A novel NMF-based authentication scheme for encrypted speech in cloud computing

Multimedia Tools and Applications - Authentication of encrypted speeches is a technique that can judge the integrity of encrypted speech in cloud computing, even the encrypted speeches have been...     

基于多模态信息融合的时间序列预测模型

针对传统单因子模型无法充分利用时间序列相关信息,以及这些模型对时间序列预测准确性和可靠性较差的问题,提出一种基于多模态信息融合的时间序列预测模型——Skip-Fusion对多模态数据中的文本数据和数值数据进行融合。首先利用BERT（Bidirectional Encoder Representations from Transformers）预训练模型和独热编码对不同类别的文本数据进行编码表示;再使用基于全局注意力机制的预训练模型获得多文本特征融合的单一向量表示;然后将得到的单一向量表示与数值数据按时间顺序对齐;最后通过时间卷积网络（TCN）模型实现文本和数值特征的融合,并通过跳跃连接完成多模态数据的浅层和深层特征的再次融合。在股票价格序列的数据集上进行实验,Skip-Fusion模型的均方根误差（RMSE）和日收益（R）分别为0.492和0.930,均优于现有的单模态模型和多模态融合模型的结果,同时在可决系数（R-Squared）上取得了0.955的拟合优度。实验结果表明,Skip-Fusion模型能够有效进行多模态信息融合并具有较高的预测准确性和可靠性。     

一种流数据立方体分析挖掘框架

流数据是目前一种重要的数据展现形式,对流数据进行OLAM(联机分析挖掘)操作可为分析人员提供多层次的数据视图。但OLAM要求在不同粒度中实现对数据的聚合操作,而流式数据内含时态特性和持续到达特性,使得数据无法被多次重复操作。使用传统OLAP(联机分析处理)方法无法生成部分物化视图且流数据规模宏大,受限于存储空间大小而无法保存全部数据单元信息。针对上述问题,提出了一种基于概要技术的流数据OLAM框架——sketch cube(概要立方体),该框架把任意维度组合映射成唯一自然数,根据上下限单调原则对维度组合裁剪,在类线性空间中保存有效数据单元信息,并构建时间序列索引提高检索效率。通过理论分析给出使用sketch cube的前提条件,同时通过真实海量流数据实验分析表明,sketch sube在有效性、存储空间效率和正确率上可以满足实时挖掘的需求。     

Speech authentication and content recovery scheme for security communication and storage

With the rapid development of Internet, it brings a lot of conveniences. However, the data transmission and storage are faced with some security issues that seem to be obstacles to overcome, such as privacy protection and integrity authentication. In this paper, an efficient speech watermarking algorithm is proposed for content authentication and recovery in encrypted domain. The proposed system consists of speech encryption, watermark generation and embedding, content authentication and recovery. In the encryption process, chaotic and block cipher are combined to eliminate the positional correlation and conceal the statistical feature. In the watermark embedding process, approximation coefficients of integer wavelet transform are used to generate watermark and the detail coefficients are reserved to carry watermark. Theoretical analysis and simulation results show that the proposed scheme has high security and excellent inaudibility. Compared with previous works, the proposed scheme has strong ability to detect de-synchronization attacks and locate the corresponding tampered area without using synchronization codes. Meanwhile, the selective encryption will not influence the selective watermarking operation. Similarly, the operation of watermarking will not affect the decryption of the encrypted speech. Additionally, the tampered samples can be recovered without any auxiliary watermark information.     

基于希尔伯特−黄变换的音频信号复制粘贴篡改检测算法

复制粘贴篡改是音频篡改常用的篡改方式。当被篡改的音频段来自于同一音频且伴随着后处理操作时,这种篡改往往难以检测。针对有后处理的音频信号复制粘贴篡改,文章提出了一种基于希尔伯特−黄变换的音频信号复制粘贴篡改检测和定位的方法。采用音高跟踪技术将音频信号的有声段和无声段区分开来,对每个有声段分别实行希尔伯特−黄变换以提取谱图特征,进而利用动态时间规整（DTW）来计算每个谱图特征的相似度,当满足设定阈值条件时,通过比较希尔伯特谱图特征的DTW值来检测音频复制粘贴篡改,同时根据DTW值对应有声段的索引来定位复制粘贴篡改的位置。实验结果表明：该方法对几种常用的音频复制粘贴后处理操作都具有较好的鲁棒性;与基于梅尔谱图的音频信号复制粘贴篡改检测方法相比,所提方法检测准确率高、召回率高、鲁棒性好。

     

学习产出驱动的大数据课程教学探索与实践

大数据技术发展变化快,综合应用性强,而大数据计算技术课程的教学内容偏重理论,脱离实际工程应用,培养的学生实践能力弱,无法适应企业的实际需求.依据新工科培养思路和OBE理念,提出构建基于综合案例的大数据计算技术的课程内容;设计贯穿整个课程体系的综合案例,激发学生的主动探索的积极性,提高其实践能力;结合产学合作模式,形成更...     

How do you visit: Identifying addicts from large-scale transit records via scenario deep embedding

Identification of individuals based on transit modes is of great importance in user tracking systems. However, identifying users in real-life studies is not trivial owing to the following challenges: 1) activity data containing both temporal and spatial context are high-order and sparse; 2) traditional two-step classifiers depend on trajectory patterns as input features, which limits accuracy especially in the case of scattered and diverse data; 3) in some cases, there are few positive instances and they are difficult to detect. Therefore, approaches involving statistics-based or trajectory-based features do not work effectively. Deep learning methods also suffer from the problem of how to represent trajectory vectors for user classification. Here, we propose a novel end-to-end scenario-based deep learning method to address these challenges, based on the observation that individuals may visit the same place for different reasons. We first define a scenario using critical places and related trajectories. Next, we embed scenarios via path-based or graph-based approaches using extended embedding techniques. Finally, a two-level convolution neural network is constructed for the classification. Our model is applied to the problem of detection of addicts using transit records directly without feature engineering, based on real-life data collected from mobile devices. Based on constructed scenario with dense trajectories, our model outperforms classical classification approaches, anomaly detection methods, state-of-the-art sequential deep learning models, and graph neural networks. Moreover, we provide statistical analyses and intuitiveexplanations to interpret the characteristics of resident and addict mobility. Our method could be generalized to other trajectory-related tasks involving scattered and diverse data.

     

人工智能在军事通信网络中的应用:基于“零信任”方法

军事团体对Artificial Intelligence(AI)技术、网络安全及遭遇威胁时的弹性作业能力的关注以应对敌方的全新网络攻击,以AI驱动的自动武器系统将面临更加严格的基于“零信任”方法的审查,以迎接全新挑战并开创智能武器的新局面。应用存在的安全隐患,并提出了相应的安全防护措施,以期为计算机网络信息系统的正常使用提供一定的参考意见。     

Research on cloud computing users’ public safety trust model based on scorecard-random forest

Traditional cloud computing trust models mainly focused on the calculation of the trust of users’ behavior.In the process of classification and evaluation,there were some problems such as ignorance of content security and lack of trust division verification.Aiming to solve these problems,cloud computing users’ public safety trust model based on scorecard-random forest was proposed.Firstly,the text was processed using Word2Vec in the data preprocessing stage.The convolution neural network (CNN) was used to extract the sentence features for user content tag classification.Then,scorecard method was used to filter the strong correlation index.Meanwhile,in order to establish the users’ public safety trust evaluation model in cloud computing,a random forest method was applied.Experimental results show that the proposed users’ public safety trust evaluation model outperforms the general trust evaluation model.The proposed model can effectively distinguish malicious users from normal users,and it can improve the efficiency of the cloud computing users management.     

基于特征对抗对的视觉特征归因网络研究

可视化图像关键特征区域是计算机视觉一个重要而需要深入研究的问题.图像关键特征区域可视化的技术已经在弱监督定位和理解数据隐藏特征的领域中得到广泛应用.近年来,基于神经网络分类器的特征定位显示已成为最新的技术,并且通常用于医学和自然图像数据集上.但存在特征定位显示不精确的缺陷.针对传统神经网络分类器在可视化图像关键特征区域标注方法上的局限性,提出了一种基于生成对抗对特征的关键特征区域可视化方法(即视觉特征归因方法).该方法通过构造关键特征区域对抗对,采用生成和鉴别对抗网络生成关键特征区域,可有效过滤冗余信息并实现精准定位,有效解决了疾病特征可视化问题.在该方法中,为了解决传统生成对抗网络难以达到负载均衡的缺陷,采用了Wasserstein距离解决协调其训练平衡的问题,同时使用梯度惩罚加速收敛过程.在人工合成数据集、肺部数据集和心脏数据集上的实验结果表明,提出的方法在视觉显示的定性和定量的问题中,均产生了理想的真实效果图,非常接近观察到的效果.