Sentiment analysis on Chinese microblogs has received extensive attention recently. Most previous studies focus identifying sentiment orientation by encoding as many word properties possible while they fail to consider contextual features (e.g., the long-range dependencies of words), which are, however, essentially important in analysis. In this paper, we propose a method incorporating word2vec model and stacked bidirectional long short-term memory (Stacked Bi-LSTM) model. We first employ...

Millimeter-wave and terahertz imaging systems have been used for security checks of human bodies, small postal packages, industrial nondestructive evaluation. The image resolution is one the key factors affecting accuracy contraband detection inner defect analysis. We propose a super-resolution reconstruction algorithm based on multi-scale channel attention (MS-CA) residual network to enhance images, which are acquired through homemade real-time system inspection. generalized block was...

The security and privacy of underwater acoustic sensor networks has received extensive attention recently due to the proliferation activities. This paper proposes an analytical model investigate eavesdropping attacks in networks. Our framework considers impacts various channel conditions (such as signal frequency, spreading factor wind speed) different hydrophones (isotropic array hydrophones) terms network nodes eavesdroppers. We also conduct simulations evaluate effectiveness accuracy our...

Abstract The problem of real-time detection welding defects is a difficult in resistance spot welding. It found that the dynamic has strong connection with growth nugget. signals are significantly different from those normal welding, and between show characteristics, same kind may also differ each other. most common practice today to realize resistive by extracting time-domain features signal waveforms. However, this approach highly subjective, so article proposes double-size mesh division...

Aiming at the problems of high manual cost, low efficiency, and precision mechanical axis health management in industrial robot applications, this paper proposes a assessment state prediction algorithm based on hidden Markov model (HMM) temporal convolutional networks (TCN). First, MPdist similarity comparison is used to construct index. Then trained with observable sensor data. After that, convolution neural network predict transition time iteratively, predicted results are decoded by HMM....

The blockchain technology becomes a key facilitator for Intelligent Manufacturing as it enables intelligent nodes to participate in global manufacturing networks with secure ledgers and smart contracts features. However, Traditional centralized storage cannot meet performance security requirements fully distributed consumes large amount of computing, network resources, which is inefficient difficult implement. In this paper, we propose clustering strategy on node community by constructing...

Abstract Background Cancer threat is relevant to age, and the of a foreshortened life coupled with lengthy treatment process negatively affects middle‐aged older adults. Understanding coping throughout cancer experience in survivors will help develop supportive care promote their physiological psychological effects. Objectives To explore experiences adults aged 40–59 over 60. Design A descriptive phenomenological study was employed. Methods Face‐to‐face, in‐depth, semistructured interviews...

To create an intelligent surface region of interests (ROI) 3D quantitative inspection strategy a reality in the continuous casting (CC) production line, improved laser image scanning system (3D-LDS) was established based on binocular imaging and deep-learning techniques. In 3D-LDS, firstly, to meet requirements industrial application, CCD method optimized high-temperature experiments secondly, we proposed novel proposal ROI initial depth location for effectively suppressing redundant...

In the past decades, due to low design cost and easy maintenance, text-based CAPTCHAs have been extensively used in constructing security mechanisms for user authentications. With recent advances machine/deep learning recognizing CAPTCHA images, growing attack methods are presented break CAPTCHAs. These machine learning/deep learning-based attacks often rely on training models massive volumes of data. The poorly constructed data also leads accuracy attacks. To investigate this issue, we...

The optimal control of denitrification system in coal-fired power plants China has recently received widespread attention. accurate prediction efficiency and formulate strategy can guide the operation better. Meanwhile, it achieve effect energy conservation Nitrogen oxides (NOx) reduction. In this paper, we take a domestic 1000 MW unit as an example, consider each major factors that affect selective catalytic reduction (SCR). We put forward deep reinforcement learning (DRL) model by...

To enhance the approximation and generalization ability of classical artificial neural network (ANN) by employing principles quantum computation, a quantum-inspired neuron based on controlled-rotation gate is proposed. In proposed model, discrete sequence input represented qubits, which, as control qubits after being rotated rotation gates, target qubit for rotation. The model output described probability amplitude state |1〉 in qubit. Then with (QNNSI) designed neurons to hidden layer layer....

One of recent proposals on standardizing quality user experience (QoE) video streaming over mobile network is Mean Opinion Score (vMOS), which can model QoE in 5 discrete grades. However, there are few studies quantifying vMOS and investigating the relationship between other service (QoS) parameters. In this paper, we address concern by proposing a novel data analytical framework based data. particular, our consists K-means clustering logistic regression. This integrates benefits both these...

Open banking brings both the opportunities and challenges to banks all over world. Due different economic development levels of each country gaps among financial environment maturity, countries have strategies regulations towards privacy protection data in scenes, such as General Data Protection Regulation (GDPR) by European Union (EU). A blockchain a continuously growing list records managed peer-to-peer network is widely used various application scenarios, can protect data. However, still...

Because of the high putrescibility body, in general condition, only part or fragmentary bones can be found when victim body was discovered. Therefore, first step criminal investigation is to reconstruct facial features order find victim. wrinkle most remarkable and intuitionistic feature human skin, generation method for reconstructed face an important reconstruction. Three improvements based on Bando's classic algorithm are represented basis analyzing special need craniofacial...