A novel unsupervised person re-identification algorithm based on soft multi-label and compound attention model

To explore discriminative information fully and keep consistence of labels, an unsupervised person re-identification algorithm based on soft multi-label and compound attention model was proposed in this study. Based on learning of reference agent labels, soft multi-label was built by constructing a mapping model of targets and reference datasets. Later, soft multi-label was added into initial samples through deep convolutional network training to realize accurate labeling of targets and fine-grain classification of features under multi-camera scenes. In the training stage of the deep network, a compound attention mechanism is added between the convolution blocks to fuse the complementary information of the multiple channels features and the spaces domain features, therefore the potential discriminative information is explored. In addition, a weight fusion of distance loss function, label consistency loss function, and reference agent loss function was performed to distinguish hard negative pair set and realize matching of multi-camera labels. Since learning rate is the key influencing factor against the improvement of identification precision and training speed, a rectified adaptive moment estimation was adopted to achieve adaptive control of learning rate, accelerate training convergence of network and increase the robustness of the proposed algorithm. The proposed algorithm is proved by an experiment that it can increase identification precision significantly. The rank-1 of the proposed algorithm is at least 3.9% higher, and its mean average precision (mAP) is at least 4.7% higher compared to those of similar representative algorithms.

     

一种等量约束聚类的无监督蒸馏哈希图像检索方法

为了进一步降低无监督深度哈希检索任务中的伪标签噪声,提出了一种等量约束聚类的无监督蒸馏哈希图像检索方法。该方法主要分为两个阶段,在第一阶段中,主要对无标签图像进行软伪标签标注,用于第二阶段监督哈希特征学习,通过所提等量约束聚类算法,在软伪标签标注过程中可以有效降低伪标签中的噪声;在第二阶段中,主要对学生哈希网络进行训练,用于提取图像哈希特征。通过所提出的无监督蒸馏哈希方法,利用图像软伪标签指导哈希特征学习,进一步提高了哈希检索性能,实现了高效的无监督哈希图像检索。为了评估所提方法的有效性,在CIFAR-10、FLICKR25K和EuroSAT三个公开数据集上进行了实验,并与其他先进方法进行了比较。在CIFAR-10数据集上,与TBH方法相比,所提方法检索精度平均提高12.7%;在FLICKR25K数据集上,与DistillHash相比,所提方法检索精度平均提高1.0%;在EuroSAT数据集上,与ETE-GAN相比,所提方法检索精度平均提高16.9%。在三个公开数据集上进行的实验结果表明,所提方法能够实现高性能的无监督哈希检索,且对各类数据均有较好的适应性。     

基于低密度分割密度敏感距离的谱聚类算法

本文提出一种基于低密度分割密度敏感距离的谱聚类算法, 该算法首先使用低密度分割密度敏感距离计算相似度矩阵, 该距离测度通过指数函数和伸缩因子实现放大不同流形体数据间的距离和缩短同一流形体数据间距离的目的, 从而有效反映数据分布的全局一致性和局部一致性特征.另外, 算法通过增加相对密度敏感项来考虑数据的局部分布特征, 从而有效避免孤立噪声和"桥"噪声的影响.文中最后给出了基于SC (Scattering criteria)指标的k近邻图k值选取办法和基于谱熵贡献率的特征向量选取方法.实验部分, 讨论了参数选择对算法性能的影响并给出取值建议, 通过与其他流行谱聚类算法聚类结果的对比分析, 表明本文提出的基于低密度分割密度敏感距离的谱聚类算法聚类性能明显优于其他算法.     

Online recognition of handwritten music symbols

In this paper, we propose an effective online method to recognize handwritten music symbols. Based on the fact that most music symbols can be regarded as combinations of several basic strokes, the proposed method first classifies all the strokes comprising an input symbol and then recognizes the symbol based on the results of stroke classification. For stroke classification, we propose to use three types of features, which are the size information, the histogram of directional movement angles, and the histogram of undirected movement angles. When combining classified strokes into a music symbol, we utilize their sizes and spatial relation together with their combination. The proposed method is evaluated using two datasets including HOMUS, one of the largest music symbol datasets. As a result, it achieves a significant improvements of about 10% in recognition rates compared to the state-of-the-art method for the datasets. This shows the superiority of the proposed method in online handwritten music symbol recognition.     

Scene Detection in Videos Using Shot Clustering and Sequence Alignment

Video indexing requires the efficient segmentation of video into scenes. The video is first segmented into shots and a set of key-frames is extracted for each shot. Typical scene detection algorithms incorporate time distance in a shot similarity metric. In the method we propose, to overcome the difficulty of having prior knowledge of the scene duration, the shots are clustered into groups based only on their visual similarity and a label is assigned to each shot according to the group that it belongs to. Then, a sequence alignment algorithm is applied to detect when the pattern of shot labels changes, providing the final scene segmentation result. In this way shot similarity is computed based only on visual features, while ordering of shots is taken into account during sequence alignment. To cluster the shots into groups we propose an improved spectral clustering method that both estimates the number of clusters and employs the fast global k-means algorithm in the clustering stage after the eigenvector computation of the similarity matrix. The same spectral clustering method is applied to extract the key-frames of each shot and numerical experiments indicate that the content of each shot is efficiently summarized using the method we propose herein. Experiments on TV-series and movies also indicate that the proposed scene detection method accurately detects most of the scene boundaries while preserving a good tradeoff between recall and precision.     

Online flowchart understanding by combining max-margin Markov random field with grammatical analysis

Flowcharts are considered in this work as a specific 2D handwritten language where the basic strokes are the terminal symbols of a graphical language governed by a 2D grammar. In this way, they can be regarded as structured objects, and we propose to use a MRF to model them, and to allow assigning a label to each of the strokes. We use structured SVM as learning algorithm, maximizing the margin between true labels and incorrect labels. The model would automatically learn the implicit grammatical information encoded among strokes, which greatly improves the stroke labeling accuracy compared to previous researches that incorporated human prior knowledge of flowchart structure. We further complete the recognition by using grammatical analysis, which finally brings coherence to the whole flowchart recognition by labeling the relations between the detected objects.     

基于特征匹配和距离加权的蓝牙定位算法

针对传统iBeacon指纹定位技术中接收信号强度值（RSSI）波动较大、指纹库聚类复杂、存在较大跳变性定位误差等问题,提出一种基于排序特征匹配和距离加权的蓝牙定位算法。在离线阶段,该算法先对RSSI进行加权滑动窗处理,然后根据RSSI向量大小生成排序特征码等值,并与位置坐标等信息组成指纹信息,形成指纹库;在在线定位阶段,根据排序特征向量指纹匹配定位算法和基于距离的最优加权K最邻近法（WKNN）实现室内行人定位。在定位仿真实验中,该算法可以自动根据特征码进行聚类,从而降低了聚类的复杂度,能实现最大误差在0.952 m内的室内行人定位精度。     

基于Bayes潜在语义模型的半监督Web挖掘

随着互联网信息的增长,Web挖掘已经成为数据挖掘研究的热点之一.网页分类是通过学习大量的带有类别标注的训练样本来预测网页的类别,人工标注这些训练样本是相当繁琐的.网页聚类通过一定的相似性度量,将相关网页归并到一类.然而传统的聚类算法对解空间的搜索带有盲目性和缺乏语义特征.提出了两阶段的半监督文本学习策略.第1阶段,利用贝叶斯潜在语义模型来标注含有潜在类别主题词变量的网页的类别;第2阶段,利用简单贝叶斯模型,在第1阶段类别标注的基础上,通过EM(expectation maximization)算法对不含有潜在类别主题词变量的文档作类别标注.实验结果表明,该算法具有很高的精度和召回率.     

Hybrid mining approach in the design of credit scoring models

Unrepresentative data samples are likely to reduce the utility of data classifiers in practical application. This study presents a hybrid mining approach in the design of an effective credit scoring model, based on clustering and neural network techniques. We used clustering techniques to preprocess the input samples with the objective of indicating unrepresentative samples into isolated and inconsistent clusters, and used neural networks to construct the credit scoring model. The clustering stage involved a class-wise classification process. A self-organizing map clustering algorithm was used to automatically determine the number of clusters and the starting points of each cluster. Then, the K-means clustering algorithm was used to generate clusters of samples belonging to new classes and eliminate the unrepresentative samples from each class. In the neural network stage, samples with new class labels were used in the design of the credit scoring model. The proposed method demonstrates by two real world credit data sets that the hybrid mining approach can be used to build effective credit scoring models.     

基于无监督集成聚类的开放关系抽取方法

开放关系抽取(Open Relation Extraction, OpenRE)旨在从开放域语料库中抽取关系事实。大多数OpenRE方法通常局限于无监督方法提取命名实体之间的关系模式,然后将语义等价的模式聚类成一个关系簇,但由于缺少监督信息且聚类精度较低,影响了最终的关系抽取效果。为了进一步提高聚类性能,该文提出一种无监督集成聚类框架(Unsupervised Ensemble Clustering,UEC),它将无监督集成学习与基于信息度量的多步聚类算法相结合自主创建高质量伪标签,并以此作为监督信息改进关系特征的学习,从而引导聚类过程,获得更好的标签质量,最后通过多次迭代聚类发现文本中的关系类型。在FewRel和NYT-FB数据集上的实验结果表明,该文方法优于其他主流的基线OpenRE模型,F₁值分别达到了65.2%和67.1%。