The skin illness forms the most hazardous type of malignancy when the skin cells explode. In recent years, biomedical imaging and analysis have grown increasingly intriguing, helpful, and promising as a means of addressing the potential challenges caused by malignant melanoma skin tissues that might form on epidermal. The human visual system has perceptual difficulties in the conversion zone intermediatory of the clinical and the lesion segmentation. In this paper, a review following Preferred Reporting Items for Systematic Review and Meta-analysis on segmentation and classification techniques used for skin lesion detection is presented. Firstly, an easy-to-read summary of skin problems, image-acquiring techniques in dermatology, and a list of numerous publicly accessible skin lesion datasets is provided. The paper focuses on the segmentation techniques, feature extraction and selection techniques and classification techniques used in detection of skin lesions. Also, this paper goes through the difficulties encountered in the region and offer potential paths for further study. This article’s main goal is to present a conceptual and methodological evaluation of recent studies on skin disease.

The speed at which internet growth changed the advertising field enabled businesses worldwide to develop online advertising as their main marketing tactic. Businesses use different digital advertising models, such as display ads and SEO, together with email marketing and mobile advertising, to connect with larger audiences across various platforms. The advertising sector has experienced sharp expansion in mobile advertising, which enables application developers to generate revenue by showing native ads within their app interfaces. Mobile ad fraud presents a major existential threat to the entire online advertising network. Click fraud is a common type of attack that happens in cost-per-click (CPC) advertising. In the mobile advertising sector, attackers create fake clicks to fool the system and make money falsely. Mobile ad fraud includes all illicit activities targeting marketing technology that mislead publishers, advertisers, and vendors, resulting in monetary damage within the mobile advertising sector. The Real-Time Bidding (RTB) system uses a large part of mobile advertising budgets for instant auctions that link advertisers straight to their target customers. RTB is vulnerable to fraudulent activity, which complicates the problems currently associated with click fraud. A thorough analysis of click fraud in online advertising is provided in this paper, with an emphasis on mobile ad fraud in RTB scenarios. In order to create a more transparent and safe digital advertising ecosystem, we examine the different kinds, methods, and financial effects of fraudulent activity. We also go over different fraud detection strategies that are intended to recognize and avoid fraudulent clicks.

Click-through rate prediction (CTR) remains a significant research topic, which is crucial for online advertising and recommendation systems. Despite numerous advantages in CTR models, performance improvements have been limited, which lack contextual learning abilities and are prone to over fitting challenges. Therefore, to overcome such limitations, this research introduces a Modified Frequency Wise Hessian Eigen Value Regularization Enabled Bidirectional Long Short-Term Memory framework (MFHESTM), which reveals a strong positive correlation between the feature frequency and top Hessian eigenvalue. The proposed MFHESTM framework leverages the benefits of fractional calculus into the stochastic gradient descent, which allows for more nuanced updates by considering the memory effect of fractional derivatives. When compared with the prevailing methods, the empirical results exhibit that the proposed model shows superior prediction performance in terms of accuracy of 98.28

Personalized video summarization involves creation of a compact yet representative version (video summary) of an input video based on individual user preferences. There exists some research gaps in existing works concerning multi-modal system, domain knowledge, effective mapping of video content and user preference. This paper covers the existing research gaps for personalized video summarization and better user entertainment experience. This paper presents a Multi-Modal Multi-module approach for dynamic and personalized video summarization of Cricket sport videos. The Multi-module architecture exploits multiple modalities of the video to select representative content according to user preferences. The proposed approach includes dynamic video segmentation strategy based on Cricket domain knowledge, and key segment selection strategy based on Umpire Detection-Umpire Pose Recognition and Score Board Optical Character Recognition. The proposed approach is quantitatively and qualitatively evaluated to observe the performance of the models and to analyse the quality of generated dynamic and personalized video summary. The performance evaluation (both quantitative and qualitative) reveals the exceptional results and promises to present this work as a standard towards video segmentation, dynamic and personalized video summarization and sports entertainment.

In this paper, we present the Multi-CNN approach for dynamic and personalized Video Summarization. The proposed approach is grounded on Cricket Sport domain knowledge to learn complex and domain features. The personalized video summary is based on individual user preferences and is dynamic (dynamic summary). The considerations of individual user preference, domain knowledge, dynamic content, and Cricket sport make the work one of its kind. The proposed Multi-CNN architecture entails two levels, CNN Level-1 and CNN Level-2. We present domain activity-based video segmentation through CNN Level-1 to generate dynamic video segments. The video segments are then forwarded to CNN Level-2, which includes a stacked organization of two models (Umpire detection and umpire pose recognition) to label the video segments. The individual user preference is matched with labeled video segments for key segment identification. We also propose two novel summary evaluation metrics based on individual user reactions. The results indicate the promising performance of the proposed system and provide significant insights for dynamic and personalized video summarization.

Artificial Intelligence is quickly emerging as a technological solution for the agriculture industry to surmount its classical challenges. Artificial Intelligence is facilitating farmers to refine their products and alleviate unfavourable impacts due to the environment. The central concern of this paper is predictive analytics to develop a machine learning model to identify and predict crop yield based on multiple environmental factors. In this paper, a hybrid learner 'RaNN' is proposed that combines the feature sampling and majority voting technique of Random Forest incombination with the multilayer Feedforward Neural Network to predict the crop yield. Research has also ascertained the essential features responsible for accurate yield prediction. The proposed model works for rice yield prediction, one of the chief grains of India. The region chosen for the work is Punjab, which is among the largest producer states of India for rice. The dataset consists of 15 attributes comprising the weather and agriculture data collected from the Indian Meteorological Department Pune, and Punjab Environment Information System (ENVIS) Center, Government of India. The study has also made a comparative assessment of 'RaNN' with machine learning methods like Multiple Linear Regression, Random Forest, Decision Tree, Boosting Regression, Support Vector Machine Regression, Ensemble Learner, and Artificial Neural Network. Our model RaNN has listed a better prediction accuracy with minimal error among the other techniques providing a 98% correlation between the actual and the predicted yield.

Personalized video summarization entails generation of short and compact video summary based exclusively on user preferences. The perception of individual user preferences and selection of relevant and salient video content requires domain knowledge understanding. This paper presents a stacked Convolutional Neural Network (CNN) approach by embedding action-based and rule-based domain knowledge of Cricket sport. The proposed approach relies on a stacked organization of Umpire detection and Umpire pose recognition modules, followed by domain rules to select video content matching the user preferences. This content selection strategy facilitates the generation of dynamic and personalized video summary grounded on domain knowledge. This paper also proposes two novel summary evaluation metrics based on user reactions, i.e., User Rating Score and Composite Summary Score (CS-Score). The proposed approach performs exceptionally well with deep action-based features and rule-based features for better perception and understanding of user preferences and synchronization-selection of relevant video content. The results indicate promising performance of both models and present a standard and benchmark platform for sports, personalized and dynamic video summarization.

The performance of the cloud-based systems is directly associated with the resource utilization. The maximum resource utilization indicates the high performance of cloud computing. Further, effective task scheduling is the primary factor that significantly impacts resource utilization. Various researchers have proposed multiple algorithms to achieve optimum utilization, but a high rejection rate of tasks occurred due to conflict among similar tasks during backfilling. A hybrid approach is proposed for selecting the backfilled tasks using a multicriteria decision-making approach (MCDM), namely Distance-based approach with a traditional task scheduling algorithm. The parameters considered to find the suitable task for backfilling are arrival time, start time, deadline time, duration, size, and the number of virtual machines. The proposed algorithm is implemented on different datasets and compared with other state-of-the-art algorithms. The result shows a significant improvement in virtual machine utilization and task acceptance/rejection rate.

ABSTRACT The paper has focussed on the global landcover for the identification of cropland areas. Population growth and rapid industrialization are somehow disturbing the agricultural lands and eventually the food production needed for human survival. Appropriate agricultural land monitoring requires proper management of land resources. The paper has proposed a method for cropland mapping by semantic segmentation of landcover to identify the cropland boundaries and estimate the cropland areas using machine learning techniques. The process has initially applied various filters to identify the features responsible for detecting the land boundaries through the edge detection process. The images are masked or annotated to produce the ground truth for the label identification of croplands, rivers, buildings, and backgrounds. The selected features are transferred to a machine learning model for the semantic segmentation process. The methodology has applied Random Forest, which has compared to two other techniques, Support Vector Machine and Multilayer perceptron, for the semantic segmentation process. Our dataset is composed of satellite images collected from the QGIS application. The paper has derived the conclusion that Random forest has given the best result for segmenting the image into different regions with 99% training accuracy and 90% test accuracy. The results are cross-validated by computing the Mean IoU and kappa coefficient that shows 93% and 69% score value respectively for Random Forest, found maximum among all. The paper has also calculated the area covered under the different segmented regions. Overall, Random Forest has produced promising results for semantic segmentation of landcover for cropland mapping.

Video Summarization is a video compression/compaction technique to create a shorter yet informative version of original video. Video summarization has offered solutions to plenty of media, user and engineering applications. Though sports video summarization has been an active research topic for some time; there still exists a void for multi-modal, dynamic, generic and domain knowledge based approach for Cricket Sport video summarization. This paper presents a multi-modal video summarization approach to summarize Cricket sport videos. This work captures the domain knowledge acquired from multi-modal (audio-visual) cues. A dual neural network architecture pipeline is proposed to dynamically segment and dynamically summarize Cricket videos for generic target audience. The former Neural Network is grounded on Cricket bowling activity (visual feature) for dynamic video segmentation of Cricket videos. The segments are then forwarded to the latter Neural Network for identification of key segments. The key segment detection module relies on Audio analysis of Cricket video stream to identify exciting, content representative and informative segments as per Cricket domain. Experimental analysis on two novel proposed benchmark datasets, i.e. DPCS (Delivery Play Cricket Sport) image dataset and EXINP (Excited Interval Normal Play) Cricket Dataset (audio based) shows promising results. The results indicate that the proposed multi-modal approach generates exciting, content representative, informative, generic and dynamic summary incorporating domain knowledge of the sport.

Plant disease detection and classification are required to identify the early symptoms of disease in plants and crops to avoid their mitigation to entire croplands. Deep learning is rapidly getting involved in plant disease identification because of the large amount of image data which is otherwise difficult to learn. The paper has performed the tomato leaf disease classification based on convolutional neural network along with transfer learning where the data is collected from the PlantVillage dataset. The dataset comprises 10 classes containing nearly 16,000 leaf images. The data is divided into 70%, 20%, and 10% ratios into training, test, and validation set, respectively. The paper has shown a CNN model developed from scratch for learning, whose results have been compared with four different transfer learning models: DenseNet121, ResNet50, Inception-V3, and VGG-16. The models are evaluated using accuracy and cross-entropy loss. VGG-16 and CNN model developed from scratch has shown promising results on the given dataset with 90% and 83% validation accuracy, respectively, on the test set.

The WWW contains huge amount of information from different areas. This information may be present virtually in the form of web pages, media, articles (research journals / magazine), blogs etc. A major portion of the information is present in web databases that can be retrieved by raising queries at the interface offered by the specific database and is thus called the Hidden Web. An important issue is to efficiently retrieve and provide access to this enormous amount of information through crawling. In this paper, we present the architecture of a parallel crawler for the Hidden Web that avoids download overlaps by following a domain-specific approach. The experimental results further show that the proposed parallel Hidden web crawler (PSHWC), not only effectively but also efficiently extracts and download the contents in the Hidden web databases

With the advancement of location-acquisition and mobile computing devices in today’s life, the trajectory data obtained from Global Positioning System (GPS) is becoming large which makes route prediction a very challenging task. The route prediction can be done in two modes-offline route prediction and online route prediction in terms of acquiring performance and accuracy. There are many different techniques of route prediction which mainly consists of three consecutive tasks- i) Route abstraction ii) Frequent route pattern mining iii) Route Prediction applied in the context of different application domains. Route prediction plays an important role in many location-based applications including vehicular ad-hoc networks, route navigation, traffic control and congestion estimation, Place recommendation and many more. Trajectory data have different useful information hiding into it like background data and contextual information. So, Privacy is a big concern while sharing this personal trajectory data with the server predicting the best optimal route. Many techniques have been proposed by researchers for ensuring privacy of user data along with maintaining availability and usability of the data. Rather than doing all pre-processing computations on a single machine, many algorithms have been proposed to make system scalable. In this paper, a comparative analysis of various existing techniques of route prediction along with privacy preservation and scalability issues is discussed with pros and cons. The main aim of this paper is to give insights of the different algorithms of the route prediction with the directions of future research.

With the advancement in the advertising industry, Real-Time Bidding (RTB) is become the most promising framework for the ad-space auction. The main challenge in RTB is to handle the highly dynamic nature of the data and do computations with this exponentially growing data, so that the involving agencies get high Return of Investment (ROI). In the RTB process, advertisers buy ad-space (impressions) published by the publisher (website owner). When user visits a webpage, an ad-space is created. If ad-space is not reserved for any specific advertiser then publisher ad-server auctions it on the open ad-market. Publisher ad-server connects to the Server-Side Platform (SSP), SSP sends the ad-request with auction information (like web cookies, unique visitors, page views per visit, sessions etc.) to Ad-Exchange (AdX), which further send the ad-request to DSPs. DSPs bid on the behalf of advertisers for the ad-space and submit it to the AdX. The winner, who bids more than other competitors, will win the auction and display her ad to the publisher website. In RTB, there are two important bid price models, from the view of advertiser. One is to estimate the utility and other is auction cost. The former deals with the probability of user responses towards displayed ad (i.e. determination of Conversion rate (CVR) and Click through Rate (CTR)) and later deals with the amount paid by the advertisers after winning auction. In this paper an attempt has been made to provide insights about the bidding strategies from demand and supply side platforms. The motive is to give detail about real time bidding strategies with the future work guidance.

In recent years, my country's smart grid has developed rapidly and is safe and stable operation of the power grid system; it is related to the healthy development of the national economy and stable life of the people. Most of the current vibration monitoring systems use wired ICP piezoelectric acceleration sensors to collect the vibration of the transformer tank wall, on-site live monitoring, and safety, but there are many inconveniences and problems. The author designed a vibration-sensing element based on the latest Micro-electromechanical systems (MEMS), a vibration sensor for wireless communication with ZigBee, applied in the online monitoring of a 110 kV three-phase power transformer in operation, and adopted wired and wireless sensor modes, respectively, to monitor and compare waveforms. The wired sensor used in the test is an ICP sensor, and its sensitivity is 500 mV/g. It can be used in a multi-vibration measuring point wireless network to monitor the surface vibration of the power transformer tank. In the waveform graph collected by wireless sensor, it can be observed that the waveform presents a periodic law, and it can be observed from the spectrogram that the energy is concentrated on the frequency multiplier of 50 Hz, which conforms to the vibration law of the transformer.

Employment of various sensors used in IoT can help create a sustainable urban life. Rapid advancements in IoT have made human existence smarter. Smartness may be associated to office, home, networks, energy consumption, agriculture, education, retail, and even healthcare. Smart health management addresses its populations, such as tracking routine activities, obesity, nutrition intake, heart rate, glucose level, oxygen level, body temperature, or even stress level monitoring. Stress is a condition which is being faced by people irrespective of their age, gender, or profession. But its identification at an early stage can help in preventing the consequences. This work presents nine machine learning techniques for identifying stress using the SWELL dataset. Hence, automated classifiers were utilised to predict working circumstances and stress-related mental states and were compared for accuracy for three stress conditions (no stress, interruption, and time pressure).

Wearable and movable lodged health monitoring gadgets, micro-sensors, human system locating gadgets, and other gadgets started to appear as low-power communication mechanisms and microelectronics mechanisms grew in popularity. More people are interested in energy capture technology, which turns the energy created by motion technology into electric energy. To understand the difference in motor skill levels, a nonlinear feature-oriented method was proposed. A bi-stable magnetic-coupled piezoelectric cantilever was designed to detect the horizontal difference of motion technology. The horizontal difference was increased by the acceleration generated by the oscillation of the leg and the impression betwixt the leg and the ground during the movement. Based on the Hamiltonian principle and motion technique signal, a nonlinear dynamic model for energy capture in motion technique is established. According to the shaking features of human leg motion, a moveable nonlinear shaking energy-gaining system was the layout, which realized the dynamic characteristics of straight, nonlinear, mono-stable, and bi-stable. The experimental outcome shows that nonlinearity can effectively detect the difference of motion techniques. The experimental results of different human movement states confirm the benefits of the uncertain bi-stable human power capture mechanism and the effectiveness of the electromechanical combining design established. The nonlinear mono-stable beam moves in the same way as the straight mono-stable beam in the assessment, but owing to its higher stiffness, its frequency concentration range (13.85 Hz) is moved to the right compared to the linear mono-stable beam, and the displacement of the cantilever beam is reduced. If the velocity is 8 km/h, the mean energy of the bi-stable method extends to the utmost value of 23.2 μW. It is proved that the nonlinear method can understand the difference in the level of motion technique effectively.