Network traffic classification has become an important part of network management, which is beneficial for achieving intelligent operation and maintenance, enhancing the quality service (QoS), security. Given rapid development various applications protocols, more encrypted emerged in networks. Traditional methods exhibited unsatisfied performance since no longer plain text. In this work, we modeled time-series by recurrent neural (RNN). Moreover, attention mechanism was introduced assisting...

In recent years, there has been increasing research on computer-aided diagnosis (CAD) using deep learning and image processing techniques. Still, most studies have focused the benign-malignant classification of nodules. this study, we propose an integrated architecture for grading thyroid nodules based Chinese Thyroid Imaging Reporting Data System (C-TIRADS). The method combines traditional handcrafted features with in extraction process. preprocessing stage, a pseudo-artifact removal...

In task-oriented dialogue systems, Dialogue State Tracking (DST) aims to extract users' intentions from the history. Currently, most existing approaches suffer error propagation and are unable dynamically select relevant information when utilizing previous states. Moreover, relations between updates of different slots provide vital clues for DST. However, rely only on predefined graphs indirectly capture relations. this paper, we propose a Distillation Network (DSDN) utilize states migrate...

As an important step in natural language processing (NLP), text classification system has been widely used many fields, like spam filtering, news classification, and web page detection. Vector space model (VSM) is generally to extract feature vectors for representing texts which very classification. In this paper, a selection algorithm based on synonym merging named SM-CHI proposed. Besides, the improved CHI formula are select words so that accuracy of can be dimension reduced. addition,...

In this paper, we propose a new learning framework named dual set multi-label learning, where there are two sets of labels, and an object has one only positive label in each set. Compared to general the exclusive relationship among labels within same set, pairwise inter-set much more explicit likely be fully exploited. To handle such kind problems, novel boosting style algorithm with model-reuse distribution adjusting mechanisms is proposed make help other. addition, theoretical analyses...

This paper focuses on the analysis of spatially correlated functional data. The between-curve correlation is modeled by correlating principal component scores We propose a Spatial Principal Analysis Conditional Expectation framework to explicitly estimate spatial correlations and reconstruct individual curves. approach works even when observed data per curve are sparse. Assuming stationarity, empirical calculated as ratio eigenvalues smoothed covariance surface $Cov(X_i(s),X_i(t))$...

Abstract Batteries are the primary energy storage for electric vehicles. Often power of battery cannot be adequate to satisfy demands heavy loads. Simultaneously with source, secondary capacity, such as a super-capacitor, can used fulfil demand where ultra-capacitors meet high-frequency specifications. Ultracapacitor inhibit recycling high-current transient battery. In course impacts life, charging device high current. This paper is focused on PI control system vehicle management and...

Abstract This paper reports on the development and validation of a formal model for an automotive adaptive exterior lights system (ELS) with multiple variants in 6, which is most recent version lightweight specification language that supports mutable relations temporal logic. We explore different strategies to address variability, one pure another through annotative extension. then show how its can be used validate systems this nature, namely by checking reference scenarios are admissible,...

The Private Aggregation of Teacher Ensembles (PATE) framework is one the most promising recent approaches in differentially private learning. Existing theoretical analysis shows that PATE consistently learns any VC-classes realizable setting, but falls short explaining its success more general cases where error rate optimal classifier bounded away from zero. We fill this gap by introducing Tsybakov Noise Condition (TNC) and establish stronger interpretable learning bounds. These bounds...

Local Coupled Extreme Learning Machine (LCELM) is a recently-proposed variant of ELM, which assigns an address for each hidden-layer node and activates the when its activated degree less than given threshold.In this paper, improved version LCELM proposed by developing new way to initialize calculating with Gaussian kernel.The experimental comparison ELM demonstrates feasibility effectiveness improve obtains higher testing accuracy without significantly increasing training time ELM.

In terms of people's lifestyles, behavioral data can be easily collected and analyzed. How to protect utilize these is a hot topic today. There are many recognition models in academia, but the means identifying behavior still based on regular matching industry. The main reason impact work circles. Data reflect patterns, circles have catastrophic effect behaviors. working circle here defined as group people daily life. related permissions. Therefore, we introduce RBAC model incorporate...

Large-scale labeled dataset is the indispensable fuel that ignites AI revolution as we see today. Most such datasets are constructed using crowdsourcing services Amazon Mechanical Turk which provides noisy labels from non-experts at a fair price. The sheer size of mandates it only feasible to collect few per data point. We formulate problem test-time label aggregation statistical estimation inferring expected voting score. By imitating workers with supervised learners and them in doubly...

In task-oriented dialogue systems, Dialogue State Tracking (DST) aims to extract users' intentions from the history. Currently, most existing approaches suffer error propagation and are unable dynamically select relevant information when utilizing previous states. Moreover, relations between updates of different slots provide vital clues for DST. However, rely only on predefined graphs indirectly capture relations. this paper, we propose a Distillation Network (DSDN) utilize states migrate...

Black-box optimization (BBO) has become increasingly relevant for tackling complex decision-making problems, especially in public policy domains such as police districting. However, its broader application policymaking is hindered by the complexity of defining feasible regions and high-dimensionality decisions. This paper introduces a novel BBO framework, termed Conditional And Generative Optimization (CageBO). approach leverages conditional variational autoencoder to learn distribution...

Large-scale labeled dataset is the indispensable fuel that ignites AI revolution as we see today. Most such datasets are constructed using crowdsourcing services Amazon Mechanical Turk which provides noisy labels from non-experts at a fair price. The sheer size of mandates it only feasible to collect few per data point. We formulate problem test-time label aggregation statistical estimation inferring expected voting score. By imitating workers with supervised learners and them in doubly...