Visible to the public Biblio

Filters: Keyword is Prediction algorithms  [Clear All Filters]
2021-10-12
Radhakrishnan, C., Karthick, K., Asokan, R..  2020.  Ensemble Learning Based Network Anomaly Detection Using Clustered Generalization of the Features. 2020 2nd International Conference on Advances in Computing, Communication Control and Networking (ICACCCN). :157–162.
Due to the extraordinary volume of business information, classy cyber-attacks pointing the networks of all enterprise have become more casual, with intruders trying to pierce vast into and grasp broader from the compromised network machines. The vital security essential is that field experts and the network administrators have a common terminology to share the attempt of intruders to invoke the system and to rapidly assist each other retort to all kind of threats. Given the enormous huge system traffic, traditional Machine Learning (ML) algorithms will provide ineffective predictions of the network anomaly. Thereby, a hybridized multi-model system can improve the accuracy of detecting the intrusion in the networks. In this manner, this article presents a novel approach Clustered Generalization oriented Ensemble Learning Model (CGELM) for predicting the network anomaly. The performance metrics of the anticipated approach are Detection Rate (DR) and False Predictive Rate (FPR) for the two heterogeneous data sets namely NSL-KDD and UGR'16. The proposed method provides 98.93% accuracy for DR and 0.14% of FPR against Decision Stump AdaBoost and Stacking Ensemble methods.
Franchina, L., Socal, A..  2020.  Innovative Predictive Model for Smart City Security Risk Assessment. 2020 43rd International Convention on Information, Communication and Electronic Technology (MIPRO). :1831–1836.
In a Smart City, new technologies such as big data analytics, data fusion and artificial intelligence will increase awareness by measuring many phenomena and storing a huge amount of data. 5G will allow communication of these data among different infrastructures instantaneously. In a Smart City, security aspects are going to be a major concern. Some drawbacks, such as vulnerabilities of a highly integrated system and information overload, must be considered. To overcome these downsides, an innovative predictive model for Smart City security risk assessment has been developed. Risk metrics and indicators are defined by considering data coming from a wide range of sensors. An innovative ``what if'' algorithm is introduced to identify critical infrastructures functional relationship. Therefore, it is possible to evaluate the effects of an incident that involves one infrastructure over the others.
2021-09-30
Ren, Xun-yi, Luo, Qi-qi, Shi, Chen, Huang, Jia-ming.  2020.  Network Security Posture Prediction Based on SAPSO-Elman Neural Networks. 2020 International Conference on Artificial Intelligence and Computer Engineering (ICAICE). :533–537.
With the increasing popularity of the Internet, mobile Internet and the Internet of Things, the current network environment continues to become more complicated. Due to the increasing variety and severity of cybersecurity threats, traditional means of network security protection have ushered in a huge challenge. The network security posture prediction can effectively predict the network development trend in the future time based on the collected network history data, so this paper proposes an algorithm based on simulated annealing-particle swarm algorithm to optimize improved Elman neural network parameters to achieve posture prediction for network security. Taking advantage of the characteristic that the value of network security posture has periodicity, a simulated annealing algorithm is introduced along with an improved particle swarm algorithm to solve the problem that neural network training is prone to fall into a local optimal solution and achieve accurate prediction of the network security posture. Comparison of the proposed scheme with existing prediction methods validates that the scheme has a good posture prediction accuracy.
Peng, Cheng, Yongli, Wang, Boyi, Yao, Yuanyuan, Huang, Jiazhong, Lu, Qiao, Peng.  2020.  Cyber Security Situational Awareness Jointly Utilizing Ball K-Means and RBF Neural Networks. 2020 17th International Computer Conference on Wavelet Active Media Technology and Information Processing (ICCWAMTIP). :261–265.
Low accuracy and slow speed of predictions for cyber security situational awareness. This paper proposes a network security situational awareness model based on accelerated accurate k-means radial basis function (RBF) neural network, the model uses the ball k-means clustering algorithm to cluster the input samples, to get the nodes of the hidden layer of the RBF neural network, speeding up the selection of the initial center point of the RBF neural network, and optimize the parameters of the RBF neural network structure. Finally, use the training data set to train the neural network, using the test data set to test the accuracy of this neural network structure, the results show that this method has a greater improvement in training speed and accuracy than other neural networks.
2021-09-21
Yan, Fan, Liu, Jia, Gu, Liang, Chen, Zelong.  2020.  A Semi-Supervised Learning Scheme to Detect Unknown DGA Domain Names Based on Graph Analysis. 2020 IEEE 19th International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom). :1578–1583.
A large amount of malware families use the domain generation algorithms (DGA) to randomly generate a large amount of domain names. It is a good way to bypass conventional blacklists of domain names, because we cannot predict which of the randomly generated domain names are selected for command and control (C&C) communications. An effective approach for detecting known DGA families is to investigate the malware with reverse engineering to find the adopted generation algorithms. As reverse engineering cannot handle the variants of DGA families, some researches leverage supervised learning to find new variants. However, the explainability of supervised learning is low and cannot find previously unseen DGA families. In this paper, we propose a graph-based semi-supervised learning scheme to track the evolution of known DGA families and find previously unseen DGA families. With a domain relation graph, we can clearly figure out how new variants relate to known DGA domain names, which induces better explainability. We deployed the proposed scheme on real network scenarios and show that the proposed scheme can not only comprehensively and precisely find known DGA families, but also can find new DGA families which have not seen before.
bin Asad, Ashub, Mansur, Raiyan, Zawad, Safir, Evan, Nahian, Hossain, Muhammad Iqbal.  2020.  Analysis of Malware Prediction Based on Infection Rate Using Machine Learning Techniques. 2020 IEEE Region 10 Symposium (TENSYMP). :706–709.
In this modern, technological age, the internet has been adopted by the masses. And with it, the danger of malicious attacks by cybercriminals have increased. These attacks are done via Malware, and have resulted in billions of dollars of financial damage. This makes the prevention of malicious attacks an essential part of the battle against cybercrime. In this paper, we are applying machine learning algorithms to predict the malware infection rates of computers based on its features. We are using supervised machine learning algorithms and gradient boosting algorithms. We have collected a publicly available dataset, which was divided into two parts, one being the training set, and the other will be the testing set. After conducting four different experiments using the aforementioned algorithms, it has been discovered that LightGBM is the best model with an AUC Score of 0.73926.
Brzezinski Meyer, Maria Laura, Labit, Yann.  2020.  Combining Machine Learning and Behavior Analysis Techniques for Network Security. 2020 International Conference on Information Networking (ICOIN). :580–583.
Network traffic attacks are increasingly common and varied, this is a big problem especially when the target network is centralized. The creation of IDS (Intrusion Detection Systems) capable of detecting various types of attacks is necessary. Machine learning algorithms are widely used in the classification of data, bringing a good result in the area of computer networks. In addition, the analysis of entropy and distance between data sets are also very effective in detecting anomalies. However, each technique has its limitations, so this work aims to study their combination in order to improve their performance and create a new intrusion detection system capable of well detect some of the most common attacks. Reliability indices will be used as metrics to the combination decision and they will be updated in each new dataset according to the decision made earlier.
2021-09-08
R, Naveen, Chaitanya, N.S.V, M, Nikhil Srinivas, Vineeth, Nandhini.  2020.  Implementation of a Methodology for Detection and Prevention of Security Attacks in Vehicular Adhoc Networks. 2020 IEEE International Conference for Innovation in Technology (INOCON). :1–6.
In the current generation, road accidents and security problems increase dramatically worldwide in our day to day life. In order to overcome this, Vehicular Ad-hoc Network (VANETs) is considered as a key element of future Intelligent Transportation Systems (ITS). With the advancement in vehicular communications, the attacks have also increased, and such architecture is still exposed to many weaknesses which led to numerous security threats that must be addressed before VANET technology is practically and safely adopted. Distributed Denial of Service (DDoS) attack, replay attacks and Sybil attacks are the significant security threats that affect the communication and privacy in VANET. An algorithm to detect and prevent various kinds of security attacks in VANET communication has been designed and proposed in this work. An analysis has also been done by applying four protocols on an existing scenario of real traffic simulator using OpenStreetMap and the best suitable protocol has been selected for further application. The evaluation has been done using SUMO, NS3 and Java simulation environment. Simulation results and extensive performance analysis shows that our proposed Algorithm performs well in detecting and preventing the attacks in VANET communication.
Ali, Jehad, Roh, Byeong-hee, Lee, Byungkyu, Oh, Jimyung, Adil, Muhammad.  2020.  A Machine Learning Framework for Prevention of Software-Defined Networking Controller from DDoS Attacks and Dimensionality Reduction of Big Data. 2020 International Conference on Information and Communication Technology Convergence (ICTC). :515–519.
The controller is an indispensable entity in software-defined networking (SDN), as it maintains a global view of the underlying network. However, if the controller fails to respond to the network due to a distributed denial of service (DDoS) attacks. Then, the attacker takes charge of the whole network via launching a spoof controller and can also modify the flow tables. Hence, faster, and accurate detection of DDoS attacks against the controller will make the SDN reliable and secure. Moreover, the Internet traffic is drastically increasing due to unprecedented growth of connected devices. Consequently, the processing of large number of requests cause a performance bottleneck regarding SDN controller. In this paper, we propose a hierarchical control plane SDN architecture for multi-domain communication that uses a statistical method called principal component analysis (PCA) to reduce the dimensionality of the big data traffic and the support vector machine (SVM) classifier is employed to detect a DDoS attack. SVM has high accuracy and less false positive rate while the PCA filters attribute drastically. Consequently, the performance of classification and accuracy is improved while the false positive rate is reduced.
2021-08-31
Ge, Chonghui, Sun, Jian, Sun, Yuxin, Di, Yunlong, Zhu, Yongjin, Xie, Linfeng, Zhang, Yingzhou.  2020.  Reversible Database Watermarking Based on Random Forest and Genetic Algorithm. 2020 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC). :239—247.
The advancing information technology is playing more and more important role in data mining of relational database.1 The transfer and sharing of databases cause the copyright-related security threats. Database watermarking technology can effectively solve the problem with copyright protection and traceability, which has been attracting researchers' attention. In this paper, we proposed a novel, robust and reversible database watermarking technique, named histogram shifting watermarking based on random forest and genetic algorithm (RF-GAHCSW). It greatly improves the watermark capacity by means of histogram width reduction and eliminates the impact of the prediction error attack. Meanwhile, random forest algorithm is used to select important attributes for watermark embedding, and genetic algorithm is employed to find the optimal secret key for the database grouping and determine the position of watermark embedding to improve the watermark capacity and reduce data distortion. The experimental results show that the robustness of RF-GAHCSW is greatly improved, compared with the original HSW, and the distortion has little effect on the usability of database.
2021-08-17
Zheng, Gang, Xu, Xinzhong, Wang, Chao.  2020.  An Effective Target Address Generation Method for IPv6 Address Scan. 2020 IEEE 6th International Conference on Computer and Communications (ICCC). :73–77.
In recent years, IPv6 and its application are more and more widely deployed. Most network devices support and open IPv6 protocol stack. The security of IPv6 network is also concerned. In the IPv6 network security technology, address scanning is a key and difficult point. This paper presents a TGAs-based IPv6 address scanning method. It takes the known alive IPv6 addresses as input, and then utilizes the information entropy and clustering technology to mine the distribution law of seed addresses. Then, the final optimized target address set can be obtained by expanding from the seed address set according to the distribution law. Experimental results show that it can effectively improve the efficiency of IPv6 address scanning.
2021-08-11
Gaikwad, Nikhil B., Ugale, Hrishikesh, Keskar, Avinash, Shivaprakash, N. C..  2020.  The Internet-of-Battlefield-Things (IoBT)-Based Enemy Localization Using Soldiers Location and Gunshot Direction. IEEE Internet of Things Journal. 7:11725–11734.
The real-time information of enemy locations is capable to transform the outcome of combat operations. Such information gathered using connected soldiers on the Internet of Battlefield Things (IoBT) is highly beneficial to create situational awareness (SA) and to plan an effective war strategy. This article presents the novel enemy localization method that uses the soldier's own locations and their gunshot direction. The hardware prototype has been developed that uses a triangulation for an enemy localization in two soldiers and a single enemy scenario. 4.24±1.77 m of average localization error and ±4° of gunshot direction error has been observed during this prototype testing. This basic model is further extended using three-stage software simulation for multiple soldiers and multiple enemy scenarios with the necessary assumptions. The effective algorithm has been proposed, which differentiates between the ghost and true predictions by analyzing the groups of subsequent shooting intents (i.e., frames). Four different complex scenarios are tested in the first stage of the simulation, around three to six frames are required for the accurate enemy localization in the relatively simple cases, and nine frames are required for the complex cases. The random error within ±4° in gunshot direction is included in the second stage of the simulation which required almost double the number of frames for similar four cases. As the number of frames increases, the accuracy of the proposed algorithm improves and better ghost point elimination is observed. In the third stage, two conventional clustering algorithms are implemented to validate the presented work. The comparative analysis shows that the proposed algorithm is faster, computationally simple, consistent, and reliable compared with others. Detailed analysis of hardware and software results for various scenarios has been discussed in this article.
2021-05-25
Satılmış, Hami, Akleylek, Sedat.  2020.  Efficient Implementation of HashSieve Algorithm for Lattice-Based Cryptography. 2020 International Conference on Information Security and Cryptology (ISCTURKEY). :75—79.
The security of lattice-based cryptosystems that are secure for the post-quantum period is based on the difficulty of the shortest vector problem (SVP) and the closest vector problem (CVP). In the literature, many sieving algorithms are proposed to solve these hard problems. In this paper, efficient implementation of HashSieve sieving algorithm is discussed. A modular software library to have an efficient implementation of HashSieve algorithm is developed. Modular software library is used as an infrastructure in order for the HashSieve efficient implementation to be better than the sample in the literature (Laarhoven's standard HashSieve implementation). According to the experimental results, it is observed that HashSieve efficient implementation has a better running time than the example in the literature. It is concluded that both implementations are close to each other in terms of the memory space used.
2021-05-18
Niloy, Nishat Tasnim, Islam, Md. Shariful.  2020.  IntellCache: An Intelligent Web Caching Scheme for Multimedia Contents. 2020 Joint 9th International Conference on Informatics, Electronics Vision (ICIEV) and 2020 4th International Conference on Imaging, Vision Pattern Recognition (icIVPR). :1–6.
The traditional reactive web caching system is getting less popular day by day due to its inefficiency in handling the overwhelming requests for multimedia content. An intelligent web caching system intends to take optimal cache decisions by predicting future popular contents (FPC) proactively. In recent years, a few approaches have proposed some intelligent caching system where they were concerned about proactive caching. Those works intensified the importance of FPC prediction using the prediction models. However, only FPC prediction may not help to get the optimal solution in every scenario. In this paper, a technique named IntellCache has been proposed that increases the caching efficiency by taking a cache decision i.e. content storing decision before storing the predicted FPC. Different deep learning models such as- multilayer perceptron (MLP), Long short-term memory (LSTM) of Recurrent Neural Network (RNN) and ConvLSTM a combination of LSTM and Convolutional Neural Network (CNN) are compared to identify the most efficient model for FPC. The information on the contents of 18 years from the MovieLens data repository has been mined to evaluate the proposed approach. Results show that this proposed scheme outperforms previous solutions by achieving a higher cache hit ratio and lower average delay and thus, ensures users' satisfaction.
2021-05-13
Ho, Tsung-Yu, Chen, Wei-An, Huang, Chiung-Ying.  2020.  The Burden of Artificial Intelligence on Internal Security Detection. 2020 IEEE 17th International Conference on Smart Communities: Improving Quality of Life Using ICT, IoT and AI (HONET). :148—150.
Our research team have devoted to extract internal malicious behavior by monitoring the network traffic for many years. We applied the deep learning approach to recognize the malicious patterns within network, but this methodology may lead to more works to examine the results from AI models production. Hence, this paper addressed the scenario to consider the burden of AI, and proposed an idea for long-term reliable detection in the future work.
Sheng, Mingren, Liu, Hongri, Yang, Xu, Wang, Wei, Huang, Junheng, Wang, Bailing.  2020.  Network Security Situation Prediction in Software Defined Networking Data Plane. 2020 IEEE International Conference on Advances in Electrical Engineering and Computer Applications( AEECA). :475–479.
Software-Defined Networking (SDN) simplifies network management by separating the control plane from the data forwarding plane. However, the plane separation technology introduces many new loopholes in the SDN data plane. In order to facilitate taking proactive measures to reduce the damage degree of network security events, this paper proposes a security situation prediction method based on particle swarm optimization algorithm and long-short-term memory neural network for network security events on the SDN data plane. According to the statistical information of the security incident, the analytic hierarchy process is used to calculate the SDN data plane security situation risk value. Then use the historical data of the security situation risk value to build an artificial neural network prediction model. Finally, a prediction model is used to predict the future security situation risk value. Experiments show that this method has good prediction accuracy and stability.
2021-05-05
Rathod, Jash, Joshi, Chaitali, Khochare, Janavi, Kazi, Faruk.  2020.  Interpreting a Black-Box Model used for SCADA Attack detection in Gas Pipelines Control System. 2020 IEEE 17th India Council International Conference (INDICON). :1—7.
Various Machine Learning techniques are considered to be "black-boxes" because of their limited interpretability and explainability. This cannot be afforded, especially in the domain of Cyber-Physical Systems, where there can be huge losses of infrastructure of industries and Governments. Supervisory Control And Data Acquisition (SCADA) systems need to detect and be protected from cyber-attacks. Thus, we need to adopt approaches that make the system secure, can explain predictions made by model, and interpret the model in a human-understandable format. Recently, Autoencoders have shown great success in attack detection in SCADA systems. Numerous interpretable machine learning techniques are developed to help us explain and interpret models. The work presented here is a novel approach to use techniques like Local Interpretable Model-Agnostic Explanations (LIME) and Layer-wise Relevance Propagation (LRP) for interpretation of Autoencoder networks trained on a Gas Pipelines Control System to detect attacks in the system.
2021-04-27
Li, Y., Zhou, Y., Hu, K., Sun, N., Ke, K..  2020.  A Security Situation Prediction Method Based on Improved Deep Belief Network. 2020 IEEE 2nd International Conference on Civil Aviation Safety and Information Technology (ICCASIT. :594–598.
With the rapid development of smart grids and the continuous deepening of informatization, while realizing remote telemetry and remote control of massive data-based grid operation, electricity information network security problems have become more serious and prominent. A method for electricity information network security situation prediction method based on improved deep belief network is proposed in this paper. Firstly, the affinity propagation clustering algorithm is used to determine the depth of the deep belief network and the number of hidden layer nodes based on sample parameters. Secondly, continuously adjust the scaling factor and crossover probability in the differential evolution algorithm according to the population similarity. Finally, a chaotic search method is used to perform a second search for the best individuals and similarity centers of each generation of the population. Simulation experiments show that the proposed algorithm not only enhances the generalization ability of electricity information network security situation prediction, but also has higher prediction accuracy.
2021-03-01
Raj, C., Khular, L., Raj, G..  2020.  Clustering Based Incident Handling For Anomaly Detection in Cloud Infrastructures. 2020 10th International Conference on Cloud Computing, Data Science Engineering (Confluence). :611–616.
Incident Handling for Cloud Infrastructures focuses on how the clustering based and non-clustering based algorithms can be implemented. Our research focuses in identifying anomalies and suspicious activities that might happen inside a Cloud Infrastructure over available datasets. A brief study has been conducted, where a network statistics dataset the NSL-KDD, has been chosen as the model to be worked upon, such that it can mirror the Cloud Infrastructure and its components. An important aspect of cloud security is to implement anomaly detection mechanisms, in order to monitor the incidents that inhibit the development and the efficiency of the cloud. Several methods have been discovered which help in achieving our present goal, some of these are highlighted as the following; by applying algorithm such as the Local Outlier Factor to cancel the noise created by irrelevant data points, by applying the DBSCAN algorithm which can detect less denser areas in order to identify their cause of clustering, the K-Means algorithm to generate positive and negative clusters to identify the anomalous clusters and by applying the Isolation Forest algorithm in order to implement decision based approach to detect anomalies. The best algorithm would help in finding and fixing the anomalies efficiently and would help us in developing an Incident Handling model for the Cloud.
2021-02-23
Ashraf, S., Ahmed, T..  2020.  Sagacious Intrusion Detection Strategy in Sensor Network. 2020 International Conference on UK-China Emerging Technologies (UCET). :1—4.
Almost all smart appliances are operated through wireless sensor networks. With the passage of time, due to various applications, the WSN becomes prone to various external attacks. Preventing such attacks, Intrusion Detection strategy (IDS) is very crucial to secure the network from the malicious attackers. The proposed IDS methodology discovers the pattern in large data corpus which works for different types of algorithms to detect four types of Denial of service (DoS) attacks, namely, Grayhole, Blackhole, Flooding, and TDMA. The state-of-the-art detection algorithms, such as KNN, Naïve Bayes, Logistic Regression, Support Vector Machine (SVM), and ANN are applied to the data corpus and analyze the performance in detecting the attacks. The analysis shows that these algorithms are applicable for the detection and prediction of unavoidable attacks and can be recommended for network experts and analysts.
Olowononi, F. O., Rawat, D. B., Liu, C..  2020.  Dependable Adaptive Mobility in Vehicular Networks for Resilient Mobile Cyber Physical Systems. 2020 IEEE International Conference on Communications Workshops (ICC Workshops). :1—6.

Improved safety, high mobility and environmental concerns in transportation systems across the world and the corresponding developments in information and communication technologies continue to drive attention towards Intelligent Transportation Systems (ITS). This is evident in advanced driver-assistance systems such as lane departure warning, adaptive cruise control and collision avoidance. However, in connected and autonomous vehicles, the efficient functionality of these applications depends largely on the ability of a vehicle to accurately predict it operating parameters such as location and speed. The ability to predict the immediate future/next location (or speed) of a vehicle or its ability to predict neighbors help in guaranteeing integrity, availability and accountability, thus boosting safety and resiliency of the Vehicular Network for Mobile Cyber Physical Systems (VCPS). In this paper, we proposed a secure movement-prediction for connected vehicles by using Kalman filter. Specifically, Kalman filter predicts the locations and speeds of individual vehicles with reference to already observed and known information such posted legal speed limit, geographic/road location, direction etc. The aim is to achieve resilience through the predicted and exchanged information between connected moving vehicles in an adaptive manner. By being able to predict their future locations, the following vehicle is able to adjust its position more accurately to avoid collision and to ensure optimal information exchange among vehicles.

2021-02-01
Behera, S., Prathuri, J. R..  2020.  Application of Homomorphic Encryption in Machine Learning. 2020 2nd PhD Colloquium on Ethically Driven Innovation and Technology for Society (PhD EDITS). :1–2.
The linear regression is a machine learning algorithm used for prediction. But if the input data is in plaintext form then there is a high probability that the sensitive information will get leaked. To overcome this, here we are proposing a method where the input data is encrypted using Homomorphic encryption. The machine learning algorithm can be used on this encrypted data for prediction while maintaining the privacy and secrecy of the sensitive data. The output from this model will be an encrypted result. This encrypted result will be decrypted using a Homomorphic decryption technique to get the plain text. To determine the accuracy of our result, we will compare it with the result obtained after applying the linear regression algorithm on the plain text.
2021-01-25
Chen, J., Lin, X., Shi, Z., Liu, Y..  2020.  Link Prediction Adversarial Attack Via Iterative Gradient Attack. IEEE Transactions on Computational Social Systems. 7:1081–1094.
Increasing deep neural networks are applied in solving graph evolved tasks, such as node classification and link prediction. However, the vulnerability of deep models can be revealed using carefully crafted adversarial examples generated by various adversarial attack methods. To explore this security problem, we define the link prediction adversarial attack problem and put forward a novel iterative gradient attack (IGA) strategy using the gradient information in the trained graph autoencoder (GAE) model. Not surprisingly, GAE can be fooled by an adversarial graph with a few links perturbed on the clean one. The results on comprehensive experiments of different real-world graphs indicate that most deep models and even the state-of-the-art link prediction algorithms cannot escape the adversarial attack, such as GAE. We can benefit the attack as an efficient privacy protection tool from the link prediction of unknown violations. On the other hand, the adversarial attack is a robust evaluation metric for current link prediction algorithms of their defensibility.
2021-01-11
Jiang, P., Liao, S..  2020.  Differential Privacy Online Learning Based on the Composition Theorem. 2020 IEEE 10th International Conference on Electronics Information and Emergency Communication (ICEIEC). :200–203.
Privacy protection is becoming more and more important in the era of big data. Differential privacy is a rigorous and provable privacy protection method that can protect privacy for a single piece of data. But existing differential privacy online learning methods have great limitations in the scope of application and accuracy. Aiming at this problem, we propose a more general and accurate algorithm, named DPOL-CT, for differential privacy online learning. We first distinguish the difference in differential privacy protection between offline learning and online learning. Then we prove that the DPOL-CT algorithm achieves (∊, δ)-differential privacy for online learning under the Gaussian, the Laplace and the Staircase mechanisms and enjoys a sublinear expected regret bound. We further discuss the trade-off between the differential privacy level and the regret bound. Theoretical analysis and experimental results show that the DPOL-CT algorithm has good performance guarantees.
Li, Y., Chang, T.-H., Chi, C.-Y..  2020.  Secure Federated Averaging Algorithm with Differential Privacy. 2020 IEEE 30th International Workshop on Machine Learning for Signal Processing (MLSP). :1–6.
Federated learning (FL), as a recent advance of distributed machine learning, is capable of learning a model over the network without directly accessing the client's raw data. Nevertheless, the clients' sensitive information can still be exposed to adversaries via differential attacks on messages exchanged between the parameter server and clients. In this paper, we consider the widely used federating averaging (FedAvg) algorithm and propose to enhance the data privacy by the differential privacy (DP) technique, which obfuscates the exchanged messages by properly adding Gaussian noise. We analytically show that the proposed secure FedAvg algorithm maintains an O(l/T) convergence rate, where T is the total number of stochastic gradient descent (SGD) updates for local model parameters. Moreover, we demonstrate how various algorithm parameters can impact on the algorithm communication efficiency. Experiment results are presented to justify the obtained analytical results on the performance of the proposed algorithm in terms of testing accuracy.