Uncertain data streams have been widely generated in many Web applications. The uncertainty makes anomaly detection from sensor far more challenging. In this article, we present a novel framework that supports uncertain streams. proposed adopts the wavelet soft-thresholding method to remove noises or errors Based on refined streams, develop effective period pattern recognition and feature extraction techniques improve computational efficiency. We use classification methods for corrected...

Diabetic eye disease (DED) is a cluster of problem that affects diabetic patients. Identifying DED crucial activity in retinal fundus images because early diagnosis and treatment can eventually minimize the risk visual impairment. The image plays significant role classification identification. An accurate diagnostic model's development using depends highly on quality quantity. This paper presents methodical study significance processing for classification. proposed automated framework was...

The growth of databases in the healthcare domain opens multiple doors for machine learning and artificial intelligence technology. Many medical devices are available field; however, errors remain a severe challenge. Different algorithms developed to identify solve errors, such as detecting anomalous readings, health conditions patient, etc. However, they fail answer why those entries considered an anomaly. This research gap leads outlying aspect mining problem. problem aims discover set...

Recent advancements in audio deepfake detection have leveraged graph neural networks (GNNs) to model frequency and temporal interdependencies data, effectively identifying artifacts. However, the reliance of GNN-based methods on substantial labeled data for construction robust performance limits their applicability scenarios with limited data. Although vast amounts exist, process labeling samples as genuine or fake remains labor-intensive costly. To address this challenge, we propose SIGNL...

Abstract The main aim of the outlying aspect mining algorithm is to automatically detect subspace(s) (a.k.a. aspect(s)), where a given data point dramatically different than rest in each those (aspect(s)). To rank subspaces for point, scoring measure required compute degree subspace. In this paper, we introduce new degree, called Simple Isolation score using Nearest Neighbor Ensemble (SiNNE), which not only detects outliers but also provides an explanation on why selected outlier. SiNNE...

With an ever-increasing number of Web services being available, finding desired service is crucial for users. Current keyword search and most existing approaches are inefficient in two main aspects: poor scalability lack semantics. Firstly, users overwhelmed by the huge irrelevant returned. Secondly, intentions semantics ignored. Inspired success divide conquer approach used to handle complex information decomposition, we use a novel partition large set results into smaller groups employing...

QoS-based service rating has made positive contributions to the area of selection. Especially for Cloud users, right decision when choosing suitable services can help them improve user satisfaction and trading revenues. This work aims address issue uncertainty in requests, descriptions, expert preferences, as well evaluation criteria a MCDM-based selection procedure. A hybrid fuzzy framework is proposed, addressing challenge using three approaches: fuzzy-ontology-based approach function...

Regularities analysis for prescriptions is a significant task traditional Chinese medicine (TCM), both in inheritance of clinical experience and improvement quality. Recently, many methods have been proposed regularities discovery, but this challenging due to the quantity, sparsity free-style prescriptions. In paper, we address specific problem discovery propose graph embedding based framework massive We model as relation prediction which correlation two herbs or herb symptom are...

Efficient document classification techniques are crucial to current legal applications, such as case-based reasoning, citations, and so on. However, Chinese judgment documents large highly complex, the traditional machine leaning-based models often inefficient due fact that they fail incorporate overall structure extra domain specific knowledge. In this paper, we propose an ontology-driven knowledge block summarization approach computing similarity for classification. First, semantic is...

Recent technological advances have accelerated the design and deployment of kinds secure applications on smartphones. Although users can access handle their data flexibly stably with mobile devices, not only computing it poses security challenges a new dimension that disclose lots sensitive privacy information over open devices networks as well. Thus, more malwares are emerging to compromise OS steal from these applications. In this paper, we propose payment framework TrustPAY TrustZone...

There has been a continuing demand for traditional and complementary medicine worldwide. A fundamental important topic in Traditional Chinese Medicine (TCM) is to optimize the prescription detect herb regularities from TCM data. In this paper, we propose novel clustering model solve general problem of categorization, pivotal task optimization regularities. The utilizes Random Walks method, Bayesian rules Expectation Maximization(EM) models complete analysis effectively on heterogeneous...

There has been a continuing demand for traditional and complementary medicine worldwide. A fundamental important topic in Traditional Chinese Medicine (TCM) is to optimize the prescription detect herb regularities from TCM data. In this paper, we propose novel clustering model solve general problem of categorization, pivotal task optimization regularities. The utilizes Random Walks method, Bayesian rules Expectation Maximization(EM) models complete analysis effectively on heterogeneous...

The insider threat continues to be a paramount cyber security challenge that threatens individuals, financial enterprises and governmental organizations. To deter threats, traditional detection, which mainly focuses on policy checks anomaly detection for users' computers network activities, has been studied widely. However, because insiders have intrinsic authorized access at attack under normal behavior profiles, it is necessary integrate the attackers' psychological characteristics. This...

Positive unlabeled time series classification (PUTSC) refers to classifying with a set PL of positive labeled examples and U ones. Model selection for PUTSC is largely untouched topic. In this paper, we look into model selection, which as far know the first systematic study in Focusing on widely adopted self-training one-nearest-neighbor (ST-1NN) paradigm, propose framework based active learning (AL). We present novel concepts label propagation, pseudo calibration principles ultimately...

Side effects of prescribed medications are a common occurrence. Electronic healthcare databases present the opportunity to identify new side efficiently but currently methods limited due confounding (i.e. when an association between two variables is identified them both being associated third variable). In this paper we propose proof concept method that learns associations and uses knowledge automatically refine effect signals exposure-outcome associations) by removing instances caused...

How to guide the visually impaired walk safely and independently in strange environment is one of urgent tasks. For such situation, we proposed a system which high precision navigation based on haptic blind spatial cognition. The combined with cognition features ultra-wideband (UWB) wireless positioning technology realize precise positioning, using different vibration prompt mode voice pedestrians. accuracy UWB achieved decimeter level used for completing route guidance. Different would tip...

Absence epilepsy is one of the most common types epilepsy. The diagnosis absence among greatest challenges faced by clinical neurologists due to a lack easily observable symptoms that are present in conventional (e.g. spasm and convulsion), highly relies on detection Spike Slow Waves (SSWs) Electroencephalogram (EEG) signals. Recently, graph representations called complex networks have been increasingly applied characterizing 1D EEG However, existing methods often fail effectively represent...