Accurate classification of voltage sag is crucial for sag responsibility allocation and optimization of management decisions. Existing voltage sag classification methods rely heavily on massive labeled data during the supervised training process. However, it is difficult to obtain sufficient labeled information from field measurements. To address this issue, this article proposes a novel voltage sag classification method based on semisupervised learning, which is suitable for the situations of limited labeled data. First, voltage sag data is transformed into a two-dimensional graphic form using the space phasor model, enabling automatic feature extraction through convolutional neural networks. Next, a new semisupervised learning algorithm named virtual adversarial mean teacher model (VAMT) is proposed to construct the classifier. The proposed method not only enables the efficient learning of unlabeled data, but also effectively improve the model's robustness against adversarial samples present in both voltage sag dataset and data augmentation process. Finally, based on field-measured dataset, the proposed method exhibits excellent classification performance in comparison with existing methods in complex environments, and is suitable for practical engineering applications.

Second order general integrator (SOGI)-based power calculation method is generally used to calculate single-phase average power owing to its ability to eliminate double frequency component with limited delay. It is important to identify the transfer function of the power calculation to facilitate the closed-loop design of a single-phase grid-forming power converter. In this paper, we investigate the nonlinearity and periodic time variability introduced by the SOGI-based power calculation method. To address this issue, we propose a method to identify the optimally linearized transfer function of the power calculation part through the system identification method based on numerical subspace state space system identification algorithm and particle swarm optimization algorithm. The optimally linearized transfer function is used for closed-loop pole placement design of a single-phase virtual synchronous generator. Simulation results verify the effectiveness of the proposed linearization method in accurately designing the closed loop of a single-phase grid-forming power converter.

The Computer Journal, Volume 62, Issue 12, December 2019, Pages 1734-1747, doi: 10.1093/comjnl/bxz033. In the originally published version of this manuscript, requested author amendments were inadvertently omitted prior to publishing and have since been corrected. The Publisher would like to apologise for any inconvenience caused to the reader.

Feature selection has been playing an important role in analyzing the high-dimension and low-sample-size gene expression profiles towards high classification performance of diseases and deep understanding of the underlying biological mechanisms. Besides classification performance, the stability of selected features is another non-ignorable factor in evaluating a feature selector, since stable feature selection results enhance the confidence of selected features for true biomarker discovery and further biological validation. In this study, we propose a novel feature selection method under the ensemble learning framework. Specifically, we take Fast Correlation Based Filter as the base feature selector to analyze subsamples of microarray data. We then present several aggregation methods to combine multiple feature subsets. Finally, two stability measures are used to quantify the robustness of feature selectors to data variations. Our comparative empirical study on publicly available datasets demonstrates the superiority of the proposed methods over its competitors in obtaining high stability scores and classification accuracy.

Advances in wireless sensor networks and increasing Internet-of-Things devices give great opportunities for smart homecare of the elderly. Smart homecare has been a promising issue and received much attention recently. Irregularity detection is one of the most important issues in smart homecare for assessing the health condition of the elderly. However, most of the researches focused on the explicit irregularity detections which are usually based on the drastic changes of sensor data, such as falling. Existing mechanisms for detecting implicit irregularity rely on the subjective assessment of behaviors' importance by the elder and simply outputs the binary detection results. This article proposes a feature-based implicit irregularity detection mechanism (FIID), which extracts the regularity features using unsupervised learning and outputs the probability of implicit irregularity. The proposed FIID identifies the regular behaviors which satisfy the time-regular and happen-frequently properties as the regularity features of daily behaviors. These features then construct a multidimensional feature space to calculate the implicit irregularity probability of the daily health condition. Performance results show that the proposed FIID outperforms the existing implicit irregularity mechanism in terms of precision, recall as well as F-measure.

Abstract Malware brings a big security threat on the Internet today. With the great increasing malware attacks. Behavior-based detection approaches are one of the major method to detect zero-day malware. Such approaches often use API calls to represent the behavior of malware. Unfortunately, behavior-based approaches suffer from behavior obfuscation attacks. In this paper, we propose a novel malware detection approach that is both effective and efficient. First, we abstract the API call to object operation. And then we generate the object operation dependency graph based on these object operations. Our approach use dependency graph to represent the behavior of malware. The evaluation results show that our approach can provide a complete resistance to all types of behavior obfuscation attacks, and outperforms existing behavior-based approaches in terms of better effectiveness and efficiency.

In indoor environment, passive human detection and localisation are important enabling technologies for elder healthcare, emergence rescue and target tracking applications. Recently, the fine‐grained channel state information (CSI) of Wi‐Fi was adopted for indoor localisation due to the low‐cost Wi‐Fi network interface card and available firmware modifications for CSI extraction. However, due to multipath fading and spatial‐temporal dynamics of wireless channel, stable CSI extraction is a challenging task to achieve reliable CSI fingerprint matching. In this study, the sensitivity of CSI is first analysed and stable CSI fingerprints can be obtained by reducing the variance from interference and white noise. The stable CSI fingerprints are then classified by quadratic discriminant analysis to achieve location matching. Extensive experiments have been conducted to justify the system performance. The results reveal that the proposed indoor passive localisation system outperforms passive CSI‐MIMO system in terms of performance.

Radio frequency based device-free passive localization has been actively studied. An important issue with this technique is that a large amount of sensor nodes are usually deployed in localization environments, causing high deployment cost, high computation complexity and other drawbacks. To reduce the number of nodes while maintaining high localization accuracy, this paper proposes a fingerprint-based node deployment mechanism, which saves hardware cost, speeds up node deployment process as well as maintains a similar accuracy. The experiments are conducted in two environments with considerably different sizes of 150 m2and 25 m2. Maintaining the accuracies above 95%, the proposed mechanism can efficiently reduce the total number of nodes in large room by 54.55% and in small room by 68.75%. Especially, the localization response time sharply reduces by 93.08% in the large room and 79.87% in the small room.

With the wide use of video-editing software, more and more similar videos are uploaded to the video-sharing platforms in cloud storage. The large number of near-duplicate videos would not only make the users not satisfactory, but also consume more resources with service providers. Since the near-duplicate videos have redundancy at the content level, the traditional de-duplication technology is not suitable to address the redundancies. In order to decrease the redundancy, we proposed a novel high performance provenance-based video-sharing system, called Provis. Provis records the provenance of the videos and compresses the near-duplicate videos via storing their provenance which replaces storing their video files to improve the space efficiency. When users attempt to upload the edited videos, Provis can rebuild the videos in the cloud via collecting and uploading their provenance to reduce the network overhead of uploading videos. We implemented Provis and compared its performance with other existing video compression techniques. Our evaluation shows that Provis achieves significant space savings.

Interesting target detection algorithm in complex natural backgrounds images is studied in this paper. Firstly, logGabor filter bank is analyzed, which is consistent with human visual system characteristics. Several kinds of local features from the filter bank can form the integrated feature. Integrated feature congruency (IFC) model is established. And upon compensating noise for IFC, an improved integrated feature congruency (IIFC) model is obtained, in which, target detecting is translated to find the interest points that are significant across scales and orientations. This model is applied to complex natural backgrounds images for target detection. Experimental results show that this method can detect interesting targets effectively from complex natural backgrounds scenes.