为了改善基于卷积编解码架构的单通道语音增强网络对语音声学特征提取不充分、解码特征丢失严重的问题,提出一种基于多路信息聚合协同解码的单通道语音增强网络MIACD,通过双路编码器充分提取融入了语音自监督学习(SSL)表征的幅度谱和复数谱特征,由4层Conformer分别从时间和频率维度对提取特征建模,采用残差连接将双路编码器提取的语音幅度、复数特征引入三路信息聚合解码器,并利用所提通道-时频注意力(CTF-Attention)机制根据语音能量分布情况调节解码器中聚合信息,有效缓解解码时可用声学信息缺失严重的问题。在公开数据集Voice Bank DEMAND上的实验结果表明,与用于单通道语音增强的协作学习框架(GaGNet)相比,MIACD在客观评价指标宽带感知评估语音质量(WB-PESQ)上提升了5.1%,短时客观可懂度(STOI)达到96.7%,验证所提方法可充分利用语音信息重构信号,有效抑制噪声并提升语音可理解性。
In the era of the Internet,widely used web applications have become the target of hacker attacks because they contain a large amount of personal information.Among these vulnerabilities,stealing private data through crosssite scripting(XSS)attacks is one of the most commonly used attacks by hackers.Currently,deep learning-based XSS attack detection methods have good application prospects;however,they suffer from problems such as being prone to overfitting,a high false alarm rate,and low accuracy.To address these issues,we propose a multi-stage feature extraction and fusion model for XSS detection based on Random Forest feature enhancement.The model utilizes RandomForests to capture the intrinsic structure and patterns of the data by extracting leaf node indices as features,which are subsequentlymergedwith the original data features to forma feature setwith richer information content.Further feature extraction is conducted through three parallel channels.Channel I utilizes parallel onedimensional convolutional layers(1Dconvolutional layers)with different convolutional kernel sizes to extract local features at different scales and performmulti-scale feature fusion;Channel II employsmaximum one-dimensional pooling layers(max 1D pooling layers)of various sizes to extract key features from the data;and Channel III extracts global information bi-directionally using a Bi-Directional Long-Short TermMemory Network(Bi-LSTM)and incorporates a multi-head attention mechanism to enhance global features.Finally,effective classification and prediction of XSS are performed by fusing the features of the three channels.To test the effectiveness of the model,we conduct experiments on six datasets.We achieve an accuracy of 100%on the UNSW-NB15 dataset and 99.99%on the CICIDS2017 dataset,which is higher than that of the existing models.
Qiurong QinYueqin LiYajie MiJinhui ShenKexin WuZhenzhao Wang
The rapid development of the Internet has accelerated the spread of rumors,posing challenges to social cohesion and stability.To address this,a multi-channel rumor propagation model incorporating individual game behavior and time delay is proposed.It depicts individuals strategically choosing propagation channels in the rumor spread process,capturing real-world intricacies more faithfully.Specifically,the model allowing spreaders to choose between text and video information base channels.Strategy adoption hinges on benefits versus costs,with payoffs dictating strategy and the propagation process determining an individual's state.By theoretical analysis of the model,the propagation threshold and equilibrium points are obtained.Then the stability of the model is further demonstrated based on Routh-Hurwitz judgment and Descartes'Rule of Signs.Numerical simulations are conducted to verify the correctness of the theoretical results and the sensitivity of the model to key parameters.The outcomes reveal that increasing the propagation cost of spreaders can significantly curb the spread of rumors.In contrast to the classical ISR model,rumors spread faster and more widely in the improved multi-channel rumor propagation model in this paper,which is a feature more aligned with real-world scenarios.Finally,the validity and predictive ability of the model are verified by using real rumor propagation data sets,indicating that the improved multi-channel rumor propagation model has good practical application and predictive value.
An advantageous porous architecture of electrodes is pivotal in significantly enhancing alkaline water electrolysis(AWE)efficiency by optimizing the mass transport mechanisms.This effect becomes even more pronounced when aiming to achieve elevated current densities.Herein,we employed a rapid and scalable laser texturing process to craft novel multi-channel porous electrodes.Particularly,the obtained electrodes exhibit the lowest Tafel slope of 79 mV dec^(-1)(HER)and 49 mV dec^(-1)(OER).As anticipated,the alkaline electrolyzer(AEL)cell incorporating multi-channel porous electrodes(NP-LT30)exhibited a remarkable improvement in cell efficiency,with voltage drops(from 2.28 to 1.97 V)exceeding 300 mV under 1 A cm^(-1),compared to conventional perforated Ni plate electrodes.This enhancement mainly stemmed from the employed multi-channel porous structure,facilitating mass transport and bubble dynamics through an innovative convection mode,surpassing the traditional convection mode.Furthermore,the NP-LT30-based AEL cell demonstrated exceptional durability for 300 h under 1.0 A cm^(-2).This study underscores the capability of the novel multi-channel porous electrodes to expedite mass transport in practical AWE applications.
