基于深度学习的系统日志异常检测研究

系统日志反映了系统运行状态,记录着系统中特定事件的活动信息,快速准确地检测出系统异常日志,对维护系统安全稳定具有重要意义。提出了一种基于GRU神经网络的日志异常检测算法,基于log key技术实现日志解析,利用执行路径的异常检测模型和参数值的异常检测模型实现日志异常检测,具有参数少、训练快的优点,在取得较高检测精度的同时提升了运行速度,适用于大型信息系统的日志分析。     

Distributed system anomaly detection using deep learning-based log analysis

Anomaly detection is a key step in ensuring the security and reliability of large-scale distributed systems. Analyzing system logs through artificial intelligence methods can quickly detect anomalies and thus help maintenance personnel to maintain system security. Most of the current works only focus on the temporal or spatial features of distributed system logs, and they cannot sufficiently extract the global features of distributed system logs to achieve a good correct rate of anomaly detection. To further address the shortcomings of existing methods, this paper proposes a deep learning model with global spatiotemporal features to detect the presence of anomalies in distributed system logs. First, we extract semi-structured log events from log templates and model them as natural language. In addition, we focus on the temporal characteristics of logs using the bidirectional long short-term memory network and the spatial invocation characteristics of logs using the Transformer. Extensive experimental evaluations show the advantages of our proposed model for distributed system log anomaly detection tasks. The optimal F1-Score on three open-source datasets and our own collected distributed system datasets reach 98.04%, 94.34%, 88.16%, and 97.40%, respectively.     

Landmark detection from mobile life log using a modular Bayesian network model

Mobile devices can now handle a great deal of information thanks to the convergence of diverse functionalities. Mobile environments have already shown great potential in terms of providing customized services to users because they can record meaningful and private information continually for long periods of time. Until now, most of this information has been generally ignored because of the limitations of mobile devices in terms of power, memory capacity and speed. In this paper, we propose a novel method that efficiently infers landmarks for users to overcome these problems. This method uses an effective probabilistic Bayesian network model for analyzing various kinds of log data in mobile environments, which were modularized in this paper to decrease complexity. We also present a cooperative inference method, and the proposed methods were evaluated with mobile log data generated and collected in the real world.     

Credit card fraud detection using a deep learning multistage model

The Journal of Supercomputing - The banking sector is on the eve of a serious transformation and the thrust behind it is artificial intelligence (AI). Novel AI applications have been already...     

TVD-MRDL: traffic violation detection system using MapReduce-based deep learning for large-scale data

Maintaining a fluid and safe traffic is a major challenge for human societies because of its social and economic impacts. Various technologies have considerably paved the way for the elimination of traffic problems and have been able to effectively detect drivers’ violations. However, the high volume of the real-time data collected from surveillance cameras and traffic sensors along with the data obtained from individuals have made the use of traditional methods ineffective. Therefore, using Hadoop for processing large-scale structured and unstructured data as well as multimedia data can be of great help. In this paper, the TVD-MRDL system based on the MapReduce techniques and deep learning was employed to discover effective solutions. The Distributed Deep Learning System was implemented to analyze traffic big data and to detect driver violations in Hadoop. The results indicated that more accurate monitoring automatically creates the power of deterrence and behavior change in drivers and it prevents drivers from committing unusual behaviors in society. So, if the offending driver is identified quickly after committing the violation and is punished with the appropriate punishment and dealt with decisively and without negligence, we will surely see a decrease in violations at the community level. Also, the efficiency of the TVD-MRDL performance increased by more than 75% as the number of data nodes increased.

     

Human activity recognition by combining external features with accelerometer sensor data using deep learning network model

Multimedia Tools and Applications - Various Human Activities are classified through time-series data generated by the sensors of wearable devices. Many real-time scenarios such as Healthcare...     

Satellite data cloud detection using deep learning supported by hyperspectral data

ABSTRACT

Deep learning methods can play an important role in satellite data cloud detection. The number and quality of training samples directly affect the accuracy of cloud detection based on deep learning. Therefore, selecting a large number of representative and high-quality training samples is a key step in cloud detection based on deep learning. For different satellite data sources, choosing sufficient and high-quality training samples has become an important factor limiting the application of deep learning in cloud detection. This paper presents a fast method for obtaining high-quality learning samples, which can be used for cloud detection of different satellite data with deep learning methods. AVIRIS (Airborne Visible Infrared Imaging Spectrometer) data, which have 224 continuous bands in the spectral range from 400–2500 nm, are used to provide cloud detection samples for different types of satellite data. Through visual interpretation, a sufficient number of cloud and clear sky pixels are selected from the AVIRIS data to construct a hyperspectral data sample library, which is used to simulate different satellite data (such as data from the Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat Operational Land Imager (OLI) satellites) as training samples. This approach avoids selecting training samples for different satellite sensors. Based on the Keras deep learning framework platform, a backpropagation (BP) neural network is employed for cloud detection from Landsat 8 OLI, National Polar-orbiting Partnership (NPP) Visible Infrared Imaging Radiometer Suite (VIIRS) and Terra MODIS data. The results are compared with cloud coverage results interpreted via artificial vision. The results demonstrate that the algorithm achieves good cloud detection results for the above data, and the overall accuracy is greater than 90%.     

Hierarchical visualization of network intrusion detection data

A technique for visualizing intrusion-detection system log files using hierarchical data based on IP addresses represents the number of incidents for thousands of computers in one display space. Our technique applies a hierarchical data visualization technique that represents leaf nodes as black square icons and branch nodes as rectangular borders enclosing the icons. This representation style visualizes thousands of hierarchical data leaf nodes equally in one display space. We applied the technique to bioactive chemical visualization and job distribution in parallel-computing environments.     

Vehicle re-identification using multi-task deep learning network and spatio-temporal model

Multimedia Tools and Applications - Vehicle re-identification (re-ID) plays an important role in the automatic analysis of the increasing urban surveillance videos and has become a hot topic in...     

BLoCNet: a hybrid,dataset-independent intrusion detection system using deep learning

International Journal of Information Security - Intrusion detection systems (IDS) identify cyber attacks given a sample of network traffic collected from real-world computer networks. As a powerful...     

BrC-MCDLM: breast Cancer detection using Multi-Channel deep learning model

Multimedia Tools and Applications - Breast cancer (BrC) is a lethal form of cancer which causes numerous deaths in women across the world. Generally, mammograms and histopathology biopsy images are...     

Influenza-like illness prediction using a long short-term memory deep learning model with multiple open data sources

The influenza problem has always been an important global issue. It not only affects people’s health problems but is also an essential topic of governments and health care facilities. Early prediction and response is the most effective control method for flu epidemics. It can effectively predict the influenza-like illness morbidity, and provide reliable information to the relevant facilities. For social facilities, it is possible to strengthen epidemic prevention and care for highly sick groups. It can also be used as a reminder for the public. This study collects information on the influenza-like illness emergency department visits to the Taiwan Centers for Disease Control, and the PM_2.5 open-source data from the Taiwan Environmental Protection Administration's air quality monitoring network. By using deep learning techniques, the relevance of short-term estimates and the outbreak calculation method can be determined. The techniques are published by the WHO to determine whether the influenza-like illness situation is still in a stage of reasonable control. Finally, historical data and future forecasted data are integrated on the web page for visual presentation, to show the actual regional air quality situation and influenza-like illness data and to predict whether there is an outbreak of influenza in the region.

     

A feed forward deep neural network model using feature selection for cloud intrusion detection system

The rapid advancement and growth of technology have rendered cloud computing services indispensable to our activities. Threats and intrusions have since multiplied exponentially across a range of industries. In such a scenario, the intrusion detection system, or simply the IDS, is deployed on the network to monitor and detect any attacks. The paper proposes a feed-forward deep neural network (FFDNN) method based on deep learning methodology using a filter-based feature selection model. The feature selection strategy aims to determine and select the most highly relevant subset of attributes from the feature importance score for training the deep learning model. Three benchmark data sets were used to assess the experiment: CIC-IDS 2017, UNSW-NB15, and NSL-KDD. In order to justify the proposed technique, a comparison was done using other learning algorithms ranging from classical machine learning to ensemble learning methods that can detect various attacks. The experiments showed that the FFDNN model with reduced feature subsets gave the highest accuracy of 99.53% and 94.45% in the NSL-KDD and UNSW-NB15 data sets, while the ensemble-based XGBoost model performed better in the CIC-IDS 2017 data set. In addition, the results show that the overall accuracy, recall, and F1 score of the deep learning algorithm are generally better for all the data sets.     

On construction of a network log management system using ELK Stack with Ceph

The Journal of Supercomputing - A log management system is essential for the networks administrator. With a log management tool, we can collect, store, analyze, archive, and finally dispose of the...     

KDSAE: Chronic kidney disease classification with multimedia data learning using deep stacked autoencoder network

In recent times, Chronic Kidney Disease (CKD) has affected more than 10% of the population worldwide and millions of people die every year. So, early-stage detection of CKD could be beneficial for increasing the life expectancy of suffering patients and reducing the treatment cost. It is required to build such a multimedia driven model which can help to diagnose the disease efficiently with higher accuracy before leading to worse conditions. Various techniques related to conventional machine learning models have been used by researchers in the past time without involvement of multimodal data-driven learning. This research paper offers a novel deep learning framework for chronic kidney disease classification using stacked autoencoder model utilizing multimedia data with a softmax classifier. The stacked autoencoder helps to extract the useful features from the dataset and then a softmax classifier is used to predict the final class. It has experimented on UCI dataset which contains early stages of 400 CKD patients with 25 attributes, which is a binary classification problem. Precision, recall, specificity and F1-score were used as evaluation metrics for the assessment of the proposed network. It was observed that this multimodal model outperformed the other conventional classifiers used for chronic kidney disease with a classification accuracy of 100%.

     

基于Preisach模型的深度学习网络迟滞建模

针对传统压电扫描器迟滞模型泛化能力较弱的问题,提出了一种基于Preisach模型的深度学习网络来建立迟滞模型,提高了模型的学习能力和泛化能力.具体而言,首先利用深度学习在深度特征提取方面的优势,建立包含卷积层、池化层、展开层以及深度特征层的深度学习层来提取输入电压信号的特征信息;其次,利用傅里叶变换层计算得到输入信号的频率,并将频率输入到非线性层,构造并输出了与输入信号频率相关的非线性项,该非线性项作为权值函数与Preisach模型的迟滞单元输出相乘,并将乘积叠加得到了频率相关的模型输出向量;最后,将深度学习层输出的特征向量与Preisach模型输出向量点乘,即可得到深度学习网络的最终输出位移.同时利用电容位移传感器采集的16组输入输出信号对深度学习网络进行训练,得到了网络中的权值参数,并利用其他8组输入输出数据对深度网络进行测试,训练和测试结果表明,本文所提出的基于Preisach模型的深度学习网络在得到高精度迟滞模型的同时,提高了模型的泛化能力.     

基于深度学习的安卓恶意应用检测

针对传统安卓恶意程序检测技术检测准确率低,对采用了重打包和代码混淆等技术的安卓恶意程序无法成功识别等问题,设计并实现了DeepDroid算法。首先,提取安卓应用程序的静态特征和动态特征,结合静态特征和动态特征生成应用程序的特征向量;然后,使用深度学习算法中的深度置信网络（DBN）对收集到的训练集进行训练,生成深度学习网络;最后,利用生成的深度学习网络对待测安卓应用程序进行检测。实验结果表明,在使用相同测试集的情况下,DeepDroid算法的正确率比支持向量机（SVM）算法高出3.96个百分点,比朴素贝叶斯（Naive Bayes）算法高出12.16个百分点,比K最邻近（KNN）算法高出13.62个百分点。DeepDroid算法结合了安卓应用程序的静态特征和动态特征,采用了动态检测和静态检测相结合的检测方法,弥补了静态检测代码覆盖率不足和动态检测误报率高的缺点,在特征识别的部分采用DBN算法使得网络训练速度得到保证的同时还有很高的检测正确率。     

Establish a patent risk prediction model for emerging technologies using deep learning and data augmentation

Technology patents are considered the source and bedrock of emerging technologies. Patents create value in any enterprise. However, obtaining patents is time consuming, expensive, and risky; especially if the patent application is rejected. The development of new patents requires extensive costs and resources, but sometimes they may be similar to other patents once the technology is fully developed. They might lack relevant patentable features and as a result, fail to pass the patent examination, resulting in investment losses. Patent infringement is also an especially important topic for reducing the risk of legal damages of patent holders, applicants, and manufacturers. Patent examinations have so far been performed manually. Due to manpower and time limitations, the examination time is exceedingly long and inefficient. Current patent similarity comparison research, and the classification algorithms of text mining are most commonly employed to provide analyses of the possibility of examination approval, but there is insufficient discussion about the possibility of infringement. However, if a new technology or innovation can be accurately determined in advance whether it likely to pass or fail (and why), or is at risk of patent infringement, losses can be mitigated.This research attempts to identify the issues involved in evaluating patent applications and infringement risks from existing patent databases. For each patent application, this research uses Convolutional Neural Networks, CNN + Long Short Term Memory Network, LSTM, prediction model, and the United States Patent and Trademark Office (USPTO) public utility patent application and reviews results based on keyword search. Then, data augmentation is utilized before performing model training; 10% of the approved and rejected applications are randomly selected as test cases, with the remaining 90% of the cases used to train the prediction model of this research in order to determine a model that can predict patent infringement and examination outcomes. Experimental results of the model in this study predicts that the accuracy of each classification is at least 87.7%, and can be used to find the classification of the reason for a rejection of a patent application failure.     

Plant leaf disease classification and damage detection system using deep learning models

Agriculture is the primary source of livelihood for about 70% of the rural population in India. The crop variety cultivated in India is very diverse. There are more than 500 crop varieties grown in India. Despite the technological advances, the agricultural practices are still manual and involve less automation than western countries. Most of the diseases affecting a plant will reflect the damage in the leaves. The diseases affecting the plant can thus be identified from the leaf images. This paper presents an automatic plant leaf damage detection and disease identification system. The first stage of the proposed method identifies the type of the disease based on the plant leaf image using DenseNet. The DenseNet model is trained on images categorized according to their nature, i.e., healthy and the type of the disease. This model is then used for testing new leaf images. The proposed DenseNet model produced a classification accuracy of 100%, with fewer images used during the training stage. The second stage identifies the damage in the leaf using deep learning-based semantic segmentation. Each RGB pixel value combination in the image is extracted, and supervised training is performed on the pixel values using the 1D Convolutional Neural Network (CNN). The trained model can detect the damage present in the leaves at a pixel level. Evaluation of the proposed semantic segmentation resulted in an accuracy of 97%. The third stage suggests a remedy for the disease based on the disease type and the damage state. The proposed method detects various defects in different plants in the experimental analysis, namely apple, grape, potato, and strawberry. The proposed model is compared with the existing techniques and obtained better performance in comparison with those methods.

     

基于深度学习的海岸线边缘检测网络模型

海岸线的动态监测对海岸带的规划管理具有非常重要的意义。由于海陆环境错综复杂,遥感影像中海陆边界光谱特征不明显,导致提取的海岸线定位不准确。提出一种融合语义分割网络和边缘检测网络的深度卷积神经网络模型(EWNet)。该模型包含2个分支流：语义分割流负责提取分层语义信息并用来指导边缘检测流获取岸线语义信息;边缘检测流通过语义分割流完善边缘语义信息。在“高分一号”遥感图像上的实验结果表明,与几种最新网络模型相比,EWNet获得了更精确的海岸线边界提取结果。