Advances in Data and Information Sciences

Delay Tolerant Network isSharma, Diksha aKumar, Sanjay newNagwani, Naresh Kumar emerging technology, delivering messages in a challenged network termed as Intermittently Connected Networks (ICNs), lacking continuous end-to-end connectivity, having low data rate and high propagation delay. Routing of bundles is an area of interest in DTN. Spray and Wait is a DTN routing protocol that outstrips other DTN routing protocols ProPHET, Epidemic in performance metric overhead ratio. The performance of the Spray and Wait protocol in other metrics is intended to be elevated in this work. The proposed algorithm implements weighted dissemination of messages instead of even dissemination in the spraying phase. Number of replicas to be transmitted to an encountered node is decided on the basis of its delivery probability. The proposed algorithm transmits less number of replicas to the node having greater delivery probability as they have more chances of encountering the destination node. The algorithm explores more possible ways to find a suitable hop as it also considers giving packets to nodes having low probability considering the situation that it might encounter a node having the best probability to deliver.

Networked control systems (NCSs) are significant and foremost multidisciplinary research areas for many decades. This paper is mainly oriented toward recent developments and challenges of network-induced delays due to inclusion of data network in NCSs. Network delays deteriorate the control performance and stability of the NCSs. The time-varying delays can be measured in real time by calculating the time difference of sending and receiving control packets. Various compensation techniques are reviewed to mitigate the effect of constant, time-varying, and stochastic delay. Lastly, some conclusions are drawn and the future research scope is directed.

Generally, if an image is received as a Cipher, it is assumed that the data (to be sent) might be an image, but text can also be encrypted in the format of an image. So to confuse the attacker, this paper proposes a technique that encrypts the text into image using piecewise linear chaotic map (PWLCM). A text file can be taken as input and the proposed algorithm will convert it into a cipher image, which can then be converted back by the decryption process. The advantage of this scheme is that it is quite secure and hard to break. Also, PWLCM Maps is the simplest among chaotic maps. At first, the text is converted from 7-bit ASCII to its double equivalent and then padding is done to get the required matrix to form a structure for the image to which the text will be converted. After that permutation is done to the matrix bits and then diffusion occurs at two stages, first for the rows and then for the columns. Both permutation and diffusion are done using PWLCM map.

The idea to connect everything to anything and at any point of time is what vaguely defines the concept of Internet of Things (IoT). The concept of IoT is not only about providing connectivity but also facilitating interaction among these connected things. Though the term IoT was introduced in 1999 but has drawn significant attention during the past few years. The pace at which new devices are being integrated into the system will profoundly impact the world in a good way but also poses some serious threats with regard to security and privacy. IoT in its current form is susceptible to a multitudinous set of attacks. One of the greatest concerns of IoT is to provide security assurance for the data exchange because data is vulnerable to a number of attacks by the attackers at each layer of IoT. The IoT has layered structure, where each layer provides a service. The security vary from layer to layer as each layer serves a different purpose. The aim of this paper is to analyze the various security and privacy threats related to IoT. Furthermore, this paper also discusses numerous existing security protocols operating at different layers, potential attacks, and suggested countermeasures.

Blockchain is the latest technology which is used in cryptocurrencies such as bitcoin and ether. Blockchain is the decentralized distributed ledger, which is based on the peer-to-peer network method. As the blockchain is mainly developed for implementation in the virtual cryptocurrency such as bitcoin, so its main purpose is clear that is security. Even though the healthcare industry is leading in majority of fields whether it is technology, equipment, researches, medicines, etc. We have even reached to remote locations through IoT devices and but one major thing is still lacking that is security of the data. Huge amount of data is being generated everyday from patient’s medical checkups, treatments, symptoms, etc., which is to be dealt with care as they are very crucial and can be tempered by hackers which can lead to serious problems. Therefore, such critical data must need to be secured with blockchain as it makes it very difficult to tamper with data. This paper deals with the implementation of the blockchain to solve the above-mentioned problems.

We are living in an era where structured and unstructured data is produced, consumed and stored in enormous amount on frequent basis. Database transactions, social media, images, audios, videos etc. are the major sources responsible for generating big data in huge capacity and diversity. This usually consists of large volumes of complex and growing data sets with numerous self-regulating sources that are difficult to process with the conventional techniques of data management. Using big data mining, organizations are able to extract useful evidences from these large data sets. In spite of big data gains, there are numerous challenges also and among these challenges maintaining data privacy is the most important concern in big data mining applications since processing large scale of sensitive data sets such as health record, banking transaction records needs to be maintained in such a way that the private data should not be revealed to any unauthorized person. This paper provides a review of big data, challenges in big data mining and the privacy concern in big data.

This paper describes a technique that encrypts images in the form of DNA sequences using PWLCM system, i.e., Piecewise Linear Chaotic Map. This method has two times DNA-encoding–decoding operations along with DNA-permutation and DNA-diffusion operations to get the cipher image. On comparison with other processes, the advantage of this algorithm is easy to compute but confuses a cryptanalyst a lot. Apart from that dual-layer DNA-encoding–decoding processes in the algorithm result in good encryption outputs. When outputs are subjected to different security analysis to find out the strength of the algorithm, results with encrypted images with higher values of UACI, NPCR, key space, information entropy, and good correlation coefficient. This results in strong resistivity toward widely used attacks.

With the development in technology, the security of transmission and storage of digital information (basically, digital images) is a challenge to all cryptographic researchers. In recent years, for securing digital images multiple encryption techniques have been proposed. Among them, one-dimensional (1D) chaotic map based image encryption techniques render better security in storage and transmission of images. 1D chaotic maps are simple in structure and hence efficient to implement in both software and hardware. An image encryption based on simple permutation and diffusion operation using various 1D chaotic maps is proposed in this paper. The proposed technique first performs pixel permutation operation using various 1D chaotic maps and then performs pixel diffusion operation using pixel key generated by the Secure Hash Algorithm-256 and the plain image. Also in this paper, a comparative analysis of security is presented using various 1D chaotic maps in image encryptions. The comparative results show the best security of using most of the 1D chaotic maps in image encryptions.

Internet-of-things (IoT) became very popular in today’s research. IoT means all devices of a particular system should be connected with each other through the internet. Cloud Computing and Wireless Sensor Networks (WSN) are integrated for efficient management in IoT. This integration is known as Sensor Cloud. This technology has a lot of applications due to the continuous development of information and communication technology. Although sensor cloud has several advantages still it has many research challenges like energy efficiency, security, QoS, etc. The wireless sensor network is the network of sensors which operate on battery. Reducing energy consumption and communication overhead are important issues of wireless sensor networks. Efficient management of WSN and cloud results in efficient management of IoT. This paper presents a survey on efficient management of IoT with sensor cloud.

The inclusion of smartphone/mobile nodes as a part of grid computing increases the computation limits of the static grid while at the same time adds to the complexity owing to the associated factors like mobility, limited power, and weak wireless connectivity. This work presents a hybrid PSO-GA (Particle Swarm Optimization––Genetic Algorithm) based resource allocation strategy for reliable execution of jobs within a reasonable time for the computational mobile grid. Before allocating the task to the resources, the best nodes as per the fitness function are selected under the given constraints in order to meet the scheduling objectives. PSO-GA is a hybrid approach, proving to be more efficient and effective than single PSO or GA. Simulation study supports the effectiveness of the proposed approach.

Internet of Things (IoT) is an emerging technique for connecting heterogeneous technologies related to our daily needs that can affect our lives tremendously. Many architectures and applications have been proposed and implemented using IoT platform from a simple supply chain to complex life support systems. There are many obvious benefits of such networks but these systems can cause great danger to finance and life if compromised. Such issues are hindering the mass adaptation of IoT. This requires a strong architecture that can provide strong user authentication, access control as well as privacy and trust to the users of the system. The IoT network is a heterogeneous network connecting many small hardware constraint devices and also where traditional security architectures and techniques cannot be applied. Therefore, it requires a different set of specialized techniques and architecture to provide security to the IoT network. This paper focuses on the security requirements, current state of art as well as future directions in the field of IoT.

Software-Defined Networking (SDN) introducesGupta, Ashish aDidwania, Bharat newSingh, Gaurav network paradigmGupta, Hari Prabhat forMishra, Rahul separating theDutta, Tamina control plane and data plane. The control plane manages the packet flow in the data plane of the network. The anomaly detection in the context of SDN is to identify potentially harmful traffic. If an anomaly occurs because of malicious packets in SDN, inspecting the payload of packets is an effective way to recognize abnormal traffic. In this paper, we consider different bandwidths and topologies of the network for the detection of an anomaly in SDN. We also evaluate the performance of the SDN on the same network. We have implemented different tree topologies on OpenFlow controller using Mininet network emulator. We considered OpenFlow messages as a performance metric for evaluating the performance of the network with different tree topologies.

The rapid advancement in networking technologies leads to the sharing of information by millions of users at a much higher rate. Information sharing over unprotected network may lead to compromise of the sensitive and confidential information to unauthorized person. Cryptography is one of the several ways to guard sensitive and confidential information. Before sharing the information over the Internet, it must be encrypted to preserve its confidentiality. Encryption is a process of converting readable information into scrambled form so that no unauthorized person can make use of it. In this paper, the Playfair encryption technique is taken into consideration for encrypting the information to be shared. In its simplest form, the Playfair encryption technique generates a 5 × 5 key table by taking a key as input and encrypts the diagraphs of actual message. The key generation process is improved by modifying it with Fibonacci series. The Fibonacci series is used create a random key, which is passed to generate 5 × 5 key table. This key table is used to encrypt the actual message using rules defined by Playfair encryption algorithm. In this paper, the limitations of 5 × 5 key table are removed by using an 8 × 8 key table, which provides much higher level of security to the message being encrypted.

Epileptic seizures are usually investigated using EEG. The dynamic and statistical properties of brain waves of an individual with seizure are different from a normal person’s brain waves. This paper exploits these underlying properties of EEG using Lyapunov exponent and approximate entropy and proposes a novel statistical feature namely Gini’s coefficient. In this paper, we propose an automated system for detecting seizure using statistical and machine learning algorithm. The data used was publicly available with five different classes (normal to seizure). Linear discriminant analysis (LDA) was used to classify the extracted features. The proposed method gives the best accuracy of 100% in detecting seizure from the EEG.

In today’s world, the need and urge for use of cloud become more popular among the public users. The cloud provides services like freeware to the end-users. The resources that the cloud users use will be in form of shared pool. If any resources are requested by the end-users, they are provided in a shared pool. Nowadays, the resources are requested only in dynamic basis. Upon the requisition by the user, the resources are provided to them. From these shared pools of resources, the cluster head or master node is selected by using Advanced Ant Colony optimization algorithm. The status of each and individual nodes should be known to neighbor nodes and master nodes; these can be achieved by using “Heartbeat messages”. The status and movement of an individual node can be known by using these messages. The services requested by end-user and they are provided to them in very secure manner using DMZ (De-militarized zone) technique. The DMZ provides very higher security, that is, three layers of security, with different algorithms at each layer. In this paper, we address data leakage security issues and dynamic load balancing issues.

Multicast or group communication enables the distribution of the content in a one-to-many fashion. In multicast communication, the major challenges are dynamicity of group, forward and backward secrecy of the data. There are issues like single-point failure in centralized Group Key Management (GKM), false participation attack in participatory GKM, member dynamicity in Logical Key Hierarchy, etc. To address the various issues of centralized, participatory and LKH GKM; in this paper, we proposed a Key Management Scheme for Secure Group Communication. In the proposed scheme, network members share the computational load of the server and scheme achieves the forward and backward secrecy. The proposed scheme is well suitable for one-to-many mode communication.

In the lightweight domain, various ciphers and their different implementations are introduced to deal with the problem of security in resource scare environment. Eight-sided fortress (ESF) is a lightweight Feistel cipher which uses substitution–permutation network based round function with Serpent Substitution-box(S-box). This work presents a study and comparison of the various hardware architectures of ESF to combat issues of security in an extremely constrained resource environment. For the design of hardware, different techniques of S-box implementation are used. Comparison and evaluation of ESF S-box implementation techniques is done on the basis of latency, throughput, area utilization, and power consumption. It is observed that the Random Access Memory (RAM)-based S-box design gave the best results with the requirement of minimum area for its implementation. This makes it the preferred architecture for resource-limited applications.

Advancement of information technology leads toward a world with process automation to perform a task more efficiently and avail the services with ease. Banking sectors are not an exception and are moving from traditional manual banking system to an electronic entity. The basic functionality of a bank, out of many, is to deposit money into user accounts and retrieve as per account holder’s necessity. However, as time is precious, eventually account holders may not expect to spend too much time in the queue for depositing or retrieving their money. That is why the need for ATM comes into the picture to make the user’s life easier. However, it comes with some questionable possibilities for false attacks as well. Thus, a proper user authentication mechanism is needed to overcome these fraudulent activities. Our proposed method gives a new dimension to this authentication which is a hybrid version of an existing authentication system for the ATM transaction by using a Graphical pattern password along with current PIN code supplied from the bank. This Graphical password is a version, which has been invented by Google’s Android pattern unlock system. In our proposed mechanism, we combine both Graphical pattern and PIN and incorporated security to enhance reliable transactions. More specifically, the secret encryption key is generated from a PIN using the PRESENT algorithm. Finally, the ciphertext is created using digit stream from the Graphical pattern and secret encryption key. This hybrid process to detect intrusion will significantly enhance security. Our primary focus is to develop a robust and flexible user authentication system to avoid common authentication problems. The proposed approach needs no additional hardware and device dependency.

With heavy demand for cloud technology, it is important to balance the cloud load to deliver seamless Quality of Services to the different cloud users. To address such issues, a new hybridized technique Artificial Neural Network based Load Balancing (ANN-LB) is introduced to calculate an optimized Virtual Machine (VM) load in cloud systems. The Particle Swarm Optimization (PSO) technique is used to perform task scheduling. The performance of the proposed ANN-LB approach has been analyzed with the existing CM-eFCFS, Round Robin, MaxMin, and MinMin algorithms based on MakeSpan, Average Resource Utilization, and Transmission Time. Calculated values and plotted graphs illustrate that the presented work is efficient and effective for load balancing. Hybridization of ANN and iK-mean methods obtains a proper load balancing among VMs and results have been remarkable.

ThisDeshpande, Amruta S. paperPatil, Sanjaykumar L. proposes the design of a hybrid controller for a solar photovoltaic system to withdraw the maximum power. This controller combines the advantages of fuzzy controller and proportional–integral controller. The fuzzy logic control does not require the exact knowledge of the plant model while proportional– integral control reduces the offset value and gives easy architecture. Simple and efficient controller development are the main objectives behind the work. Enhanced tracking efficiency in the presence of varying environmental conditions is one of the main advantages of the controller. The controller is simulated for various irradiation and temperature conditions and the outputs are compared with fuzzy logic and traditional perturb observe controller.

Nature-inspired optimization isSingh, Shailendra Pratap the field of study forSingh, Deepak Kumar planning, simulation, and execution of problems using scientific methodologies. In this paper, a novel mutation-based modified differential evolution (DE) algorithm has been proposed. Enhance-based adaption mutation operator helps in avoiding the local optimum problem. The proposed approach is named as enhance-based adaption (EBA) in the existing mutation vector to provide more diversity for selecting effective mutant solutions. The proposed approach provides more promising solutions to guide the evolution and helps DE escaping the situation of the local optimum problem. Comparisons with other DE variants such as CPI-DE, TSDE, ToPDE, MPEDE, and JADEcr establish that the proposed Environment adaption-based operator is able to improve the performance of differential evolution algorithms.

Rough Set Theory isAgarwal, Vanita a powerfulPatil, Rajendrakumar A. Artificial Intelligence based tool used for data analysis and mining Inconsistent Information Systems. In the presence of inconsistent, incomplete, imprecise or vague data, normal statistical-based data analytic techniques lag behind. The various software used for the analysis of inconsistent data using Rough Set Theory runs on x86 kind of processors for various operating systems. Unlike the other software implementations, the main objective of undertaking this experimentation is to describe a new and standard library tool set for the computation of inconsistent data using Rough Set Theory which is completely synthesizable on FPGA. Further, the authors have also studied the effect of implemented design on Zybo FPGA for understanding the area, timing, and power efficiency criteria. A Rough Set Theory based Data Analytic Engine can be used as a potential candidate for knowledge discovery and data mining of inconsistent data in IoT applications at fog and/or edge interfaces. This paper defines the standard library tool for Rough Set Theory on FPGA.

Given a set of n demand points, the objective of the p-centre problem is to identify a subset of the demand points having p ≪ n nodes (called centres) such that the maximum distance of any demand point to its nearest centre is minimized. The problem is NP-hard and finds application in facility location. This paper presents two novel heuristics for the p-centre problem that requires O(n3) time. One of these is a deterministic heuristic that uses a minimum spanning tree-based clustering approach, and the other is a randomized heuristic that uses greedy clustering. Bounds on the computational time requirements of both heuristics are proved. The relative performance of the two heuristics is evaluated in the course of several computational experiments on a wide range of benchmark problems used in the literature for the p-centre problem.

Teaching-learning-based optimization algorithm (TLBOA) isMishra, Ruchi aSharma, Nirmala significantSharma, Harish metaheuristic algorithm. It is a proficient approach for solving multidimensional, linear, and nonlinear optimization problems. It is based on teaching-learning (TL) process that searches for a global optimum through two modules of learning: (a) teacher-phase (TP) and (b) learner-phase (LP). For avoiding the premature convergence of TLBOA, half-life teaching factor is discovered in this paper. The proposed strategy is known as half-life teaching factor based TLBO (HLTLBO) algorithm. The performance of HLTLBO is calculated over 20 benchmark functions and compared with various state-of-art algorithms namely, TLBOA, global-Best inspired biogeography-based optimization (GBBO), particle swarm optimization (PSO), and covariance matrix adaptation evolution strategy (CMA-ES). The obtained outcomes validate the authenticity of the discovered HLTLBO.

The analysis of sentiments consists in identifying and classifying the opinions, attitudes, and sentiments of people expressed in original sentences. The advancement of social media, different critics, forum discussions, blogging, and working on social networks can be divided into different ways. Users who generate huge amounts of sentiment data on the website are in large quantities in the form of tweets and status updates. The sentiment analysis of this data is useful for market analysis and product research organizations. They are increasingly using public opinions in these media for their decision-making. In this paper, we propose an approach for analyzing the sentiment or opinion in an efficient manner. For this, we have proposed a technique that focuses multilingual data analysis to classify sentiment analysis for the tweets.

Autonomous Vehicles are on rise around the globe, millions of them are already there on road with medium levels of automation but still there is a long way to go for full autonomy. One of the biggest roadblocks for autonomous vehicles to reach full autonomy is driving in urban environments. To make autonomous vehicles fully autonomous, they require the ability to communicate with other road users (pedestrian, vehicles, and other road users) and understand their intentions. Social interaction is a complex task, there are uncountable scenarios that happen on roads that require human interaction both verbal and nonverbal. Deciding whether a person standing on the sidewalk is about to cross the road, or they are just waiting near the sidewalk is a difficult task for an autonomous vehicle, and it could be a matter of life-and-death in case of a vehicle driving at very high speed. So, it is very important for self-driving cars to identify true intentions of on-road pedestrians and understand social interaction norms. In this paper, we go through some of the challenges in Pedestrian and Autonomous vehicles interaction that autonomous vehicles might face while driving in an urban environment; after that we propose a novel architecture for identifying pedestrian’s intention using pedestrian’s detection, pose estimation, and classification algorithms while discussing different methods of each.

Rough set theory isAgarwal, Vanita a powerfulPatil, Rajendrakumar A. artificialAdwani, Jyoti intelligence based tool used for data analysis and mining inconsistent information systems. In the presence of inconsistent, incomplete, imprecise, or vague data, normal statistical based data analytic techniques lag behind. This paper discusses the code profiling for rough set theory on DSP and ARM processors. This work was undertaken to understand the performance of rough set theory on existing processors for mining/analyzing inconsistent nature of IoT application at fog/edge interface.

In this paper, we present an IoT-based solution that can reduce the complexity of crowd estimation. About the human crowd estimation many techniques are in existence but nowadays more work is going on in this field because this is era of IoT and most of the organization is shifted toward IoT-based system. So in our proposed system we are using the Raspberry Pi-3 which are having quad-core processor that can be very useful and gives better result and accurate number even when the humans are very close to each other. This IoT-based model can easily be implemented in crowded areas and monitor the same. The camera module in this model also helps to differentiate between human and other bodies. As this is a mobile model, it can be easily fixed on the walls of street light and in the time of darkness or in night the camera captures clear images for process in the presence of street light. So that this model gives better result almost 70% better result in compare to exiting approaches.

Sun is a clean and renewable source of energy and produces solar energy by converting solar radiations into useful electricity through solar cells. Solar energy is obtained from solar radiations by the phenomenon of photoelectric effect. Solar panels are installed in open atmospheric conditions and undergo with environmental effects. These solar panels are subjected to various defects and faults during operation; therefore, proper condition monitoring is needed. Data loggers are used to remotely monitor the condition of solar panels, i.e., voltage, current, temperature, and other atmospheric parameters on the screen of personal computer. QR codes are normally used in commercial applications for information exchange but in this research work a novel technique based on QR code is used to view the variations of values and graphs for different parameters of solar panels via well-designed recording system through an Android app generated for particular QR code. By scanning the QR code, live variations of graphs and values in video form for solar panel data can be visualized which are not possible after time of visualization through data logger is lost. Therefore the research work presents a unique technique for visualizing recorded data in video form for solar panels as off-line whenever needed through a simple Android mobile generated QR code.

Artificial Neural Networks are being used in many applications and forecasting is one of such application where it solves the purpose like stock market predictions, sales forecasting, etc., over the past. In this paper, ANN models are used for forecasting solar power. Multilayer perceptron (MLP) neural network models have been tested for different combinations of transfer functions and net input function on different number of neurons and layers for forecasting solar power. The evaluation and implementation of models are being measured by mean square error.

This paper discusses an integrated approach toward smart parking implementation for smart cities in India using radar detection and ultrasonic technology. The aim of this research is to develop an autonomous parking system which could reduce traffic congestion and toxic emission from vehicles and also make it easier for customers to find a place to park during peak hours, hence saving their time. The autonomous parking system will have minimum amount of human interaction, and all the data will be sent directly to a cloud server wirelessly using a Wi-Fi module. In addition to that, the interface will also guide the customer to an available parking space. The paper discusses a system that can perform real-time monitoring of parking space availability by individual space and can be embedded in software. The software will raise a system alert when the number of vehicles in transit and more exceeds. The project can be implemented in various situations in places like schools, colleges, institutions, etc., and can provide an end-to-end solution for a safe parking mechanism with low-cost maintenance.

The main aim of the distributed system is to maximize the utilization of resources with minimal response time and overall execution time. For this, optimal scheduling of tasks is desirable, which is also a well-known NP-complete problem. This paper presents an algorithm to generate an optimal schedule for distributed system considering the parameters like processor load, peak load, average processor utilization, and communication cost. Further, algorithms have been experimentally compared, and the proposed algorithm has performed better than existing algorithm.

A car that can navigate by itself without being dependent on human for inputs is known as a self-driving car. There has been a great advancement in automobile industry which is bringing new technologies every day. There are various types of autonomous cars and they are divided based on their level of automation, which includes level 0 to level 5. Advanced methodologies are used to build these cars, and concepts like machine learning and computer vision play a vital role in development of these cars. The accuracy varies based on lots of factors including both internal and external factors. This paper presents survey done on various technologies used in these cars with their results and also about their current trends.

Bridge between human and artificial intelligence is the app technology, which has capability of connecting every physical object to app such that fusion of physical world and virtual world of software is enabling enterprises and all its stakeholders to deliver more efficiently which in turn promotes effective decision-making and simplicity.

Storage management is a challenging issue due to exponential growth of data and computing workloads transacted per day. Although storage virtualization plays a prominent role in addressing this problem, there exist complexities in the adoption process of virtualization technologies because the transition from a non-virtualized to a virtualized environment is not always a smooth process. Using a survey of 24 public and private institutions in Tanzania, the organizational readiness for managing data storage in virtualized servers is studied. While there is a satisfactory awareness level for most adopters of the existing virtual-based data storage technologies and disk partition techniques, the attention given to securing adopter’s practices for virtual resource allocation is inefficient for countering virtual machine attacks. Adopters are prone to starve their virtual machines as soon as their computing workloads expand. The study reveals a serious ad hoc allocation of virtual HDDs with high discrepancy between adopters from the same sector and with similar IT infrastructure, and computing level and demands. Organizations are prompted about the limitations of their current practices which would hardly bring maximum benefits of storage virtualization when expanding from small- to large-scale data workloads.

Natural resource planning isKumar, Arvind anSinha, Nishant important aspect for any society. Knowing forest cover type is one of them. Multiple statistical and machine learning approaches are already proposed in past for classification. In the current work, publicly available dataset of Forest Cover type (FC) from UCI repository was taken and classified for the forest cover type, using random forests machine learning algorithm. On ten-fold cross validation, we got accuracy of 94.6% over 70.8% accuracy of original work presented at UCI repository. The result is also compared with existing work done in past and it have been shown that random forests algorithm performed better than most of existing works.

Noisy data isNigam, Nitika theDutta, Tanima mainGupta, Hari Prabhat issue in classification. The possible sources of noise label can be insufficient availability of information or encoding/communication problems, or data entry error by experts/nonexperts, etc., which can deteriorate the model’s performance and accuracy. However, in a real-world dataset, like Flickr, the likelihood of containing the noisy label is high. Initially, few methods such as identification, correcting, and elimination of noisy data was used to enhance the performance. Various machine learning algorithms are used to diminish the noisy environment, but in the recent studies, deep learning models are resolving this issue. In this survey, a brief introduction about the solution for the noisy label is provided.

State-of-the-art database paradigm allows data to generate from different types of devices but these data have no fixed format according to RDBMS structure. Due to industrial need for business expansion, these data need to be restructured into effective database. Therefore, there is a need for migrating existing data into new database technology such as NoSQL which can efficiently handle them. NoSQL is a group of technologies, which does not follow the concept of relational database. NoSQL stores information in different types of data model such as column oriented, document oriented, or graph based. These models have their own way of storing information and schema design. Various research works have been performed in finding appropriate schema design for migration of data from relational model to NoSQL. In this paper, we have performed performance analysis of different database schema design for migration from RDBMS to NoSQL databases. The selected replication schema design provides better performance in execution time.

Automatic Speech Recognition (ASR) systemsKumar, Ankit have becomeAggarwal, R. K. more popular recently for low resource languages. India has 22 official language and more than two thousands other regional languages, the majority have low resources. The standard resources are also limited for the Hindi language. In this paper, the implementation of continuous Hindi ASR system has been done using Time Delay Neural Network (TDNN) based acoustic modeling significantly improves the performance of baseline Gaussian Mixture Model-Hidden Markov Model (GMM-HMM) based Hindi ASR system up to 11%. Further improvement of 3% and 2% have been recorded by applying i-vector adaptation, interpolated language modeling in this work.

Offensive language, hate speech, and bullying behavior is prevalent during textual communication happening online. Users usually misuse the anonymity available online social media, use this as an advantage, and engage in behavior that is not acceptable socially in actual world. Social media platforms, analytics companies, and online communities had shown much interest and involvement in this field to cope up with this problem by stopping its propagation in social media and its usage. In this paper, we will propose the work done by researchers to form effective strategies for tackling this problem of identifying offense, aggression, and hate speech in user’s textual posts, comments, microblogs, etc.

Security is one of the major challenges in the field of sensor cloud. There is a need to implement security over sensor cloud using a technique which involves fine-grained access in virtualized wireless sensor networks. The existing security model to secure the data transmission and stored data at the sensor-cloud environment uses different encryption techniques, but its effectiveness, efficiency, and performance can be further increased. Earlier approaches provided fine-grained access such as Ciphertext-Policy Attribute-Based Encryption (CP-ABE) which involves some complex computations. In this paper, we proposed a security model based on Elliptic Curve Encryption (ECC) and attributes to ensure the overall security of sensor data that guarantees the confidentiality and integrity. It also provides a fine-grained access control. The proposed approach reduces the overall computational overhead as compared to other existing approaches.

Big data is the collection of thousands of datasets from different application sources just as social media, banking, sales, marketing, etc. In every field, big data technologies are used for analyzing, preprocessing, storing, and generating new patterns for the benefits of the organization. The era of big data technology is nowadays booming [1]. Health care is one of the most important applications of big data. In health care, data exist in different forms like heart rate, blood pressure, blood test, sugar test, cholesterol, and many more. Diagnosis of diseases at an early stage is also very important in healthcare services. Cancer disease is an abnormal cell that negatively affects our body texture and regular functioning body organs. Due to cancer, the death rate is increased as it gets diagnosed at a later stage. Early diagnosis of cancer increases the survival rate of a patient. This paper focuses on the prediction model for the breast cancer diagnosis at an early stage as it increases the chances for successful treatment because of the advanced diagnostics technologies like MRI scans, ductogram, diagnostics mammogram, ultrasound, and many more. So predicting the prognosis of breast cancer increases the survival rate of women. Data mining classification algorithm like SVM, naive Bayes, k-NN, decision tree, etc. combined with analytical tool, which is a promising independent tool for handling huge datasets, is proven better in prediction of the breast cancer diagnosis.

Nowadays, huge amount of data has been generated and collected in every instance of time. So to analyze them is the toughest task to do. Data are generated and collected in a huge amount from unlike sources such as social media, business transactions, public data, etc. This greater amount of data may be structured, semi-structured, and unstructured one. The data in which analysis is to be performed these days are not only of massive amount but also varies each other by its types, at which speed it is generated and by its value and also varies by different characteristics which is termed as big data. So to examine this vast amount of data and get the relevant information from it, analysis should be done and to analyze this huge amount of data is a greater challenge these days. So to analyze these vast amount of data we need the help of several data analytics tools and methods so that it will be easier to deal with it. This survey paper talks about different tools and techniques used for big data analytics. This survey paper tries to provide a clear idea about the genesis of big data, features of big data, and different tools and techniques used to analyze these huge collections of data.

Social media generates massive volume of unstructured data by means of blogs, social networking sites, purchasing sites, etc. This huge amount of data plays a very crucial role in determining the opinions and sentiments of people toward a product, services, popularity of artists, celebrities, etc. The accurate and meaningful analysis of this online media further helps in developing product quality, making strategies, and decision for a company or a personality. The online media consists of both text and emoji. Emoji is pictorial representation of facial expression, objects, weather, animals, etc. In this paper, the web document of various worldwide famous male and female personalities has been examined to determine the amount of usage of emoji and presence of text and emoji both in expressing sentiments. This paper also presents the comparative analysis of the total usage of emoji among the most followed male and female personalities in Twitter. The major application of our paper is that it exhibits that the sentiment analysis is more accurate and complete when done for both text and emoji.

The best of brain working does not mean the absence of some sort of mental illness. It is beyond that. The ability to deal with the pressure of negative situation varies greatly from one person to another. The fast-paced lifestyle influenced humans’ communication and thinking power. This paper identifies the significant reasons behind this contemporary but devastating lifestyle. Some emotional state of mind like being loved, self-esteem, confidence, and breakups leads to adopting difficult behavior like smoking, drinking, and usage of various drugs. To examine this problem among youth, the data was collected from a reputed university’s students those who were hosteler or non-hosteler and using factor analysis, it was found that major contribution in the drinking, smoking, and drugs was social status, loneliness, and depression.

E-commerce is very much in trend in the current era for selling/purchasing products online. Hence, consumers tend to visit question answer forums to know about a product before making a purchase. The proposed work is to build a web application which would form an alternative to a traditional question answer system. Rather than focusing on a knowledge-based question answer system, the proposed work attempts to mine reviews related to the product and provides critical reviews on the product which are relevant to the question asked by the consumer. This application would be used by any user looking for supporting critical reviews related to product functionalities. Given a question answer dataset and a review dataset on a product, the similarity between the questions and the reviews is calculated and three top reviews which are most relevant to the question along with their relevance score are the output of the system. The model uses powerful similarity measures based on WordNet and Word embedding in addition to the basic similarity measures based on cosine similarity and TF-IDF. The model is evaluated in terms of how well the sentiment extracted from the output reviews of the proposed model confines with that of the answer in the question answer dataset.

Currently, social systems have caused a substantial amount of users connecting together over a couple of years, while the link gold mining is certainly a crucial analysis trail in this area. It attracted the factor of some researchers to study and understand the associations between nodes in the social network. The key concern experienced simply by authorities is normally to deal with the problem of new links forming in the network. For this purpose, all of our design and style, a new model entails Internet site survey approaches with semantics to perform hyperlink mining on data parts. To test our model, we use highlighting node degree technique to find out the future relationships between users. Our main focus in link prediction is to predict future links in the network. Our analysis normally focuses on the scoring-based methods and provides latest methodologies which are based on deep learning methods.

In earlier days, it was easier to distribute the following assets as death of a person. But with the huge amount of data with the expansion of technology, there would be a great difficulty in storing that large amount of data after the person’s demise. Nowadays, people wallow too much in social networking sites, while they do not have any idea that how much data they provoke in their daily life. To safeguard and handle the data online, they need someone who can handle the accounts. A lot of problems can arise when executors attempt to access these digital assets left behind by the deceased. Many people do not have the clear idea about digital estate which they are handling. So, to prevent those unauthorized access to digital estate, one should know all the laws and the privacy rights related to the digital assets.

Among several types of biometric systems, ear recognition is a bustling research area. Due to the minimal cooperation of the user, this biometric trait proves to be a good application in security and surveillance. Over the period of last two decades, various contributions have been reported with robust techniques and approaches in ear biometrics. This paper provides an overview of various ear recognition and detection techniques using 2D ear images, among which some are automated and some are not. Also, a comparative review of the available databases for research purposes is provided. A comparative vision of ear detection and recognition is presented in this paper in chronological order.

Sleep spindles are one of the unique rhythmic activities observed in sleep electroencephalogram (EEG). Detecting sleep spindles visually by sleep spindles is a difficult task as high skills and efforts are required. In this study, a methodology for detecting sleep spindles automatically has been proposed using band-pass filtering. Time domain features (energy and entropy) are used for classification. The extracted features have been used as inputs to Linear, Quadratic, and Mahalanobis classifier for spindle detection. Results show that the proposed method yields best results when using a Mahalanobis Classifier. The accuracy, sensitivity, and specificity recorded are 91.11%, 84.86%, and 89.73%, respectively. The sensitivity obtained in this study is more than most of the work done in sleep spindles detection using the same dataset.

Computer Aided Diagnosis (CAD) systems forKamble, Bhawana automatic detection ofSahu, Satya Prakash pulmonary diseases andDoriya, Rajesh lung cancer mainly depend on the segmentation of different pulmonary components like right and left lung lobes, airways, vessels, and nodules from the medical imaging modalities like CTs, MRIs, etc. Lung segmentation and nodule segmentation are the important steps to detect any lung related abnormalities. It requires many image processing operations to be performed on the medical images. Computed Tomography (CT) imaging is the most preferred modal because of its popularity, ease of use, and capability of showing different anatomical structures of thorax region. This review paper includes a study of various state of the art techniques explaining the methods applied on CT scans to find the ROIs along with their segmentation accuracies parameters in terms of similarity coefficient, mean error, and overlap ratio.

Incorrect information or documentation of the patient details and a wrong direction provided by any personnel to the patient can largely influence the lab reports or any other confidential data. Once the medical image is verified as secure and safe by the authorized access, one can review the medical image and the patient information. This paper focuses on providing a helping tool to ensure higher accuracy as well as authentication of the medical data and the related information. It proposes an authentication scheme to address the issue of security and privacy preservation of medical details. This technique involves generating a secure identification pattern from the fusion of the patient’s information with the features of cone-beam CT images obtained through wavelet analysis without any tampering with the medical details. The patient’s information can be successfully restored at the authorized receiver’s end. This authentication scheme can be helpful and implemented in telecare medical information systems.

ThereVashistha, Piyush areSingh, Juginder Pal different methods orKhan, Mohd Aamir techniques used for identifying violence from video, such as hitting some object, kicking, fighting, and punching someone but still there is a big challenge for us to identify violence. However, some of the earlier mechanism generally extract descriptors around the spatiotemporal interesting points (STIP) or extract statistic features but there is limited effectiveness in detecting video-based violence. Therefore, the objective is to develop a better violence identification system that identifies the violence and triggers an alarm so that prompt assistance will be provided. This paper helps researchers who wish to study violent activity recognition and gather different insights on the main challenges and issues to solve in this emerging field.

TheDutta, Tanima compressed domain video watermarking hasSoni, Aishwarya been less expensive sinceGupta, Hari Prabhat complete decoding and then re-encoding is not required for watermark embedding. Handling synchronization error due to embedding is a challenging task, especially, due to motion compensation. In this paper, we propose an embedding scheme that can minimize desynchronization in the watermark extraction process at the decoder for embedding in compressed videos with better visual quality. Experimental results show the effectiveness of the proposed scheme.

Deep learning has becomeBagi, Randheer one ofDutta, Tanima theGupta, Hari Prabhat most preferred solution for many complex problems. It shows outstanding performance in the field of computer vision to perform tasks like, image classification, object detection, and image generation. Recently, many research efforts are focused on changing the deep learning architecture for widespread application domain. In this paper, we present a comprehensive survey on the various issues and challenges faced by deep learning techniques. Furthermore, we analyze different deep learning architectures to provide the solution for the computer vision tasks along with their importance.

Robust reversible watermarking algorithms appliedDutta, Tanima to protectBagi, Randheer medical imaging fromGupta, Hari Prabhat misdiagnosis for any slight distortion. It suited for tasks like copyright protection while preserving the visual quality of watermarked images. It also shows robustness against transmission errors. We propose a novel robust reversible watermarking technique for grayscale medical images. We investigate the high embedding capacity watermarking technique also keeping low distortion in the watermarked images. We demonstrate the accuracy of the reversibility even in error-prone transmission channels. Our simulations show that the proposed technique effectively retains the perceptual quality.

The medical images are often corrupted by various noises and blurriness. In particular, the noises presented in ultrasound images may lead to an inaccurate diagnosis of smaller kidney stones and affect its treatment. This paper proposes an improved technique for detection of kidney stone from the ultrasound images of kidney. The ultrasound kidney images are preprocessed to remove labels and change from RGB to Gray images. Further, image contrast is enhanced by adjusting the image intensity. To remove the noises, median filtering is used. The filtered image is taken as input for morphological segmentation process; initially applied dilation and then seed region growing algorithm is used to segment the renal calculi from ultrasound image of kidney. The region parameters are extracted from the segmented region. Finally, area of each renal calculi is calculated. The various performance evaluation GLCM features such as entropy, contrast, angular second moment and correlation are used to judge the quality of output images. The confusion matrix is also prepared to analyze the sensitivity, specificity, and accuracy of the final system. The overall accuracy of classification system is around 90%. The proposed technique may help medical professionals in easy detection of kidney stones and benefit the patients.

Fingerprint sweat pores as Level 3 features have the capability to improve the accuracy of the fingerprint recognition process. Extraction of pores is the foremost step in the designing of fingerprint high-level features based applications such as migration control at the borders, identification of fake fingerprints, etc. Filtering based approach is one of the prominent techniques for pores extraction. All the available filtering based methods are categorized into non-adaptive and adaptive techniques. This paper presents an extended description of both the techniques. Some practical concerns while implementing the methods have also been highlighted. Experimental results have been carried out on the high resolution database of 500 dpi fingerprint images. Performance has been measured and compared on the basis of true detection rate (RT) and false detection rate (RF). Simulation results show that the adaptive filtering techniques achieve better RT and RF.

3D video game assets, small props to larger mesh in a video game, make a huge difference in performance of video games. Non-optimized game assets may not support its presentation in wide range of hardware. Many people want to play video game but cannot play because of their lower end hardware setup. So, what is the point of developing a video game if vast amount of people cannot enjoy it? In this paper, we have introduced a simple but an effective methodology to optimize 3D mesh to support in wide range of hardware, which can satisfy the need of gamers. We have shown that reducing poly count and removing unnecessary parts from meshes contribute significantly to optimal performance.

The work presented in this paper proposes a robust and optimized algorithm to be used in image watermarking. It makes use of cuckoo search algorithm (CSA) for optimization. A binary watermark is embedded within the host images in transform domain using a fitness function. The locations used for inserting the watermark bits are selected using CSA. The fitness function is a linear summation of similarity correlation coefficients SIM(W, Wʹ) obtained for four different operations performed on the watermarked image. The simulation reports indicate that the PSNR values are high enough. As a result, the watermarked images have high visual quality. The values of similarity correlation coefficient show that the scheme presented in this paper is also robust against the said operations. It can be concluded that this CSA-based scheme shows good result and also reports better results than other similar works.

Over the past few years, fingerprints have been considered the most sensitive and crucial identification basis for low enforcement agencies. In crime scene and forensics, recording of latent fingerprints from uneven and noisy surface is a difficult task and conventional algorithm fails in most of the times. A robust orientation field estimation algorithm is the need of the time to recognize the poor quality latent. To overcome the limitations of conventional algorithm, various techniques have been proposed in the last decade. In this paper, a comparative study has been done of state-of-the-art techniques with their advancements and limitations. Our proposal aims at effectively minimizing the difficulties faced to separate ridges and segmentation of latent images reducing search time and computational complexity while optimizing the system retrieval performance.