Abstract: Frauds are known to be dynamic and have no patterns, hence they are not easy to identify. Fraudsters use recent technological advancements to their advantage. They somehow bypass security checks, leading to the loss of millions of dollars. Analyzing and detecting unusual activities using data mining techniques is one way of tracing fraudulent transactions. transactions. This paper aims to benchmark multiple machine learning methods such as k-nearest neighbor (KNN), random forest and support vector machines (SVM), while the deep learning methods such as autoencoders, convolutional neural networks (CNN), restricted boltzmann machine (RBM) and deep belief networks (DBN). The datasets which will be used are the European (EU) Australian and German dataset. The Area Under the ROC Curve (AUC), Matthews Correlation Coefficient (MCC) and Cost of failure are the 3-evaluation metrics that would be used.

Abstract: The real-time sign language recognition system is developed for recognising the gestures of Indian Sign Language (ISL). Generally, sign languages consist of hand gestures and facial expressions. For recognising the signs, the Regions of Interest (ROI) are identified and tracked using the skin segmentation feature of OpenCV. The training and prediction of hand gestures are performed by applying fuzzy c-means clustering machine learning algorithm. The gesture recognition has many applications such as gesture controlled robots and automated homes, game control, Human-Computer Interaction (HCI) and sign language interpretation. The proposed system is used to recognize the real-time signs. Hence it is very much useful for hearing and speech impaired people to communicate with normal people.

Abstract: Having diseases is quite natural in crops due to changing climatic and environmental conditions. Diseases affect the growth and produce of the crops and often difficult to control. To ensure good quality and high production, it is necessary to have accurate disease diagnosis and control actions to prevent them in time. Grape which is widely grown crop in India and it may be affected by different types of diseases on leaf, stem and fruit. Leaf diseases which are the early symptoms caused due to fungi, bacteria and virus. So, there is a need to have an automatic system that can be used to detect the type of diseases and to take appropriate actions. We have proposed an automatic system for detecting the diseases in the grape vines using image processing and machine learning technique. The system segments the leaf (Region of Interest) from the background image using grab cut segmentation method. From the segmented leaf part the diseased region is fruther segmented based on two different methods such as global thresholding and using semi-supervised technique. The features are extracted from the segmented diseased part and it has been classified as healthy, rot, esca, and leaf blight using different machine learning techniques such as Support Vector Machine (SVM), adaboost and Random Forest tree. Using SVM we have obtained a better testing accuracy of 93%.

Abstract: This paper examines five key questions surrounding computer generated art. Driven by the recent public auction of a work of “AI Art” we selectively summarise many decades of research and commentary around topics of autonomy, authenticity, authorship and intention in computer generated art, and use this research to answer contemporary questions often asked about art made by computers that concern these topics. We additionally reflect on whether current techniques in deep learning and Generative Adversarial Networks significantly change the answers provided by many decades of prior research.

Abstract: Credit scoring focuses on the development of empirical models to support the financial decision‐making processes of financial institutions and credit industries. It makes use of applicants' historical data and statistical or machine learning techniques to assess the risk associated with an applicant. However, the historical data may consist of redundant and noisy features that affect the performance of credit scoring models. The main focus of this paper is to develop a hybrid model, combining feature selection and a multilayer ensemble classifier framework, to improve the predictive performance of credit scoring. The proposed hybrid credit scoring model is modeled in three phases. The initial phase constitutes preprocessing and assigns ranks and weights to classifiers. In the next phase, the ensemble feature selection approach is applied to the preprocessed dataset. Finally, in the last phase, the dataset with the selected features is used in a multilayer ensemble classifier framework. In addition, a classifier placement algorithm based on the Choquet integral value is designed, as the classifier placement affects the predictive performance of the ensemble framework. The proposed hybrid credit scoring model is validated on real‐world credit scoring datasets, namely, Australian, Japanese, German‐categorical, and German‐numerical datasets.

Abstract: Real time object detection is one of the challenging task as it need faster computation power in identifying the object at that time. However the data generated by any real time system are unlabelled data which often need large set of labeled data for effective training purpose. This paper proposed a faster detection method for real time object detection based on convolution neural network model called as Single Shot Multi-Box Detection(SSD).This work eliminates the feature resampling stage and combined all calculated results as a single component. Still there is a need of a light weight network model for the places which lacks in computational power like mobile devices( eg: laptop, mobile phones, etc). Thus a light weight network model which use depth-wise separable convolution called MobileNet is used in this proposed work. Experimental result reveal that use of MobileNet along with SSD model increase the accuracy level in identifying the real time household objects.

Abstract: In order to increase the automatic quality control level in the textile industry, depending on the big data collected by the Internet of things of the textile factories, this paper proposes a novel visual saliency–based defect detection algorithm, which has the capability of automatically detecting defect in both nonpatterned and patterned fabrics. The algorithm employs the histogram features extracted from the saliency maps to detect the fabric defects. The algorithm involves three main steps: (1) saliency map generation to highlight the defective regions and suppress the defect‐free regions, (2) saliency histogram features extraction and selection to obtain the feature vectors that can effectively discriminate between the defective and defect‐free fabric images, and (3) fabric defect detection using a two‐class support vector machine classifier that has been trained using sets of feature vectors extracted from defective and defect‐free fabric samples. Experimental results show that our method yields accurate detections, outperforming other state‐of‐the‐art algorithms.

Abstract: Stock market is a very volatile in-deterministic system with vast number of factors influencing the direction of trend on varying scales and multiple layers. Efficient Market Hypothesis (EMH) states that the market is unbeatable. This makes predicting the uptrend or downtrend a very challenging task. This research aims to combine multiple existing techniques into a much more robust prediction model which can handle various scenarios in which investment can be beneficial. Existing techniques like sentiment analysis or neural network techniques can be too narrow in their approach and can lead to erroneous outcomes for varying scenarios. By combing both techniques, this prediction model can provide more accurate and flexible recommendations. Embedding technical indicators will guide the investor to minimize the risk and reap better returns.

Abstract: Numerous parts of southern India have recently encountered severe damage to lives and properties due to floods. Floods are one among the most destructive natural hazard and recovering to normal life takes ample time. During hazards, various technologies are in use for speeding up relief operations and to minimize the amount of damage, one such being the use of drones. Many algorithms are in need for automatic analysis of remote sensing and aerial images. Nowadays, drones are being used for taking images from varied heights similar to aerial images, as they have cameras with exceptional features and effective sensors. This paper proposes a hybrid approach to classify whether a region in an aerial image is flood affected or not. A combination of Support Vector Machine(SVM) and k-means clustering proved capable of detecting flooded areas with good accuracy, classifying about 92% of flooded images correctly. Performance analysis is done by changing various kernel functions in SVM. The results show that there is a decrease in the prediction and training time when quadratic SVM is used.

Abstract: Today, everyone has their own personal devices that connects to the internet. Every user tries to get the information that they require through internet. Most of the information is in the form of a database. A user who wants to access a database but having limited or no knowledge of database languages faces a challenging and difficult situation. Hence, there is a need for a system that enables the users to access the information in the database. This paper aims to develop such a system using NLP by giving structured natural language question as input and receiving SQL query as the output, to access the related information from the railways reservation database with ease. The steps involved in this process are tokenization, lemmatization, parts of speech tagging, parsing and mapping. The dataset used for the proposed system has a set of 2880 structured natural language queries on train fare and seats available. We have achieved 98.89 per cent accuracy. The paper would give an overall view of the usage of Natural Language Processing (NLP) and use of regular expressions to map the query in English language to SQL.

Abstract: Ensemble learning is a popular classification method where many individual simple learners contribute to a final prediction. Constructing an ensemble of learners has been shown to often improve prediction accuracy over a single learner. Bagging and boosting are the most common ensemble methods, each with distinct advantages. While boosting methods are typically very tunable with numerous parameters, to date, the type of flexibility this allows has been missing for general bagging ensembles. In this paper, we propose a new tunable weighted bagged ensemble methodology, resulting in a very flexible method for classification. We explore the impact tunable weighting has on the votes of each learner in an ensemble and compare the results with pure bagging and the best known bagged ensemble method, namely, the random forest.

Abstract: Recommendation systems are being enforced to offer personalized set of services to the users. They are basically build to produce recommendations or suggestions (like restaurants, places…) that comply with user’s concern and that can be applied to multiple fields. To enhance the quality and service of Recommendation systems and to resolve any issues related to it, various effective techniques linked to data management can be made use of. The current paper proposes a machine learning algorithms to resolve the issue of personalized Restaurant selection relying upon tripadvisor.com search data. The facilities provided by the hotel along with user’s comments are being utilized. The NLP - Natural Language Processing is imbibed for examining and tagging all the previous user’s comments (whether positive or negative) for every hotel, thereafter computing the overall % of the comments and storing the output. In the process of Restaurant recommendation, first the user chooses the hotel’s features according to his interest and centered on this, the corresponding hotels are fetched and the user comments are examined to identify the hotel with the highest ranking. Eventually, the highest rated hotel is being recommended to the user by the restaurant recommended system. The proposed sentimental score measure NLP algorithm is used for finding the aspect and sentiments of the user comments. Natural language processing (NLP) is one of the machines learning technique to analyze, understand, and derive meaning from human language in a smart and useful way. The evaluation results reveal that the proposed NLP algorithm improves the performance when compared to existing algorithms. The focus of the research work is to offer list of recommended restaurants that is more precise and accessible. The conclusion and results reveal that the suggested approach yields high accuracy.

Abstract: In recent times, the advancements in Brain-Computer Interface has not only been instrumental in achieving its fundamental purpose of aiding disabled people, but also in creating novel applications like playing games without physical controls or operating home appliances merely by the power of your brain. The electrical activity generated in the brain is measured by an EEG device after which the collected raw data undergoes through various steps, namely: Signal acquisition, Data Preprocessing, Feature Extraction, and Classification. This paper helps the reader in understanding the different algorithms and methods used in each of these processes. A detailed survey of various applications of BCI using different feature extraction and classification techniques is done. Finally, we have compiled all the current issues which hinder the efficiency of BCI systems.

Abstract: Lateral epicondylitis, also known as 'tennis elbow', is a degenerative rather than inflammatory tendinopathy, causing chronic recalcitrant pain in elbow joints. Although most patients with lateral epicondylitis resolve spontaneously or with standard conservative management, few refractory lateral epicondylitis are candidates for alternative non-operative and operative modalities. Other than standard conservative treatments including rest, analgesics, non-steroidal anti-inflammatory medications, orthosis and physical therapies, nonoperative treatments encompass interventional therapies include different types of injections, such as corticosteroid, lidocaine, autologous blood, platelet-rich plasma, and botulinum toxin, which are available for both short-term and long-term outcomes in pain resolution and functional improvement. In addition, newly emerging biologic enhancement products such as bone marrow aspirate concentrate and autologous tenocyte injectates are also under clinical use and investigations. Despite all non-operative therapeutic trials, persistent debilitating pain in patients with lateral epicondylitis for more than 6 months are candidates for surgical treatment, which include open, percutaneous, and arthroscopic approaches. This review addresses the current updates on emerging non-operative injection therapies as well as arthroscopic intervention in lateral epicondylitis.

Abstract: The application of machine learning algorithms on text data is challenging in several ways, the greatest being the presence of sparse, high dimensional feature set. Feature selection methods are effective in reducing the dimensionality of the data and helps in improving the computational efficiency and the performance of the learned model. Recently, evolutionary computation (EC) methods have shown success in solving the feature selection problem. However, due to the requirement of a large number of evaluations, EC based feature selection methods on text data are computationally expensive. This paper examines the different evaluation classifiers used for EC based wrapper feature selection methods. A two-stage feature selection method is applied to twitter data for sentiment classification. In the first stage, a filter feature selection method based on Information Gain (IG) is applied. During the second stage, a comparison is made between 4 different EC feature selection methods, Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Cuckoo Search (CS) and Firefly Search, with different classifiers as subset evaluators. LibLinear, K Nearest neighbours (KNN) and Naive Bayes (NB) are the classifiers used for wrapper feature subset evaluation. Also, the time required for evaluating the feature subset for the chosen classifiers is computed. Finally, the effect of the application of this combined feature selection approach is evaluated using six different learners. Results demonstrate that LibLinear is computationally efficient and achieves the best performance.

Abstract: Recently, in designing a good microstrip patch antenna for different wireless applications, researchers have been focusing on the best techniques used in improving gain, directivity, and bandwidth of an antenna. This paper, we have presented state-of-the-art and potential demand of the low profile microstrip antenna. Further, we have overviewed various techniques which have been employed to improve the performance particularly gain and bandwidth of the microstrip antenna. Various techniques in improving gain and bandwidth have been investigated, which include the use of feeding, parasitic patch, defective ground structure, air gap, slot, shorting pin, metamaterial, and a dielectric substrate. The different performance characteristics have been illustrated using the impendence matching network, miniaturization techniques, and patch geometry modification. This paper provides a review showing the performance characteristics using various methods, and their performance characteristics were compared in terms of percentage improvement of gain and bandwidth so that the best option can be selected for various applications.

Abstract: This paper focuses on the classification of dermoscopic images to identify the type of Skin lesion whether it is benign or malignant. Dermoscopic images provide deep insight for the analysis of any type of skin lesion. Initially, a custom Convolutional Neural Network (CNN) model is developed to classify the images for lesion identification. This model is trained across different train-test split and 30% split of train data is found to produce better accuracy. To further improve the classification accuracy a Batch Normalized Convolutional Neural Network (BN-CNN) is proposed. The proposed solution consists of 6 layers of convolutional blocks with batch normalization followed by a fully connected layer that performs binary classification. The custom CNN model is similar to the proposed model with the absence of Batch normalization and presence of Dropout at Fully connected layer. Experimental results for the proposed model provided better accuracy of 89.30%. Final work includes analysis of the proposed model to identify the best tuning parameters.

Abstract: The sector manufacturing rebellion (Manufacturing generation) has been enticing consideration from entirel over the biosphere over the past few years. Digital technology progression has positioned companies in the face of a projected change in pari gms and management methods. Manufacturing 4.0 is altering the manufacturing sector. The novel scientific tendencies, the arrival of manufacturing 4.o and artificial Intelligence (AI) and machine-learning make it imperative to replicate on manufacturing practices and their efficiency with appreciate to the new technological framework. AI and Machine Learning in industry 4.0 paves way for immense revolution for manufacturing companies with different industry opportunities. Industry 4.0 embrace IoT, AI, cyber-physical systems, Machine learning and cognitive computing which enables the startups for smart factories. The production processes in smart factory are allied to interfaces, machines and modules communicate with one another where large amount of information can be collected to enhance the manufacturing process. Artificial intelligence and machine learning in manufacturing is focused on industrial automation in the pursuit of efficacy.

Abstract: Path planning and trajectory tacking are the fundamental task in mobile robotic science, and they enable the robot to navigate autonomously. In this work, the path planning task is carried out using three bio-inspired optimization algorithms, including PSO, ABC and FA. The duty of the algorithms is to determine a collision-free path through fixed obstacles in the working environment. The maximum speed of the robot is applied to the optimization problem as a constraint. In order to evaluate the performance of the algorithms, four workspaces with different obstacle layout are simulated in MATLAB, and the quality of path planning task is analyzed statistically and numerically, considering four different criteria, including, convergency quality, convergency time, path length and success rate. In the next step, a control model is designed to track the path curve determined by the path planning algorithms. A PID-based control structure is simulated in MATLAB Simulink and the controller was able to track the pre-determined traj ectories with proper approximation. The controller is applied on a dynamic model of a two-wheeled mobile robot offered by [1]. In order to validate the control inputs it is necessary to apply them on a real platform. The experimental study is implemented on a two-wheeled mobile robot which is designed and built based on the authors' previous paper [2] in various enverioment and obstacles. The result shows control inputs were applied to the real robot and the robot was able to imitate the applied path curve, and find its way toward the target point without colliding obstacles in real and simulation task.

Abstract: Floods are a natural calamity which leads the dry land to be submerged by water due to a resurgence of a waterbody capacity which goes well beyond its natural limits leading to an overflow. Floods are most commonly caused by excessive precipitation and runoffs which lead the adjoining land areas to be submerged by water which causes huge loss to human lives and infrastructure, which includes damaging buildings, bridges, power supply network and crippling the transportation and bringing economic hardships on the people. Over the years, multiple measures have been taken to predetermine flood warnings which have been implemented using sensor technology and active monitoring of the parameters. This had led to the creation of a wide number of data-sets which can be employed for future purposes and with the availability of data analytics techniques heralded by the resurgence of Machine Learning and the concept of Intelligent Machines, the datasets can be directly employed to allow algorithms to "learn" directly from the collected data and based upon this, create a predetermined equation as a model to help predict future outcomes. In the proposed method, we propose a Flood Detection mechanism using the Gradient Boost Algorithm which will be used to classify the data sets and perform regression on it to produce the best outcomes from the datasets we will use to train it, to create a weak prediction model based on a Decision Tree. The outcome can henceforth be used to display it to the concerned authorities who can employ preemptive actions to tackle the threat. This approach is developed to be better suited in such ends providing predictions with high accuracy and additionally employs various other technologies like Remote Sensing and Sensor Technology to develop accurate datasets required to train the model.

Abstract: The lighting condition of the environment are uncontrolled, so the segmentation of a leaf from the background is considered as a complex task. Here we propose a system which can identify the plant species based on the input leaf sample. An improved vegetation index, ExG-ExR is used to obtain more vegetative information from the images. The reason here is, it fixes a built-in zero threshold and hence there is no need to use otsu or any threshold value selected by the user. Inspite of the existence of more vegetative information in ExG with otsu method, our ExG-ExR index works well irrespective of the lighting background. Therefore, the ExG-ExR index identifies a binary plant region of interest. The original color pixel of the binary image serves as the mask which isolates leaves as sub-images. The plant species are classified by the color and texture features on each extracted leaf using Logistic Regression classifier with the accuracy of 93.3%.

Abstract: The proliferation of data from diverse sources makes humans insufficient in utilizing the knowledge properly at some instance. To quickly have an overview of abundant information, Text Summarization (TS) comes into play. TS will effectively extract the candidate sentences from the source and represent the saliency of whole knowledge. Over the decades Text Summarization techniques have been transformed by the usage of linguistics to advanced machine learning models, this study explores summarization approaches along with their recent state-of-art models in single and multi-document summarization. This survey is intended to make an extensive study from features representation to sentence selection and summary generation using machine learning, recent graph and evolutionary based methods. The overall investigation will help the researchers to effectively handle large quantities of data in building effective Natural Language Processing applications. Eventually, this study draws popular abstractive mechanisms and observations that would be helpful for the intended research.

Abstract: Internet of Things (IoT) is a rapidly expanding technology that has recently got significant recognition in the field of studies. In IoT networks, huge traffic in network causes congestion at nodes that influences the quality of routing metrics the overall performance of the network. Therefore in this paper, congestion aware algorithm using fuzzy logic (CAUF) has been proposed to avoid congestion by selecting the best parent in a tree structured IoT network to find the optimal routing path. It models the problem of parent selection into multi attribute decision making (MADM) based problem using fuzzy weighted sum model. CAUF has been implemented and simulated on cooja simulator and a comparison of performance is carried out with queue utilization based RPL (QU-RPL) and optimization based hybrid congestion alleviation (OHCA) algorithms. Simulation results indicate that proposed has 15% more throughput and 4.5% packets less dropped over OHCA and QU-RPL algorithms.

Abstract: According to statistics, most road accidents take place due to lack of response time to instant traffic events. With the self-driving cars, this problem can be addressed by implementing automated systems to detect these traffic events. To design such recognition system in self-driving automated cars, it is important to monitor and manoeuvre through real-time traffic events. This involves correctly identifying the traffic signs that can be faced by an automated vehicle, classifying them, and responding to them. In this paper, an attempt is made to design such system, by applying image recognition to capture traffic signs, classify them correctly using Convolutional Neural Network, and respond to it in real-time through an Arduino controlled autonomous car. To study the performance of this road sign recognition system, various experiments were conducted using Belgium Traffic Signs dataset and an accuracy of 83.7% has been achieved by this approach.

Abstract: Cloud Computing is one of the most widely emergent areas in today’s IT sector. It helps a number of people to use various services through their own devices with the help of the internet. It provides an environment which has low cost, easy to use and also consumes less power with virtualization. VM’s are required to be managed by a number of task scheduling algorithms to make sure that the less amount of energy consumption takes place. This paper deals with the importance of green cloud computing and how it can provide alternatives to the IT sector in terms of the energy consumption, load on data centers, VM average load and the task distribution. All these experiments have been performed by using a green cloud simulator in which three different algorithms are being used such as DENS, Round Robin and Green Schedulers.

Abstract: Recently, many computational intelligence algorithms have been proposed to address software remodularization problem. Unfortunately, it has been observed that the performance of optimizers degrades with the optimization problem containing more than three objectives. In this paper, we propose a many‐objective discrete harmony search (MaDHS) to address the software remodularization problem having more than three objectives. The basic idea of MaDHS is that it uses the quality indicator Iϵ + and external archive to rank and store the nondominated solutions. Along with MaDHS, five remodularization objectives, ie, low coupling, high cohesion, low modification degree, quality of class distribution, and low package instability have also been adapted to improve the package structure of existing object‐oriented software systems. To improve the accuracy of modularization solution, the coupling and cohesion objectives are formulated in terms of various dimensions of direct coupling relationships. To test the supremacy of the proposed approach, it is evaluated over eight real‐world object‐oriented software systems. Simulation results show that the proposed approach outperforms the other existing approaches in terms of couplings, cohesion, modularization quality, modularization merit factor, rate per refactoring of achieved improvement, and external developers view.

Abstract: Various studies have demonstrated that convolutional neural networks (CNNs) can be directly applied to different levels of text embedding, such as character‐, word‐, or document‐levels. However, the effectiveness of different embeddings is limited in the reported result and there is a lack of clear guidance on some aspects of their use, including choosing the proper level of embedding and switching word semantics from one domain to another when appropriate. In this paper, we propose a new architecture of CNN based on multiple representations for text classification, by constructing multiple planes so that more information can be dumped into the networks, such as different parts of text obtained through named entity recognizer or part‐of‐speech tagging tools, different levels of text embedding, or contextual sentences. Various large‐scale, domain‐specific datasets are used to validate the proposed architecture. Tasks analyzed include ontology document classification, biomedical event categorization, and sentiment analysis, showing that multi‐representational CNNs, which learns to focus attention to specific representations of text, can obtain further gains in performance over state‐of‐the‐art deep neural network models.

Abstract: With the applications of Industrial Internet of Things (IIoT) and service based on manufacturing, the 4th industrial revolution which is termed as Industry 4.0, is believe to be proposed in the upcoming years, where the factories will not work as used to work. In this paper, we present a smart factory framework which combine industrial network, cloud and various control terminal with several work pieces such as conveyor, product and machine. This work has used ‘Controllino mega’ which is a controller to control the entire production as it has a cloud connectivity feature. It is an IoT Arduino compatible open source PLC device for industrial uses in automation and for controlling and monitoring. This will allow separate part of the production line to communicate to make the entire manufacturing process easier to control and monitor in real time. In this work, a lot of new features and functionalities have been incorporated, which will change the way, a factory used to look and be. A wireless communication using smart sensors has been used to demonstrate that the devices such as sensors and actuators work intelligently rather than depending on the main control hub. These devices will have its own microcontroller through which it comes to action. Another feature which will help the factories to continue the production rather than getting downtime issue has also been addressed successfully by adding "the notification feature".

Abstract: currently, traditional libraries, despite using information systems, are facing problems managing books. In this era of technology, traditional libraries need to innovate and adapt to the smart society. Technologies such as Internet-of-things (IoT) could be used to capture data in real time. Making use of business process re-engineering, a number of processes could be refined. Using a smart library management system that integrates IoT and automating a traditional library’s core processes will propel the library towards a next-generation library. Users will be able to communicate smartly with IoT devices to perform relevant tasks. The paper therefore describes a traditional library system on a university campus and highlights challenges faced by such a system. Additionally, the paper explains how processes in the traditional library could be converted into a smart library using BPMN notation. The smart library aims to efficiently perform library management and solve traditional library problems.