融入注意力机制的轻量化可回收垃圾检测方法

目的 针对目前智能垃圾分类设备使用的垃圾检测方法存在检测速度慢且模型权重文件较大等问题,提出一种基于YOLOv4的轻量化方法,以实现可回收垃圾的检测。方法 采用MobileNetV2轻量级网络为YOLOv4的主干网络,用深度可分离卷积来优化颈部和头部网络,以减少参数量和计算量,提高检测速度;在颈部网络中融入CBAM注意力模块,提高模型对目标特征信息的敏感度;使用K-means算法重新聚类,得到适合自建可回收数据集中检测目标的先验框。结果 实验结果表明,改进后模型的参数量减少为原始YOLOv4模型的17.0%,检测的平均精度达到96.78%,模型权重文件的大小为46.6 MB,约为YOLOv4模型权重文件的19.1%,检测速度为20.46帧/s,提高了约25.4%,检测精度和检测速度均满足实时检测要求。结论 改进的YOLOv4模型能够在检测可回收垃圾时保证较高的检测精度,同时具有较好的实时性。     

Multi-Zone-Wise Blockchain Based Intrusion Detection and Prevention System for IoT Environment

Blockchain merges technology with the Internet of Things (IoT) for addressing security and privacy-related issues. However, conventional blockchain suffers from scalability issues due to its linear structure, which increases the storage overhead, and Intrusion detection performed was limited with attack severity, leading to performance degradation. To overcome these issues, we proposed MZWB (Multi-Zone-Wise Blockchain) model. Initially, all the authenticated IoT nodes in the network ensure their legitimacy by using the Enhanced Blowfish Algorithm (EBA), considering several metrics. Then, the legitimately considered nodes for network construction for managing the network using Bayesian-Direct Acyclic Graph (B-DAG), which considers several metrics. The intrusion detection is performed based on two tiers. In the first tier, a Deep Convolution Neural Network (DCNN) analyzes the data packets by extracting packet flow features to classify the packets as normal, malicious, and suspicious. In the second tier, the suspicious packets are classified as normal or malicious using the Generative Adversarial Network (GAN). Finally, intrusion scenario performed reconstruction to reduce the severity of attacks in which Improved Monkey Optimization (IMO) is used for attack path discovery by considering several metrics, and the Graph cut utilized algorithm for attack scenario reconstruction (ASR). UNSW-NB15 and BoT-IoT utilized datasets for the MZWB method simulated using a Network simulator (NS-3.26). Compared with previous performance metrics such as energy consumption, storage overhead accuracy, response time, attack detection rate, precision, recall, and F-measure. The simulation result shows that the proposed MZWB method achieves high performance than existing works     

基于交叉验证SVM的网络入侵检测

针对传统入侵检测系统漏报率和误报率高的问题,将支持向量机(SVM)应用于入侵检测中,提出了在SVM学习过程中引入交叉验证的方法,采用径向基函数(RBF)作为核,将训练集分成若干子集,每一子集使用其它子集训练得到的分类器进行测试,获得RBF的两个最佳参数后,将其应用于最终的分类器.实验结果表明,该方法能够有效检测入侵攻击,具有更高的检测率和更强的泛化能力,同时具有较低的误报率和漏报率,可以有效地运用于入侵检测系统中.     

智能神经网络在Internet入侵检测中的应用

探讨了一个基于智能神经网络的网络入侵检测系统模型，在对网络中的IP数据包进行分析处理以及特征提取的基础上，采用智能神经网络进行学习或判别，以达到对未知数据包进行检测的目的，智能神经网络可以将多种多样的入侵检测任务划分为多个单一的检测任务，并将这些任务分配给功能专一，结构简单的较小的智能神经网络来完成，实验证明这是一种行之有效的网络入侵检测的解决方法。     

Long Short Term Memory Networks Based Anomaly Detection for KPIs

In real-world many internet-based service companies need to closely monitor large amounts of data in order to ensure stable operation of their business. However, anomaly detection for these data with various patterns and data quality has been a great challenge, especially without labels. In this paper, we adopt an anomaly detection algorithm based on Long Short-Term Memory (LSTM) Network in terms of reconstructing KPIs and predicting KPIs. They use the reconstruction error and prediction error respectively as the criteria for judging anomalies, and we test our method with real data from a company in the insurance industry and achieved good performance.     

基于注意力机制的轻量级RGB-D图像语义分割网络

目的针对卷积神经网络在RGB-D(彩色-深度)图像中进行语义分割任务时模型参数量大且分割精度不高的问题,提出一种融合高效通道注意力机制的轻量级语义分割网络。方法文中网络基于RefineNet,利用深度可分离卷积(Depthwiseseparableconvolution)来轻量化网络模型,并在编码网络和解码网络中分别融合高效的通道注意力机制。首先RGB-D图像通过带有通道注意力机制的编码器网络,分别对RGB图像和深度图像进行特征提取;然后经过融合模块将2种特征进行多维度融合;最后融合特征经过轻量化的解码器网络得到分割结果,并与RefineNet等6种网络的分割结果进行对比分析。结果对提出的算法在语义分割网络常用公开数据集上进行了实验,实验结果显示文中网络模型参数为90.41 MB,且平均交并比(mIoU)比RefineNet网络提高了1.7%,达到了45.3%。结论实验结果表明,文中网络在参数量大幅减少的情况下还能提高了语义分割精度。     

基于全局注意力机制和LSTM的连续手语识别算法

目的 为提高连续手语识别准确率,缓解听障人群与非听障人群的沟通障碍。方法 提出了基于全局注意力机制和LSTM的连续手语识别算法。通过帧间差分法对视频数据进行预处理,消除视频冗余帧,借助ResNet网络提取特征序列。通过注意力机制加权,获得全局手语状态特征,并利用LSTM进行时序分析,形成一种基于全局注意力机制和LSTM的连续手语识别算法,实现连续手语识别。结果 实验结果表明,该算法在中文连续手语数据集CSL上的平均识别率为90.08%,平均词错误率为41.2%,与5种算法相比,该方法在识别准确率与翻译性能上具有优势。结论 基于全局注意力机制和LSTM的连续手语识别算法实现了连续手语识别,并且具有较好的识别效果及翻译性能,对促进听障人群无障碍融入社会方面具有积极的意义。     

Network Intrusion Detection Based on Feature Selection and Hybrid Metaheuristic Optimization

Applications of internet-of-things (IoT) are increasingly being used in many facets of our daily life, which results in an enormous volume of data. Cloud computing and fog computing, two of the most common technologies used in IoT applications, have led to major security concerns. Cyberattacks are on the rise as a result of the usage of these technologies since present security measures are insufficient. Several artificial intelligence (AI) based security solutions, such as intrusion detection systems (IDS), have been proposed in recent years. Intelligent technologies that require data preprocessing and machine learning algorithm-performance augmentation require the use of feature selection (FS) techniques to increase classification accuracy by minimizing the number of features selected. On the other hand, metaheuristic optimization algorithms have been widely used in feature selection in recent decades. In this paper, we proposed a hybrid optimization algorithm for feature selection in IDS. The proposed algorithm is based on grey wolf (GW), and dipper throated optimization (DTO) algorithms and is referred to as GWDTO. The proposed algorithm has a better balance between the exploration and exploitation steps of the optimization process and thus could achieve better performance. On the employed IoT-IDS dataset, the performance of the proposed GWDTO algorithm was assessed using a set of evaluation metrics and compared to other optimization approaches in the literature to validate its superiority. In addition, a statistical analysis is performed to assess the stability and effectiveness of the proposed approach. Experimental results confirmed the superiority of the proposed approach in boosting the classification accuracy of the intrusion in IoT-based networks.     

Statistical Process Control‐Based Intrusion Detection and Monitoring

Intrusion detection systems have a vital role in protecting computer networks and information systems. In this article, we applied a statistical process control (SPC)–monitoring concept to a certain type of traffic data to detect a network intrusion. We proposed an SPC‐based intrusion detection process and described it and the source and the preparation of data used in this article. We extracted sample data sets that represent various situations, calculated event intensities for each situation, and stored these sample data sets in the data repository for use in future research. This article applies SPC charting methods for intrusion detection. In particular, it uses the basic security module host audit data from the MIT Lincoln Laboratory and applies the Shewhart chart, the cumulative sum chart, and the exponential weighted moving average chart to detect a denial of service intrusion attack. The case study shows that these SPC techniques are useful for detecting and monitoring intrusions. Copyright © 2013 John Wiley & Sons, Ltd.     

粒子群算法和时间序列分析在入侵检测系统中的应用

入侵检测系统是一种被动的安全防御方法。它是通过分析各种收集到的数据来发现可能的入侵行为。常用的入侵检测分类方法不仅算法复杂而且效率还偏低。本文提出一种基于粒子群算法和时间序列相结合的半监督入侵检测方法来提高入侵检测的分类效率。实验结果表明,该方法用于入侵检测系统具有较高的检测率。     

A Fused Machine Learning Approach for Intrusion Detection System

The rapid growth in data generation and increased use of computer network devices has amplified the infrastructures of internet. The interconnectivity of networks has brought various complexities in maintaining network availability, consistency, and discretion. Machine learning based intrusion detection systems have become essential to monitor network traffic for malicious and illicit activities. An intrusion detection system controls the flow of network traffic with the help of computer systems. Various deep learning algorithms in intrusion detection systems have played a prominent role in identifying and analyzing intrusions in network traffic. For this purpose, when the network traffic encounters known or unknown intrusions in the network, a machine-learning framework is needed to identify and/or verify network intrusion. The Intrusion detection scheme empowered with a fused machine learning technique (IDS-FMLT) is proposed to detect intrusion in a heterogeneous network that consists of different source networks and to protect the network from malicious attacks. The proposed IDS-FMLT system model obtained 95.18% validation accuracy and a 4.82% miss rate in intrusion detection.     

Intrusion Detection System Using FKNN and Improved PSO

Intrusion detection system (IDS) techniques are used in cybersecurity to protect and safeguard sensitive assets. The increasing network security risks can be mitigated by implementing effective IDS methods as a defense mechanism. The proposed research presents an IDS model based on the methodology of the adaptive fuzzy k-nearest neighbor (FKNN) algorithm. Using this method, two parameters, i.e., the neighborhood size (k) and fuzzy strength parameter (m) were characterized by implementing the particle swarm optimization (PSO). In addition to being used for FKNN parametric optimization, PSO is also used for selecting the conditional feature subsets for detection. To proficiently regulate the indigenous and comprehensive search skill of the PSO approach, two control parameters containing the time-varying inertia weight (TVIW) and time-varying acceleration coefficients (TVAC) were applied to the system. In addition, continuous and binary PSO algorithms were both executed on a multi-core platform. The proposed IDS model was compared with other state-of-the-art classifiers. The results of the proposed methodology are superior to the rest of the techniques in terms of the classification accuracy, precision, recall, and f-score. The results showed that the proposed methods gave the highest performance scores compared to the other conventional algorithms in detecting all the attack types in two datasets. Moreover, the proposed method was able to obtain a large number of true positives and negatives, with minimal number of false positives and negatives.     

A Novel Bidirectional LSTM and Attention Mechanism Based Neural Network for Answer Selection in Community Question Answering

Deep learning models have been shown to have great advantages in answer selection tasks. The existing models, which employ encoder-decoder recurrent neural network (RNN), have been demonstrated to be effective. However, the traditional RNN-based models still suffer from limitations such as 1) high-dimensional data representation in natural language processing and 2) biased attentive weights for subsequent words in traditional time series models. In this study, a new answer selection model is proposed based on the Bidirectional Long Short-Term Memory (Bi-LSTM) and attention mechanism. The proposed model is able to generate the more effective question-answer pair representation. Experiments on a question answering dataset that includes information from multiple fields show the great advantages of our proposed model. Specifically, we achieve a maximum improvement of 3.8% over the classical LSTM model in terms of mean average precision.     

A New Database Intrusion Detection Approach Based on Hybrid Meta-Heuristics

A new secured database management system architecture using intrusion detection systems (IDS) is proposed in this paper for organizations with no previous role mapping for users. A simple representation of Structured Query Language queries is proposed to easily permit the use of the worked clustering algorithm. A new clustering algorithm that uses a tube search with adaptive memory is applied to database log files to create users’ profiles. Then, queries issued for each user are checked against the related user profile using a classifier to determine whether or not each query is malicious. The IDS will stop query execution or report the threat to the responsible person if the query is malicious. A simple classifier based on the Euclidean distance is used and the issued query is transformed to the proposed simple representation using a classifier, where the Euclidean distance between the centers and the profile’s issued query is calculated. A synthetic data set is used for our experimental evaluations. Normal user access behavior in relation to the database is modelled using the data set. The false negative (FN) and false positive (FP) rates are used to compare our proposed algorithm with other methods. The experimental results indicate that our proposed method results in very small FN and FP rates.     

一种基于注意机制和卷积神经网络的视觉模型

针对目前的深度卷积神经网络(CNN)模型规模大、训练参数多、计算速度慢以及难以移植到移动端等问题,提出了一种深度可分离卷积结合3重注意机制模块(DSC-TAM)的视觉模型.首先,通过深度可分离卷积网络来减少模型参数,提高网络模型的计算速度;其次,引入3重注意机制模块提高网络的特征提取能力,改善网络性能.实验结果表明:该...     

基于注意力与特征融合的工程机械目标检测方法

目的 针对施工环境中工程机械目标大小不一、相互遮挡、工作形态各异等问题,提出一种基于注意力与特征融合的目标检测方法(AT–FFRCNN)。方法 在主干网络中采用ResNet50和特征路径聚合网络PFPN,融合不同尺度的特征信息,在区域建议网络(RPN)和全连接层引入注意力机制,提高目标识别的能力,在损失函数中使用广义交并比(GIoU),提高目标框的准确性。结果 实验表明,文中提出方法检测准确率比其他方法有较大提高,检测平均准确率(mAP)达到90%以上。结论 能够较好地完成工程机械目标的检测任务。     

基于改进注意力机制的认知障碍病程分类

阿尔茨海默病(AD)是一种发病进程缓慢、随着时间不断恶化的神经退化性疾病,在老龄化的趋势下,AD患者数量日渐增加。因此,如何对其予以早期精准诊断并进行正向干预是急需解决的问题。为提高计算机辅助诊断的效率,同时促进疾病的病理生理机制研究,提出了改进的基于SE模块二维双路径融合网络的分类方法,在网络中加入缩减系数模块,增加图片有用信息占比;对通道注意模块的权重函数重新设计,增大特征图间差异,联合二维双路径网络,增大网络倚重点,达到更好分类性能的同时,防止模型过拟合。使用ADNI数据集对AD、EMCI、NC进行二分类,实验表明所提出模型准确度相比于VGG和二维双路径融合模型分别提高了5.59%和8.11%,与其它先进方法进行比较验证了所提方法的可行性。     

Unknown Attack Detection: Combining Relabeling and Hybrid Intrusion Detection

Detection of unknown attacks like a zero-day attack is a research field that has long been studied. Recently, advances in Machine Learning (ML) and Artificial Intelligence (AI) have led to the emergence of many kinds of attack-generation tools developed using these technologies to evade detection skillfully. Anomaly detection and misuse detection are the most commonly used techniques for detecting intrusion by unknown attacks. Although anomaly detection is adequate for detecting unknown attacks, its disadvantage is the possibility of high false alarms. Misuse detection has low false alarms; its limitation is that it can detect only known attacks. To overcome such limitations, many researchers have proposed a hybrid intrusion detection that integrates these two detection techniques. This method can overcome the limitations of conventional methods and works better in detecting unknown attacks. However, this method does not accurately classify attacks like similar to normal or known attacks. Therefore, we proposed a hybrid intrusion detection to detect unknown attacks similar to normal and known attacks. In anomaly detection, the model was designed to perform normal detection using Fuzzy c-means (FCM) and identify attacks hidden in normal predicted data using relabeling. In misuse detection, the model was designed to detect previously known attacks using Classification and Regression Trees (CART) and apply Isolation Forest (iForest) to classify unknown attacks hidden in known attacks. As an experiment result, the application of relabeling improved attack detection accuracy in anomaly detection by approximately 11% and enhanced the performance of unknown attack detection in misuse detection by approximately 10%.     

Evaluation of Intrusion Detection Systems

This paper presents a comprehensive method for evaluating intrusion detection systems (IDSs). It integrates and extends ROC (receiver operating characteristic) and cost analysis methods to provide an expected cost metric. Results are given for determining the optimal operation of an IDS based on this expected cost metric. Results are given for the operation of a single IDS and for a combination of two IDSs. The method is illustrated for: 1) determining the best operating point for a single and double IDS based on the costs of mistakes and the hostility of the operating environment as represented in the prior probability of intrusion and 2) evaluating single and double IDSs on the basis of expected cost. A method is also described for representing a compound IDS as an equivalent single IDS. Results are presented from the point of view of a system administrator, but they apply equally to designers of IDSs.     

基于注意力机制的包装命名实体识别

目的 为了解决包装行业相关文本命名实体识别困难问题,提出在BiLSTM（Bidirectional Long Short-Term Memory）神经网络中加入注意力机制（Attention）和字词联合特征,构建一种基于注意力机制的BiLSTM深度学习模型（简称Attention-BiLSTM）,以识别包装命名实体。方法 首先构建包装领域词典匹配包装语料中词语的类别特征,同时将包装语料转换为字特征和词特征联合的向量特征,并且在过程中加入POS（词性）信息。然后将以上特征联合馈送到BiLSTM网络,以获取文本的全局特征,并利用注意力机制获取局部特征。最后根据文本的全局特征和局部特征使用CRF（Conditional Random Field）解码整个句子的最优标注序列。结果 通过对《中国包装网》新闻数据集的实验,获得了85.6%的F值。结论 所提方法在包装命名实体识别中优于传统方法。