Showing papers by "N. R. Sunitha published in 2019"

Deep Learning-based Model to Fight Against Ad Click Fraud

Abstract: Click fraud is a fast-growing cyber-criminal activity with the aim of deceptively clicking on the advertisements to make the profit to the publisher or cause loss to the advertiser. Due to the popularity of smartphones since the last decade, most of the modern-day advertisement businesses have been shifting their focus toward mobile platforms. Nowadays, in-app advertisement on mobile platforms is the most targeted victim of click fraud. Malicious entities launch attacks by clicking ads to artificially increase the click rates of specific ads without the intention of using them for legitimate purposes. The fraud clicks are supposed to be caught by the ad providers as part of their service to the advertisers; however, there is a lack of research in the current literature for addressing and evaluating different techniques of click fraud detection and prevention. Another challenge toward click fraud detection is that the attack model can itself be an active learning system (smart attacker) with the aim of actively misleading the training process of fraud detection model via polluting the training data. In this paper, we propose a deep-learning based model to address the challenges as mentioned above. The model is a hybrid of artificial neural network (ANN), autoencoder and semi-supervised generative adversarial network (GAN). Our proposed approach triumphs excellent accuracy than other models.

Mini-Batch Normalized Mutual Information: A Hybrid Feature Selection Method

Abstract: Feature Selection has been a significant preprocessing procedure for classification in the area of Supervised Machine Learning. It is mostly applied when the attribute set is very large. The large set of attributes often tend to misguide the classifier. Extensive research has been performed to increase the efficacy of the predictor by finding the optimal set of features. The feature subset should be such that it enhances the classification accuracy by the removal of redundant features. We propose a new feature selection mechanism, an amalgamation of the filter and the wrapper techniques by taking into consideration the benefits of both the methods. Our hybrid model is based on a two phase process where we rank the features and then choose the best subset of features based on the ranking. We validated our model with various datasets, using multiple evaluation metrics. Furthermore, we have also compared and analyzed our results with previous works. The proposed model outperformed many existent algorithms and has given us good results.

A Multi-time-scale Time Series Analysis for Click Fraud Forecasting using Binary Labeled Imbalanced Dataset

Abstract: Click fraud refers to the practice of generating random clicks on a link in order to extract illegitimate revenue from the advertisers. We present a generalized model for modeling temporal click fraud data in the form of probability or learning based anomaly detection and time series modeling with time scales like minutes and hours. The proposed approach consists of seven stages: Pre-processing, data smoothing, fraudulent pattern identification, homogenizing variance, normalizing auto-correlation, developing the AR and MA models and fine tuning along with evaluation of the models. The objective of the proposed work is to first, model multi-time-scale time series data on AR/MA by relying only on time and the label without the need of too many attributes and secondly, to model different time scales separately on Auto-regression (AR) and Moving Average (MA) models. Then, we evaluate the models by tuning forecasting errors and also by minimizing Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) to obtain a best fit model for all time scale data. Through our experiments we also demonstrated that the Probability based model approach is better as compared to the Learning based probabilistic estimator model.

Learning-Based Model to Fight against Fake Like Clicks on Instagram Posts

Abstract: Online social networks (OSN) are one of the favorite places where people share posts like their photos, videos, and text to gain popularity. On the other hand, the marketing industry tries to gain the popularity of their advertisement using such OSNs. Popularity of a particular post depends on the number of likes received by that post. To increase one’s social worth, people try to use this market by artificially increasing the likes on their posts. There is a lack of research in the current literature on Instagram which is one of the growing OSNs. Our work focuses on detecting valid and fake like of posts with the application of learning model taking into consideration several popular factors. We developed an automated learning model to detect fake liking behavior on the Instagram post. The learned model can accurately differentiate between the legitimate and fake liker with an accuracy of 97% with ensemble-based learning model and also autoencoder is used to detect bots activity.

CKMI: Comprehensive Key Management Infrastructure Design for Industrial Automation and Control Systems

Abstract: Industrial Automation and Control Systems (IACS) are broadly utilized in critical infrastructures for monitoring and controlling the industrial processes remotely. The real-time transmissions in such systems provoke security breaches. Many security breaches have been reported impacting society severely. Hence, it is essential to achieve secure communication between the devices for creating a secure environment. For this to be effective, the keys used for secure communication must be protected against unauthorized disclosure, misuse, alteration or loss, which can be taken care of by a Key Management Infrastructure. In this paper, by considering the generic industrial automation network, a comprehensive key management infrastructure (CKMI) is designed for IACS. To design such an infrastructure, the proposed scheme employs ECDH, matrix method, and polynomial crypto mechanisms. The proposed design handles all the standard key management operations, viz. key generation, device registration, key establishment, key storage, device addition, key revocation, key update, key recovery, key archival, and key de-registration and destruction. The design supports secure communication between the same and different levels of IACS devices. The proposed design can be applied for major industrial automation networks to handle the key management operations. The performance analysis and implementation results highlight the benefits of the proposed design.

A Variation in the Working of Playfair Cipher

Abstract: Cryptography has decidedly been in the field of research for decades with the motif of enhancing the security of information exchange. This paper exhibits a variation in the implementation of a classical cipher technique, the Playfair cipher. The motive is to make the ciphertext produced less vulnerable to attacks; we have tested the same with a common attack, the brute-force attack. The proposed model is also statistically analyzed for vulnerability against the performance of the classical encryption technique.

Formal Verification of Authentication and Confidentiality for TACACS+ Security Protocol using Scyther

Abstract: Designing a perfect security protocol is a difficult task and requires a good effort and knowledge of Cryptography which is an art of secret writing. In order to achieve high reliability of security protocols, the testing technique is not suitable, because the testing technique has got many drawbacks. To achieve high reliability of security protocols, proving the correctness of security protocols is very much essential. To prove and verify the correctness of security protocols the Formal Verification technique is the best solution because it provides the mathematical proof for the correctness of security protocols. TACACS+ (Terminal Access Controller Access-Control System Plus) [6] is one the important security protocol used by most of the Cisco network communication devices to provide Authentication, Authorization, and Accountability (popularly known as AAA services) services to the host devices. In the proposed work, the TACACS+ security protocol is formally verified using the Model Checking technique. Using the Scyther [12] model checker the Confidentiality and Authentication security properties of TACACS+ security protocol is successfully verified.

Vehicle Detection Using Point Cloud and 3D LIDAR Sensor to Draw 3D Bounding Box

Abstract: In recent times, Autonomous driving functionalities is being developed by car manufacturers and is revolutionizing the automotive industries. Hybrid cars are prepared with a wide range of sensors such as ultrasound, LiDAR, camera, and radar. The results of these sensors are integrated in order to avoid collisions. To achieve accurate results a high structured point cloud surroundings can be used to estimate the scale and position. A point cloud is a set of Data points used to represent the 3D dimension in X, Y, Z direction. Point cloud divides data points into clusters that are processed in a pipeline. These clusters are collected to create a training set for object detection. In this paper, the cluster of vehicle objects and other objects are extracted and a supervised neural network is trained with the extracted objects for the binary classification of the objects representing vehicles or other objects. By learning global features and local features the vehicle objects represented in the point cloud are detected. These detected objects are fitted with a 3D bounding box to represent as a car object.

Ranked Keyword Search Result Verification to Detect Misbehaving Cloud Servers

Abstract: Nowadays, so many people outward their delicate information to the cloud in an encrypted format to maintain its privacy. On this basis, financial cloud computing becomes more efficient and useful. Protected keyword go through encrypted cloud information has fascinated many scholars for performing an effective data utilization. Current research is based on the impression that a cloud server is curious but truthful and therefore the search outcomes are not proved. But in real-world, sometimes cloud server might compromise and behave untruthfully. To maintain the reliability and secrecy, the outsourced delicate information should remain in the encrypted form. It is a task to go through the encoded data. There is a search facility to deliberate a ranked keyword search and there will be various data users and important files imported into the cloud. Hence, we propose a Ranked keyword search result verification for the detection of misbehaving cloud servers. We demonstrate two misbehaving scenarios of cloud server that will be detected by data user by using the relevance score of each file. In the whole process cloud server operates on the encrypted data.

AI-NLP Analytics: A thorough Comparative Investigation on India-USA Universities Branding on the Trending Social Media Platform “Instagram”

Abstract: The branding of the universities through social media has enormous influence among individuals. Social media has prospered through different networking sites such as Twitter, Facebook, Instagram, etc. However, in recent times, Instagram has become popular due to its unique features, which includes posting images, writing description alongside these images, and many more. People can even comment on these posts. The emotions expressed by people through their posts represent their real feelings and can be used for the analysis of public opinion regarding any topic. In our paper, we are applying the concepts of NLP on these textual descriptions and comments to perform a comparative investigation on the branding of the colleges/universities of India and USA. Our method provides a completely innovative way to view social media data. Our analysis can help the colleges/universities of both the country in providing an insight into their recognition among students through Instagram considering that the younger generation is largely influenced by Instagram. It can also help them in their rapid development and future goals.

A Novel Private Information Retrieval Technique for Private DNS Resolution

Abstract: Recently, advancements in various analytical techniques have enabled to encourage the unethical business to violate user privacy and market the analytical results. Existing user privacy-preserving techniques based on intractability assumptions have proved to offer only conditional user privacy. Thus, interest in perfect (i.e., unconditional) user privacy-preserving information retrieval techniques are receiving enormous attention. We have successfully constructed a single database perfect privacy-preserving information retrieval technique using Private Information Retrieval (PIR). We have proposed a novel perfect privacy-preserving PIR technique in a single database setting with non-trivial communication cost for private Domain Name System (DNS) resolution. We have further extended the proposed scheme to a computationally efficient scheme by varying the security parameter without losing the level of user privacy.

An efficient user privacy preserving multi-server private information retrieval: an efficient privacy preserving information storage and retrieval framework

Abstract: The thirst of acquiring and sharing the knowledge has been increased exponentially in these days due to the availability of the Internet at the finger tips. As a consequence, the need of privacy at various levels and various contexts has been comprehensively studied by various cryptographers. One of the user privacy preserving concepts called Private Information Retrieval (PIR) also provides user privacy at various levels. Several years of efforts on PIR have succeeded to provide almost practical communication efficient privacy preserving solutions.