Document Clustering With Dual Supervision Through Feature Reweighting

Traditional semi‐supervised clustering uses only limited user supervision in the form of instance seeds for clusters and pairwise instance constraints to aid unsupervised clustering. However, user supervision can also be provided in alternative forms for document clustering, such as labeling a feature by indicating whether it discriminates among clusters. This article thus fills this void by enhancing traditional semi‐supervised clustering with feature supervision, which asks the user to label discriminating features during defining (labeling) the instance seeds or pairwise instance constraints. Various types of semi‐supervised clustering algorithms were explored with feature supervision. Our experimental results on several real‐world data sets demonstrate that augmenting the instance‐level supervision with feature‐level supervision can significantly improve document clustering performance.     

ICA mixture model algorithm for unsupervised classification of remote sensing imagery

Conventional remote sensing classification algorithms assume that the data in each class can be modelled using a multivariate Gaussian distribution. As this assumption is often not valid in practice, conventional algorithms do not perform well. In this paper, we present an independent component analysis (ICA)‐based approach for unsupervised classification of multi/hyperspectral imagery. ICA used for a mixture model estimates the data density in each class and models class distributions with non‐Gaussian (sub‐ and super‐Gaussian) probability density functions, resulting in the ICA mixture model (ICAMM) algorithm. Independent components and the mixing matrix for each class are found using an extended information‐maximization algorithm, and the class membership probabilities for each pixel are computed. The pixel is allocated to the class having maximum class membership probability to produce a classification. We apply the ICAMM algorithm for unsupervised classification of images obtained from both multispectral and hyperspectral sensors. Four feature extraction techniques are considered as a preprocessing step to reduce the dimensionality of the hyperspectral data. The results demonstrate that the ICAMM algorithm significantly outperforms the conventional K‐means algorithm for land cover classification produced from both multi‐ and hyperspectral remote sensing images.     

互信息与多条元路径融合的异质网络表示学习方法

异质信息网络能够对真实世界的诸多复杂应用场景进行建模,其表示学习研究也得到了众多学者的广泛关注.现有的异质网络表示学习方法大多基于元路径来捕获网络中的结构和语义信息,已经在后续的网络分析任务中取得很好的效果.然而,此类方法忽略了元路径的内部节点信息和不同元路径实例的重要性;仅能捕捉到节点的局部信息.因此,提出互信息与多条元路径融合的异质网络表示学习方法.首先,利用一种称为关系旋转编码的元路径内部编码方式,基于相邻节点和元路径上下文节点捕获异质信息网络的结构和语义信息,采用注意力机制来建模各元路径实例的重要性;然后,提出一种互信息最大化与多条元路径融合的无监督异质网络表示学习方法,使用互信息捕获全局信息以及全局信息和局部信息之间的联系.最后,在两个真实数据集上进行实验,并与当前主流的算法进行比较分析.结果表明,所提方法在节点分类和聚类任务上性能都有提升,甚至和一些半监督算法相比也表现出强劲性能.     

A thousand words in a scene

This paper presents a novel approach for visual scene modeling and classification, investigating the combined use of text modeling methods and local invariant features. Our work attempts to elucidate (1) whether a textlike bag-of-visterms (BOV) representation (histogram of quantized local visual features) is suitable for scene (rather than object) classification, (2) whether some analogies between discrete scene representations and text documents exist, and 3) whether unsupervised, latent space models can be used both as feature extractors for the classification task and to discover patterns of visual co-occurrence. Using several data sets, we validate our approach, presenting and discussing experiments on each of these issues. We first show, with extensive experiments on binary and multiclass scene classification tasks using a 9,500-image data set, that the BOV representation consistently outperforms classical scene classification approaches. In other data sets, we show that our approach competes with or outperforms other recent more complex methods. We also show that probabilistic latent semantic analysis (PLSA) generates a compact scene representation, is discriminative for accurate classification, and is more robust than the BOV representation when less labeled training data is available. Finally, through aspect-based image ranking experiments, we show the ability of PLSA to automatically extract visually meaningful scene patterns, making such representation useful for browsing image collections.     

A Stochastic Map Building Method for Mobile Robot using 2-D Laser Range Finder

This paper presents a stochastic map building method for mobile robot using a 2-D laser range finder. Unlike other methods that are based on a set of geometric primitives, the presented method builds a map with a set of obstacle regions. In building a map of the environment, the presented algorithm represents the obstacles with a number of stochastic obstacle regions, each of which is characterized by its own stochastic parameters such as mean and covariance. Whereas the geometric primitives based map sometimes does not fit well to sensor data, the presented method reliably represents various types of obstacles including those of irregular walls and sets of tiny objects. Their shapes and features are easily extracted from the stochastic parameters of their obstacle regions, and are used to develop reliable navigation and obstacle avoidance algorithms. The algorithm updates the world map in real time by detecting the changes of each obstacle region. Consequently, it is adequate for modeling the quasi-static environment, which includes occasional changes in positions of the obstacles rather than constant dynamic moves of the obstacles. The presented map building method has successfully been implemented and tested on the ARES-II mobile robot system equipped with a LADAR 2D-laser range finder.     

Missing Value Imputation with Unsupervised Backpropagation

Many data mining and data analysis techniques operate on dense matrices or complete tables of data. Real‐world data sets, however, often contain unknown values. Even many classification algorithms that are designed to operate with missing values still exhibit deteriorated accuracy. One approach to handling missing values is to fill in (impute) the missing values. In this article, we present a technique for unsupervised learning called unsupervised backpropagation (UBP), which trains a multilayer perceptron to fit to the manifold sampled by a set of observed point vectors. We evaluate UBP with the task of imputing missing values in data sets and show that UBP is able to predict missing values with significantly lower sum of squared error than other collaborative filtering and imputation techniques. We also demonstrate with 24 data sets and nine supervised learning algorithms that classification accuracy is usually higher when randomly withheld values are imputed using UBP, rather than with other methods.     

基于激光测距仪的障碍物检测的仿真研究

为了进行准确和有效的导航,提出了利用机器人的车载激光测距仪来检测环境中的障碍物.利用激光测距仪对机器人的局部环境建立栅格地图,在该模型下对障碍物进行检测.针对连续变化时刻(t-1)和(t)下的两帧“图像”,在极坐标系中进行障碍物分割,根据分割结果建立障碍物链Object_ List (t-1)和Object_ List (t),计算出各特征参数,并作进一步的匹配和分类分析.如果存在动态障碍物,估计出相应的运动参数,更新得到t时刻的动态障碍物集Dob (N).为了验证算法的有效性,利用visual C++进行仿真,仿真实验结果表明,该方法进行环境中的动静态障碍物的检测是可行的和有效的.     

地图构建中动态障碍干扰滤除的一种方法

在移动机器人环境建图中,动态障碍物的存在直接影响传感器的读数,导致产生不一致的环境地图,因此,移动机器人构建地图必须滤除动态障碍物干扰。采用直线插补的方法在先前的局部图上搜寻机器人与目标点之间是否存在障碍物,若存在障碍,则可判定该障碍物已移走(即为动态障碍),应该予以滤除。实验结果证明,该方法能在建图过程中有效地滤除动态障碍,并能有效减少静态障碍物探测的误差累积,算法复杂度小。     

Multimodal obstacle detection in unstructured environments with conditional random fields

Reliable obstacle detection and classification in rough and unstructured terrain such as agricultural fields or orchards remains a challenging problem. These environments involve large variations in both geometry and appearance, challenging perception systems that rely on only a single sensor modality. Geometrically, tall grass, fallen leaves, or terrain roughness can mistakenly be perceived as nontraversable or might even obscure actual obstacles. Likewise, traversable grass or dirt roads and obstacles such as trees and bushes might be visually ambiguous. In this paper, we combine appearance‐ and geometry‐based detection methods by probabilistically fusing lidar and camera sensing with semantic segmentation using a conditional random field. We apply a state‐of‐the‐art multimodal fusion algorithm from the scene analysis domain and adjust it for obstacle detection in agriculture with moving ground vehicles. This involves explicitly handling sparse point cloud data and exploiting both spatial, temporal, and multimodal links between corresponding 2D and 3D regions. The proposed method was evaluated on a diverse data set, comprising a dairy paddock and different orchards gathered with a perception research robot in Australia. Results showed that for a two‐class classification problem (ground and nonground), only the camera leveraged from information provided by the other modality with an increase in the mean classification score of 0.5%. However, as more classes were introduced (ground, sky, vegetation, and object), both modalities complemented each other with improvements of 1.4% in 2D and 7.9% in 3D. Finally, introducing temporal links between successive frames resulted in improvements of 0.2% in 2D and 1.5% in 3D.     

Incremental nonlinear dimensionality reduction by manifold learning

Understanding the structure of multidimensional patterns, especially in unsupervised cases, is of fundamental importance in data mining, pattern recognition, and machine learning. Several algorithms have been proposed to analyze the structure of high-dimensional data based on the notion of manifold learning. These algorithms have been used to extract the intrinsic characteristics of different types of high-dimensional data by performing nonlinear dimensionality reduction. Most of these algorithms operate in a "batch" mode and cannot be efficiently applied when data are collected sequentially. In this paper, we describe an incremental version of ISOMAP, one of the key manifold learning algorithms. Our experiments on synthetic data as well as real world images demonstrate that our modified algorithm can maintain an accurate low-dimensional representation of the data in an efficient manner.     

A novel unsupervised classification approach for network anomaly detection by k-Means clustering and ID3 decision tree learning methods

This paper presents a novel host-based combinatorial method based on k-Means clustering and ID3 decision tree learning algorithms for unsupervised classification of anomalous and normal activities in computer network ARP traffic. The k-Means clustering method is first applied to the normal training instances to partition it into k clusters using Euclidean distance similarity. An ID3 decision tree is constructed on each cluster. Anomaly scores from the k-Means clustering algorithm and decisions of the ID3 decision trees are extracted. A special algorithm is used to combine results of the two algorithms and obtain final anomaly score values. The threshold rule is applied for making the decision on the test instance normality. Experiments are performed on captured network ARP traffic. Some anomaly criteria has been defined and applied to the captured ARP traffic to generate normal training instances. Performance of the proposed approach is evaluated using five defined measures and empirically compared with the performance of individual k-Means clustering and ID3 decision tree classification algorithms and the other proposed approaches based on Markovian chains and stochastic learning automata. Experimental results show that the proposed approach has specificity and positive predictive value of as high as 96 and 98%, respectively.     

Spatio-Temporal Tube data representation and Kernel design for SVM-based video object retrieval system

In this article, we propose a new video object retrieval system. Our approach is based on a Spatio-Temporal data representation, a dedicated kernel design and a statistical learning toolbox for video object recognition and retrieval. Using state-of-the-art video object detection algorithms (for faces or cars, for example) we segment video object tracks from real movies video shots. We then extract, from these tracks, sets of spatio-temporally coherent features that we call Spatio-Temporal Tubes. To compare these complex tube objects, we design a Spatio-Temporal Tube Kernel (STTK) function. Based on this kernel similarity we present both supervised and active learning strategies embedded in Support Vector Machine framework. Additionally, we propose a multi-class classification framework dealing with unbalanced data. Our approach is successfully evaluated on two real movies databases, the french movie “L’esquive” and episodes from “Buffy, the Vampire Slayer” TV series. Our method is also tested on a car database (from real movies) and shows promising results for car identification task.     

Land use classification of polarimetric SAR data by visual interpretation and comparison with an automatic procedure

A study is presented in which several representations of polarimetric SAR data were evaluated for the purpose of obtaining land use classification. Two methods comprising visual interpretation and an automatic procedure were used. For the study, fully polarimetric SAR data with a resolution of 3?m were obtained with the Dutch PHARUS sensor from a test area in the Netherlands showing various classes of land use. The land use classes consisted of bare soil, water, grass, urban areas, and forest. The visual inspection was performed by different groups of non-expert interpreters for each representation. It was found that people are quite successful by visual interpretation in performing land use classification using SAR data. Multi-polarized data are required for this purpose, although these data need not be fully polarimetric, since the best results were obtained with the hh- and hv-polarization combinations displayed in the red and green colour channels. The results show that land use classification by visual inspection is more effective than the automatic classification procedure.     

Bird's‐Eye ‐ Large‐Scale Visual Analytics of City Dynamics using Social Location Data

The analysis of behavioral city dynamics, such as temporal patterns of visited places and citizens' mobility routines, is an essential task for urban and transportation planning. Social media applications such as Foursquare and Twitter provide access to large‐scale and up‐to‐date dynamic movement data that not only help to understand the social life and pulse of a city but also to maintain and improve urban infrastructure. However, the fast growth rate of this data poses challenges for conventional methods to provide up‐to‐date, flexible analysis. Therefore, planning authorities barely consider it. We present a system and design study to leverage social media data that assist urban and transportation planners to achieve better monitoring and analysis of city dynamics such as visited places and mobility patterns in large metropolitan areas. We conducted a goal‐and‐task analysis with urban planning experts. To address these goals, we designed a system with a scalable data monitoring back‐end and an interactive visual analytics interface. The monitoring component uses intelligent pre‐aggregation to allow dynamic queries in near real‐time. The visual analytics interface leverages unsupervised learning to reveal clusters, routines, and unusual behavior in massive data, allowing to understand patterns in time and space. We evaluated our approach based on a qualitative user study with urban planning experts which demonstrates that intuitive integration of advanced analytical tools with visual interfaces is pivotal in making behavioral city dynamics accessible to practitioners. Our interviews also revealed areas for future research.     

Vision‐based navigation through urban canyons

We address the problem of navigating unmanned vehicles safely through urban canyons in two dimensions using only vision‐based techniques. Two commonly used vision‐based obstacle avoidance techniques (namely stereo vision and optic flow) are implemented on an aerial and a ground‐based robotic platform and evaluated for urban canyon navigation. Optic flow is evaluated for its ability to produce a centering response between obstacles, and stereo vision is evaluated for detecting obstacles to the front. We also evaluate a combination of these two techniques, which allows a vehicle to detect obstacles to the front while remaining centered between obstacles to the side. Through experiments on an unmanned ground vehicle and in simulation, this combination is shown to be beneficial for navigating urban canyons, including T‐junctions and 90‐deg bends. Experiments on a rotorcraft unmanned aerial vehicle, which was constrained to two‐dimensional flight, demonstrate that stereo vision allowed it to detect an obstacle to the front, and optic flow allowed it to turn away from obstacles to the side. We discuss the theory behind these techniques, our experience in implementing them on the robotic platforms, and their suitability to the urban canyon navigation problem. © 2009 Wiley Periodicals, Inc.     

An evaluation of dimension reduction techniques for one-class classification

Dimension reduction (DR) is important in the processing of data in domains such as multimedia or bioinformatics because such data can be of very high dimension. Dimension reduction in a supervised learning context is a well posed problem in that there is a clear objective of discovering a reduced representation of the data where the classes are well separated. By contrast DR in an unsupervised context is ill posed in that the overall objective is less clear. Nevertheless successful unsupervised DR techniques such as principal component analysis (PCA) exist—PCA has the pragmatic objective of transforming the data into a reduced number of dimensions that still captures most of the variation in the data. While one-class classification falls somewhere between the supervised and unsupervised learning categories, supervised DR techniques appear not to be applicable at all for one-class classification because of the absence of a second class label in the training data. In this paper we evaluate the use of a number of up-to-date unsupervised DR techniques for one-class classification and we show that techniques based on cluster coherence and locality preservation are effective.     

Geospatial variability in the egg raft distribution and abundance of Culex pipiens and Culex restuans in Urbana‐Champaign,Illinois

The aim of this study was to determine whether areas of high Culex pipiens and Culex restuans abundance in an urban environment, based on fixed oviposition surveillance sites, corresponded to remotely sensed data. A land use land cover (LULC) classification, based on Landsat‐7 Enhanced Thematic Mapper Plus (ETM+) data acquired in July 2003 and Landsat‐5 TM data from July 1991, was compared to the abundance of Culex egg rafts in Urbana‐Champaign, East‐Central Illinois. We performed a maximum likelihood, unsupervised classification and generated three land cover classifications: urban, non‐urban and water. Ground coordinates of the Culex surveillance sites were overlaid onto LULC maps with 10 m² grid cells. The grid was stratified based on levels of drainage: 0 = poorly drained and 1 = well‐drained. Total LULC change from 1991 to 2003 in the Urbana‐Champaign study site was relatively low, at 12.1%. The most frequent LULC category was maintained urban. The egg raft rate was significantly higher in urban LULC habitats. Remote stratification of the urban land cover using QuickBird visible and near‐infrared (NIR) data revealed that high‐density canopy coverage was most frequently associated with high Culex abundance in oviposition traps. We demonstrate that optical remote sensing can identify land use parameters associated with high Culex oviposition.     

Anticipated velocity based guidance strategy for wheeled mobile evader amidst stationary and moving obstacles in bounded environment

This paper is concerned with a class of pursuit‐evasion game problems amidst stationary and moving obstacles in a bounded environment. We concentrate on evader's strategy taking into account the following challenges: (i) pursuer and evader are nonholonomic wheeled mobile robots and the evader is slower than the pursuer; (ii) pursuer follows a proportional navigation law; and (iii) geometry of the environment is not known to the players, a priori. We propose an efficient evader‐centric anticipated velocity based guidance strategy. Pursuer's trajectory is anticipated at each step by the evader using quadratic polynomial interpolation. The aim of the evader is to escape interception with the pursuer for maximum possible time. To deal with static obstacles, a technique based on a well‐known tangent bug algorithm is presented. While dealing with dynamic obstacles, a recently introduced reciprocal orientation method is employed to avoid collision in situations when the dynamic obstacle also cooperates in the process. In case dynamic obstacles do not participate in the process of collision avoidance, a well‐known velocity obstacle method is employed for planning safe collision‐free paths. Efficiency of the proposed algorithms is analyzed with respect to the interception time and the distance traveled by the players. Copyright © 2014 John Wiley & Sons, Ltd.