the conventional eigenspace-based (ESB) methods. In conjunction with the spatial smoothing scheme or a proposed cross-correlation method, this technique can also deal with the case of coherent signals. The effectiveness of the proposed technique is demonstrated by several computer simulations     

Modeling of the Electrical Conductivity of DNA

The authors have developed a PSpice model of the electrical behavior of DNA molecules for use in nanoelectronic circuit design. To describe the relationship between the current through DNA and the applied voltage we used published results of the direct measurements of electrical conduction through DNA molecules. The experimental dc current-voltage (-) curves show a nonlinear conduction mechanism as well as the existence of a temperature dependent semiconductive voltage gap. A weighted least-squares polynomial fit to the experimental data at one temperature, with fitted temperature dependent polynomial coefficient of the linear term, was used as a mathematical model of electrical behavior of DNA. An equivalent electrical circuit was created in PSpice in which DNA was modeled as a voltage-controlled current source described by the mathematical model that includes temperature dependence . PSpice simulations with this model generated - curves at other temperatures that were in excellent agreement with the corresponding experimental data (average deviation 5%). This is important because having models of DNA molecules in the form of equivalent electronic circuits would be useful in the design of nanoelectronic circuits and devices.     

数据驱动的自动化机器学习流程生成方法

为有效解决自动化机器学习的数据处理时间较长等问题，文中提出了数据驱动的自动化机器学习流程生成方法。该方法的数据集匹配模块通过余弦相似性来计算数据集之间的相关性，获取和当前数据集时间匹配程度最高的数据集，并将其和历史数据搜索结果相结合，先选择最优选择动作，扩展和选择MCTS节点。该方法引入了服务关联关注约束模型，可以依据历史数据对所有的存在关联约束强度的选择动作进行指导，生成自动化的机器学习流程。测试结果显示，该方法对不同数据集的学习流程生成耗时均低于580s，生成的流程对单标签和多标签的数据分类结果均在0.92以上，数据完整度均高于95.5%，应用性良好。