一种面向任务需求的群智感知任务分配模型

针对群智感知平台中的任务分配问题,提出了一种任务需求特征提取算法和用户标签分类方法相结合的T REA U LCM任务分配模型.首先,通过任务需求特征提取算法提取感知任务的类别关键词;然后,通过多线性神经网络和多核学习对数据集进行训练得到分类器,通过分类器对用户的类型标签进行预测;最后,根据任务的类别关键词结合空间位置信息和用户参与度筛选有该任务类别标签且最大化满足任务需求的用户分发任务.仿真结果表明,T REA U LCM任务分配模型在任务匹配度和任务分配效率方面有较好的可行性.     

数据驱动的层次场景序列识别模型研究（英文）

Scene classification is a complicated task, because it includes much content and it is difficult to capture its distribution.A novel hierarchical serial scene classification framework is presented in this paper. At first, we use hierarchical feature to present both the global scene and local patches containing specific objects. Hierarchy is presented by space pyramid match, and our own codebook is built by two different types of words. Secondly, we train the visual words by generative and discriminative methods respectively based on space pyramid match, which could obtain the local patch labels efficiently. Then, we use a neural network to simulate the human decision process, which leads to the final scene category from local labels. Experiments show that the hierarchical serial scene image representation and classification model obtains superior results with respect to accuracy.     

数据驱动的层次场景序列识别模型研究

针对层次场景图像序列,本文提出了一种数据驱动的基于快速序列视觉表述任务（rapid serial visual presentation task,RSVP）的场景识别模型. 首先基于金字塔模型提取三层尺度图像块,然后构建包括全局和局部特征的词汇字典,接着分别利用生成模型和判决模型训练视觉词汇,最后通过神经网络从图像块标记中获得场景类别. 实验表明算法能够获得更为精确的分类结果.     

基于fcmpCNN模型的网络文本情感多分类标注

由于微博等网络文本所含的上下文信息有限,网络文本情感分析更具有挑战性。针对网络文本情感分析,提出了一种基于全卷积—多池化单元的卷积神经网络模型,实现情感多分类标注。无需手动指定多种上下文窗口大小和尽量保留文本的多层次语义,模型通过堆叠多级全卷积—多池化单元,提取出文本特征向量。该文本特征向量包含多个抽象级别、多种上下文窗口大小和不同层次语义的文本特征。模型最后基于此向量计算情感多分类标注。实验表明：模型的网络文本情感多分类标注正确率达到56.3%,与同类模型比较,提高了情感多分类标注的正确率。     

基于端到端的多任务商标分卡模型

目前商标分卡处理方法是先进行文本检测再进行区域分类, 最后对不同的区域进行拆分组合形成商标分卡. 这种分步式的处理耗时长, 并且因为误差的叠加会导致最终结果准确率下降. 针对这一问题, 本文提出了多任务的网络模型TextCls, 通过设计多任务学习模型来提升商标分卡的检测和分类模块的推理速度和精确率. 该模型包含一个特征提取网络, 以及文本检测和区域分类两个任务分支. 其中, 文本检测分支采用分割网络学习像素分类图, 然后使用像素聚合获得文本框, 像素分类图主要是学习文本像素和背景像素的信息; 区域分类分支对区域特征细分为中文、英文和图形, 着重学习不同类型区域的特征. 两个分支通过共享特征提取网络, 像素信息和区域特征相互促进学习, 最终两个任务的精确率得以提升. 为了弥补商标图像的文本检测数据集的缺失以及验证TextCls的有效性, 本文还收集并标注了一个由2000张商标图像构成的文本检测数据集trademark_text (https://github.com/kongbailongtian/trademark_text), 结果表明: 与最佳的文本检测算法相比, 本文的文本检测分支将精确率由94.44%提升至95.16%, 调和平均值F1 score达92.12%; 区域分类分支的F1 score也由97.09%提升至98.18%.     

结合图自编码器与聚类的半监督表示学习方法

节点标签是复杂网络中广泛存在的监督信息,对网络表示学习具有重要作用。基于此,提出了一种结合图自编码器与聚类的半监督表示学习方法（GAECSRL）。首先,以图卷积网络（GCN）和内积函数分别作为编码器和解码器,并构建图自编码器以形成信息传播框架;然后,在编码器生成的低维表示基础上增加k-means聚类模块,从而使图自编码器的训练过程和节点的类别分布划分形成自监督机制;最后,利用节点标签的判别信息对网络低维表示的类别划分进行指导,将网络表示生成、类别划分以及图自编码器的训练构建在一个统一的优化模型中,并获得融合节点标签信息的有效网络表示结果。在仿真实验中,将GAECSRL用于节点分类和链接预测任务。实验结果表明,相比DeepWalk、node2vec、全局结构信息图表示学习（GraRep）、结构化深度网络嵌入（SDNE）和用数据的转导式或归纳式嵌入预测标签和邻居（Planetoid）,在节点分类任务中GAECSRL的Micro?F1指标提高了0.9~24.46个百分点,Macro?F1指标提高了0.76~24.20个百分点;在链接预测任务中,GAECSRL的AUC指标提高了0.33~9.06个百分点,说明GAECSRL获得的网络表示结果能有效提高节点分类和链接预测任务的性能。     

Ranking with uncertain labels and its applications

The techniques for image analysis and classification generally consider the image sample labels fixed and without uncertainties. The rank regression problem studied in this paper is based on the training samples with uncertain labels, which often is the case for the manual estimated image labels. A core ranking model is designed first as the bilinear fusing of multiple candidate kernels. Then, the parameters for feature selection and kernel selection are learned simultaneously by maximum a posteriori for given samples and uncertain labels. The provable convergency Expectation Maximization (EM) method is used for inferring these parameters in an iterative manner. The effectiveness of the proposed algorithm is finally validated by the extensive experiments on age ranking task and human tracking task. The popular FG-NET and the large scale Yamaha aging database are used for the age estimation experiments, and our algorithm outperforms those state-of-the-art algorithms ever reported by other interrelated literatures significantly. The experiment result of human tracking task also validates its advantage over conventional linear regression algorithm. A short version of this paper appeared in ICME07.     

Feature extraction based on ICA for binary classification problems

In manipulating data such as in supervised learning, we often extract new features from the original features for the purpose of reducing the dimensions of feature space and achieving better performance. In this paper, we show how standard algorithms for independent component analysis (ICA) can be appended with binary class labels to produce a number of features that do not carry information about the class labels-these features will be discarded-and a number of features that do. We also provide a local stability analysis of the proposed algorithm. The advantage is that general ICA algorithms become available to a task of feature extraction for classification problems by maximizing the joint mutual information between class labels and new features, although only for two-class problems. Using the new features, we can greatly reduce the dimension of feature space without degrading the performance of classifying systems.     

A novel ant colony optimization based single path hierarchical classification algorithm for predicting gene ontology

There exist numerous state of the art classification algorithms that are designed to handle the data with nominal or binary class labels. Unfortunately, less attention is given to the genre of classification problems where the classes are organized as a structured hierarchy; such as protein function prediction (target area in this work), test scores, gene ontology, web page categorization, text categorization etc. The structured hierarchy is usually represented as a tree or a directed acyclic graph (DAG) where there exist IS-A relationship among the class labels. Class labels at upper level of the hierarchy are more abstract and easy to predict whereas class labels at deeper level are most specific and challenging for correct prediction. It is helpful to consider this class hierarchy for designing a hypothesis that can handle the tradeoff between prediction accuracy and prediction specificity. In this paper, a novel ant colony optimization (ACO) based single path hierarchical classification algorithm is proposed that incorporates the given class hierarchy during its learning phase. The algorithm produces IF–THEN ordered rule list and thus offer comprehensible classification model. Detailed discussion on the architecture and design of the proposed technique is provided which is followed by the empirical evaluation on six ion-channels data sets (related to protein function prediction) and two publicly available data sets. The performance of the algorithm is encouraging as compared to the existing methods based on the statistically significant Student's t-test (keeping in view, prediction accuracy and specificity) and thus confirm the promising ability of the proposed technique for hierarchical classification task.     

A framework for knowledge-based temporal abstraction

A new domain-independent knowledge-based inference structure is presented, specific to the task of abstracting higher-level concepts from time-stamped data. The framework includes a model of time, parameters, events and contexts. A formal specification of a domain's temporal abstraction knowledge supports acquisition, maintenance, reuse and sharing of that knowledge.

The knowledge-based temporal abstraction method decomposes the temporal abstraction task into five subtasks. These subtasks are solved by five domain-independent temporal abstraction mechanisms. The temporal abstraction mechanisms depend on four domain-specific knowledge types: structural, classification (functional), temporal semantic (logical) and temporal dynamic (probabilistic) knowledge. Domain values for all knowledge types are specified when a temporal abstraction system is developed.

The knowledge-based temporal abstraction method has been implemented in the RÉSUMÉ system and has been evaluated in several clinical domains (protocol-based care, monitoring of children's growth and therapy of diabetes) and in an engineering domain (monitoring of traffic control), with encouraging results.     

The speaker identification by using genetic wavelet adaptive network based fuzzy inference system

In this paper, an intelligent speaker identification system is presented for speaker identification by using speech/voice signal. This study includes both combination of the adaptive feature extraction and classification by using optimum wavelet entropy parameter values. These optimum wavelet entropy values are obtained from measured Turkish speech/voice signal waveforms using speech experimental set. It is developed a genetic wavelet adaptive network based on fuzzy inference system (GWANFIS) model in this study. This model consists of three layers which are genetic algorithm, wavelet and adaptive network based on fuzzy inference system (ANFIS). The genetic algorithm layer is used for selecting of the feature extraction method and obtaining the optimum wavelet entropy parameter values. In this study, one of the eight different feature extraction methods is selected by using genetic algorithm. Alternative feature extraction methods are wavelet decomposition, wavelet decomposition – short time Fourier transform, wavelet decomposition – Born–Jordan time–frequency representation, wavelet decomposition – Choi–Williams time–frequency representation, wavelet decomposition – Margenau–Hill time–frequency representation, wavelet decomposition – Wigner–Ville time–frequency representation, wavelet decomposition – Page time–frequency representation, wavelet decomposition – Zhao–Atlas–Marks time–frequency representation. The wavelet layer is used for optimum feature extraction in the time–frequency domain and is composed of wavelet decomposition and wavelet entropies. The ANFIS approach is used for evaluating to fitness function of the genetic algorithm and for classification speakers. It has been evaluated the performance of the developed system by using noisy Turkish speech/voice signals. The test results showed that this system is effective in detecting real speech signals. The correct classification rate is about 91% for speaker classification.     

Modeling in the design of a KBS validation system

We describe and illustrate the modeling issues in the design of a system for validation of knowledge based systems (KBSs). the domain of such a validation system is “KBSs and their validation problems.” the basic idea in our solution is the following. Since different KBSs may use different knowledge representation languages, we first represent the target KBS (i.e., the KBS to be validated) in a general formal model of KBS, and then validate it in this form. the advantage of this strategy is that validation problem solving needs only to refer to the common language of the general formal model. We present a set of possible conceptual abstraction levels in such a model, and argue that each level is associated with a related view on validation problems. Since high level characterizations are difficult to abstract from current knowledge representation languages, we consider the formal aspects of modeling mainly at the “lowest” level, the so-called inference primitive level. We illustrate the approach by formalizing a solution for selected modeling issues at this level.     

Abstraction-guided synthesis of synchronization

We present a novel framework for automatic inference of efficient synchronization in concurrent programs, a task known to be difficult and error-prone when done manually. Our framework is based on abstract interpretation and can infer synchronization for infinite state programs. Given a program, a specification, and an abstraction, we infer synchronization that avoids all (abstract) interleavings that may violate the specification, but permits as many valid interleavings as possible. Combined with abstraction refinement, our framework can be viewed as a new approach for verification where both the program and the abstraction can be modified on-the-fly during the verification process. The ability to modify the program, and not only the abstraction, allows us to remove program interleavings not only when they are known to be invalid, but also when they cannot be verified using the given abstraction. We implemented a prototype of our approach using numerical abstractions and applied it to verify several example programs.     

基于Wasserstein距离分层注意力模型的跨域情感分类

跨领域情感分类任务旨在利用已知情感标签的源域数据对缺乏标记数据的目标域进行情感倾向性分析.文中提出基于Wasserstein距离的分层注意力模型,结合Attention机制,采用分层模型进行特征提取,将Wasserstein距离作为域差异度量方式,通过对抗式训练自动捕获领域共享特征.进一步构造辅助任务捕获与共享特征共现的领域独有特征,结合两种特征表示完成跨域情感分类任务.在亚马逊评论等数据集上的实验表明,文中模型仅利用领域共享特征就达到较高的正确率,在不同的跨领域对之间具有较好的稳定性.     

Stacking model of multi-label classification based on pruning strategies

Exploiting dependencies between the labels is the key of improving the performance of multi-label classification. In this paper, we divide the utilizing methods of label dependence into two groups from the perspective of different ways of problem transformation: label grouping method and feature space extending method. As to the feature space extending method, we find that the common problem is how to measure the dependencies between labels and to select proper labels to add to the original feature space. Therefore, we propose a ReliefF-based pruning model for multi-label classification (ReliefF-based stacking, RFS). RFS measures the dependencies between labels in a feature selection perspective and then selects the more relative labels into the original feature space. Experimental results of 9 multi-label benchmark datasets shows that RFS is more effective compared to other advanced multi-label classification algorithms.

     

An expert system for speaker identification using adaptive wavelet sure entropy

In this study, an expert speaker identification system is presented for speaker identification using Turkish speech signals. Here, a discrete wavelet adaptive network based fuzzy inference system (DWANFIS) model is used for this aim. This model consists of two layers: discrete wavelet and adaptive network based fuzzy inference system. The discrete wavelet layer is used for adaptive feature extraction in the time–frequency domain and is composed of discrete wavelet decomposition and discrete wavelet entropy. The performance of the used system is evaluated by using repeated speech signals. These test results show the effectiveness of the developed intelligent system presented in this paper. The rate of correct classification is about 90.55% for the sample speakers.     

一种基于标签相关性的类属属性多标签分类算法

多标签学习中一个样本可同时属于多个类别标签,每个标签都可能拥有反映该标签特定特点的特征,即类属属性,目前已经出现了基于类属属性的多标签分类算法LIFT。针对LIFT算法中未考虑标签之间相互关系的问题,提出一种基于标签相关性的类属属性多标签分类算法CLLIFT。该算法使用标签距离度量标签之间的相关性,通过在类属属性空间附加相关标签的方式完成标签相关性的引入,以达到提升分类性能的目的。在四个多标签数据集上的实验结果表明,所提算法与LIFT算法相比在多个多标签评价指标上平均提升21.1%。     

一种新的基于统计的自动文本分类方法

自动文本分类就是在给定的分类体系下,让计算机根据文本的内容确定与它相关联的类别。为了提高分类性能,本文提出了中文文本多层次特征提取方法和基于核的距离加权KNN算法。多层次特征提取方法在汉字、常用词表和专业词表三个层次上提取文档的统计特征,能够更好地反映文档的统计分布。基于核的距离加权KNN算法解决了样本的多峰分布、边界重叠问题和分类器的精确分类决策问题。实际应用中,互联网和文本库提供了大量经过粗分类的训练文本,但普遍存在样本质量较差的问题,本文通过样本重要性分析技术解决此问题。实验系统证明了新方法的有效性。     

Objectness Region Enhancement Networks for Scene Parsing

Semantic segmentation has recently witnessed rapid progress, but existing methods only focus on identifying objects or instances. In this work, we aim to address the task of semantic understanding of scenes with deep learning. Different from many existing methods, our method focuses on putting forward some techniques to improve the existing algorithms, rather than to propose a whole new framework. Objectness enhancement is the first effective technique. It exploits the detection module to produce object region proposals with category probability, and these regions are used to weight the parsing feature map directly. “Extra background” category, as a specific category, is often attached to the category space for improving parsing result in semantic and instance segmentation tasks. In scene parsing tasks, extra background category is still beneficial to improve the model in training. However, some pixels may be assigned into this nonexistent category in inference. Black-hole filling technique is proposed to avoid the incorrect classification. For verifying these two techniques, we integrate them into a parsing framework for generating parsing result. We call this unified framework as Objectness Enhancement Network (OENet). Compared with previous work, our proposed OENet system effectively improves the performance over the original model on SceneParse150 scene parsing dataset, reaching 38.4 mIoU (mean intersectionover-union) and 77.9% accuracy in the validation set without assembling multiple models. Its effectiveness is also verified on the Cityscapes dataset.