Detecting human-object interactions is essential for comprehensive understanding of visual scenes. In particular, spatial connections between humans and objects are important cues for reasoning interactions. To this end, we propose a skeleton-aware graph convolutional network for human-object interaction detection, named SGCN4HOI. Our network exploits the spatial connections between human keypoints and object keypoints to capture their fine-grained structural interactions via graph convolutions. It fuses such geometric features with visual features and spatial configuration features obtained from human-object pairs. Furthermore, to better preserve the object structural information and facilitate human-object interaction detection, we propose a novel skeleton-based object keypoints representation. The performance of SGCN4HOI is evaluated in the public benchmark V-COCO dataset. Experimental results show that the proposed approach outperforms the state-of-the-art pose-based models and achieves competitive performance against other models.

We present a USB-based, highly directional, and portable microphone array device that delivers a crisp, clear and noise-reduced speech signal. This device consists of four linearly distributed microphone sensors and a filter-and-sum beamformer designed using broadband beam-forming algorithm. The device has a narrow acoustic beam pattern and identical frequency responses for almost all speech bands. In addition to beamforming, an adaptive noise reduction algorithm is used to further reduce the background noise. By utilizing both the spatial and temporal information, the SNR of speech signals is improved and speech recognition performance in noisy environments is significantly improved as reported in our experiments.

BACKGROUND: Microarray-based tumor classification is characterized by a very large number of features (genes) and small number of samples. In such cases, statistical techniques cannot determine which genes are correlated to each tumor type. A popular solution is the use of a subset of pre-specified genes. However, molecular variations are generally correlated to a large number of genes. A gene that is not correlated to some disease may, by combination with other genes, express itself. RESULTS: In this paper, we propose a new classiification strategy that can reduce the effect of over-fitting without the need to pre-select a small subset of genes. Our solution works by taking advantage of the information embedded in the testing samples. We note that a well-defined classification algorithm works best when the data is properly labeled. Hence, our classification algorithm will discriminate all samples best when the testing sample is assumed to belong to the correct class. We compare our solution with several well-known alternatives for tumor classification on a variety of publicly available data-sets. Our approach consistently leads to better classification results. CONCLUSION: Studies indicate that thousands of samples may be required to extract useful statistical information from microarray data. Herein, it is shown that this problem can be circumvented by using the information embedded in the testing samples.

The success of linear discriminant analysis (LDA) is due in part to the simplicity of its formulation, which reduces to a simultaneous diagonalization of two symmetric matrices A and B;. However, a fundamental drawback of this approach is that it cannot be efficiently applied wherever the matrix A is singular or when some of the smallest variances in A are due to noise. In this paper, we present a factorization of A(-1) and a correlation-based criterion that can be readily employed to solve these problems. We provide detailed derivations for the linear and nonlinear classification problems. The usefulness of the proposed approach is demonstrated thoroughly using a large variety of databases.

With mobile devices and wireless hotspots becoming more prevalent, customers can desire greater access to media and services that can be achieved from the marrying of these two technologies. However, often existing devices need to be augmented with hardware and software to achieve this connectivity, and the existing services are heterogeneous and incompatible with one another. By leveraging the captive portal mechanism and proposed context-aware technologies, we propose a lightweight framework that can aggregate these services in smart spaces, and allow mobile users to spontaneously discover them without any specialized installation. In this paper, we present the system requirements for spontaneous interaction, the service framework design and implementation details.

People have the desire to be always informed by the information to their importance everywhere and anytime. However, for some casual and relaxing occasions such as in home environment, information access and exhibition should be offered in a non-distracting and non-disruptive manner. To facilitate this goal and make use of embedded displays in smart home environments, this paper proposes a framework, i.e., context-aware informative display which offers a novel design for information representation catering for home and shared space users. The system architecture and techniques used are illustrated in details. Lastly, a prototype with several user cases is described to prove the workability of the framework.

Linear Discriminant Analysis (LDA) is a well-known and important tool in pattern recognition with potential applications in many areas of research. The most famous and used formulation of LDA is that given by the Fisher-Rao criterion, where the problem reduces to a simple simultaneous diagonalization of two symmetric, positive-definite matrices, A and B; i.e. B^-1 AV = VA. Here, A defines the metric to be maximized, while B defines the metric to be minimized. However, when B has near-zero eigenvalues, the Fisher-Rao criterion gets dominated by these. While this works well when such small variances describe vectors where most of the discriminant information is, the results will be incorrect when these small variances are caused by noise. Knowing which of these near-zero values are to be used and which need to be eliminated is a challenging yet fundamental task in LDA. This paper presents a criterion for the selection of those vectors of B that are best for classification. The proposed solution is based on a simple factorization of B^-1 A that permits the re-ordering of the eigenvectors of B without the need to effect the end result. This allows us to readily eliminate the noisy vectors while keeping the most discriminant ones. A theoretical basis for these results is presented along with extensive experimental results to validate the claims.

k-closest pair query is a useful type of query in many practical applications involving spatial data for decision making. The traditional techniques to handle k-closest pair queries generally assume that the objects are static. In this paper, we study the problem of k-closest pair monitoring (kCPM) over moving objects. Aiming at reducing communication cost for location updates between the clients and the server, our proposed kCPM approach achieves high monitoring accuracy with less CPU utilization compared to existing periodical location update schemes.

Over the years, many Discriminant Analysis (DA) algorithms have been proposed for the study of high-dimensional data in a large variety of problems. Each of these algorithms is tuned to a specific type of data distribution (that which best models the problem at hand). Unfortunately, in most problems the form of each class pdf is a priori unknown, and the selection of the DA algorithm that best fits our data is done over trial-and-error. Ideally, one would like to have a single formulation which can be used for most distribution types. This can be achieved by approximating the underlying distribution of each class with a mixture of Gaussians. In this approach, the major problem to be addressed is that of determining the optimal number of Gaussians per class, i.e., the number of subclasses. In this paper, two criteria able to find the most convenient division of each class into a set of subclasses are derived. Extensive experimental results are shown using five databases. Comparisons are given against Linear Discriminant Analysis (LDA), Direct LDA (DLDA), Heteroscedastic LDA (HLDA), Nonparametric DA (NDA), and Kernel-Based LDA (K-LDA). We show that our method is always the best or comparable to the best.

Despite the existence of obstacles in many database applications, traditional spatial query processing assumes that points in space are directly reachable and utilizes the Euclidean distance metric. In this paper, we study spatial queries in the presence of obstacles, where the obstructed distance between two points is defined as the length of the shortest path that connects them without crossing any obstacles. We propose efficient algorithms for the most important query types, namely, range search, nearest neighbours, e‐distance joins, closest pairs and distance semi‐joins, assuming that both data objects and obstacles are indexed by R‐trees. The effectiveness of the proposed solutions is verified through extensive experiments.

A fundamental problem in computer vision and pattern recognition is to determine where and, most importantly, why a given technique is applicable. This is not only necessary because it helps us decide which techniques to apply at each given time. Knowing why current algorithms cannot be applied facilitates the design of new algorithms robust to such problems. In this paper, we report on a theoretical study that demonstrates where and why generalized eigen-based linear equations do not work. In particular, we show that when the smallest angle between the ith eigenvector given by the metric to be maximized and the first i eigenvectors given by the metric to be minimized is close to zero, our results are not guaranteed to be correct. Several properties of such models are also presented. For illustration, we concentrate on the classical applications of classification and feature extraction. We also show how we can use our findings to design more robust algorithms. We conclude with a discussion on the broader impacts of our results.

Given two spatial data sets A and B, a top-k spatial join retrieves the k objects from A or B that intersect the largest number of objects from the other data set. Depending on the application requirements, there exist several variations of the problem. For instance, B may be a point data set, and the goal may be to retrieve the regions of A that contain the maximum number of points. The processing of such queries with conventional spatial join algorithms is expensive. However, several improvements are possible based on the fact that we only require a small subset of the result (instead of all intersection/containments pairs). In this paper, we propose output-sensitive algorithms for top-k spatial joins that utilize a variety of optimizations for reducing the overhead.

Spatial databases are entering an era of mass deployment in various real-life applications, especially mobile and location-based services. The real-time processing of spatial queries to meet different performance goals poses new problems to the real-time and parallel processing communities. We investigate how multiple window queries can be parallelized, decomposed, scheduled and processed in real time workloads to optimize system performance, such as I/O cost, response time and miss rate. We devise in-memory R-trees to decompose queries into independent jobs. Jobs from different queries can be combined according to their spatial locality to eliminate redundant I/Os. Runtime job schedulers are elaborately devised to optimize response time or miss rate for various systems. Empirical results show a significant performance improvement over the sequential, unparalleled approach

In this paper we propose an approach that enables mobile clients to determine the validity of previous queries based on their current locations. In order to make this possible, the server returns in addition to the query result, a validity region around the client's location within which the result remains the same. We focus on two of the most common spatial query types, namely nearest neighbor and window queries, define the validity region in each case and propose the corresponding query processing algorithms. In addition, we provide analytical models for estimating the expected size of the validity region. Our techniques can significantly reduce the number of queries issued to the server, while introducing minimal computational and network overhead compared to traditional spatial queries.