Search

In multimedia content, text play major role for transmission and it is crucial to protect text data while transmitting over network. This can be achieved by text encryption algorithm. However, the method relies on many resources that increase the processing unit time, memory, and battery power. There are so many different techniques should be used to protect confidential text from unauthorized access. This paper summarizes on the existing works which used different techniques for text encryption by using deep learning algorithms. Combination of these approaches helps to analyze different algorithms for different text, accelerates the processing time while maintaining accurate preservation and retrieval. [ABSTRACT FROM AUTHOR]

This paper analyzed the feature selection for reducing the number of input variables when developing a predictive model. Boruta Algorithm is using in this paper as a wrapper around a Random Forest classification algorithm. Boruta algorithm is one of the algorithms used to determine the significant variables (feature selection) in a classification model in the machine learning approach, as supervised learning. Our results show that on the German Credit Data from the UCI Machine Learning with 20 variables, feature selection using Boruta Algorithm with Python Programming obtains 4 significant features. [ABSTRACT FROM AUTHOR]

Various machine learning algorithms are developed with an aim to create precise and trustworthy models and extract knowledge from data sources. Deep expertise in the field of machine learning is required for the challenging task of choosing the right algorithms for a specific dataset. There is no single algorithm that outperforms all others across all applications and different datasets. The difficulty of choosing an appropriate algorithm for a specific task in specific domain is related to the properties of the dataset. Properties of the dataset are measured through meta-features. Meta-features describe task and can provide explanation how one machine learning approach outperforms other algorithms on a given dataset. Learning about the effectiveness of learning algorithms, or meta-learning was developed to deal with this issue. Focus is required because previous research papers have not successfully identified meta-features in particular domains. In this research, we have evaluated various meta-feature characterization methodologies and have concentrated on basic meta-features. Business domain data is in the focus of this paper. We computed basic (general) meta-features and illustrated several use cases for their applications. [ABSTRACT FROM AUTHOR]

The sine and cosine algorithm has become a widely researched swarm optimization method in recent years due to its simplicity and effectiveness. Based on the advantages, the study in this paper delves deeper into the key parameters that influence the performance of the algorithm, and has implemented modifications such as integrating the reverse learning algorithm and adding elite opposition solution to create the modified Sine and Cosine Algorithm (the modified SCA). Furthermore, by combining the fuzzy k-nearest neighbor method with the modified SCA, the study simulates numeric datasets with two or multiple classes, and analyzes the results. The accuracy rate (ACC) achieved by the modified SCA FKNN in this paper is compared to other models, with data comparison results and tables presented for each. The modified SCA FKNN proposed in this paper has obvious advantages on accuracy rate(ACC). [ABSTRACT FROM AUTHOR]

Twitter is known to be one of the familiar social networking platform these days, among many others, with a lot of user engagement. This microblogging site encourages social interactions, allowing users to stay up to date on the latest news and events and share them with others in real time. Tweets are limited to 280 characters and is allowed to include links to related websites and tools. With a platform having such wide reach, it is prone to be targeted negatively and spams are one way to do it. Spammers use this platform to display malicious content that is inappropriate and harmful to users worldwide. Machine Learning uses various approaches that can be used to detect spam and overcome it. However, with the advent of recent technologies it has been observed that the properties of tweets vary overtime making it difficult to detect spam leading to the "Twitter Spam Drift" problem. This paper reviews the papers published since 2018 that have focused on the spam drift problem and gives a comparative analysis of the different algorithms that are utilized on the various data sets to tackle such a problem. [ABSTRACT FROM AUTHOR]

Current research in artificial intelligence has been primarily focused on machine learning as applied to specific computer and engineering problems. For example, facial recognition or malware detection. This aspect of artificial intelligence has seen numerous advances and continues to advance. Research regarding synthetic consciousness has not progressed on par with research into other sub domains of artificial intelligence. While there has been speculation regarding the eventuality of some level of consciousness being developed via artificial intelligence, no specific research modalities have been adequately developed. This paper focuses on two synthetic consciousness issues. The first issue is to provide a basis for developing synthetic consciousness. The second issue is outlining specific research modalities that have a significant probability of achieving synthetic consciousness. The premise of this paper is to outline how to most effectively create algorithmic systems that have a high probability of leading to the emergent property of consciousness. [ABSTRACT FROM AUTHOR]

In their highest grade, brain tumors are the most widespread and dangerous diseases with a very short life span. Therefore, early automatic brain tumor detection is required to lower the fatality rate. Due to this, MRI is a commonly used imaging technology for diagnosis; however, it is practically impossible to do manual segmentation of the volume of data generated by MRI promptly. This paper is intended to analyze the suitable tools, techniques and language for automatic detection of Brain Tumor. From the nature of the problem, it is quite evident that it requires high precision of accuracy in detecting such a deadly disease in a very short period and if possible, in real-time, for a large number of datasets will be required not only to train the algorithm but also for its testing. Spark is an open-source platform to deal with lots of data. Spark's API, PySpark is coupled with Python language to enable the developers to develop a python script for the Spark processing engine. Deep learning algorithms are the most useful and appropriate for such types of tasks where a large amount of data is involved and requires high precision training and testing of the models and the algorithms. In this paper, we have discussed the selection of the tools, techniques and language for the implementation of the model for detecting the brain tumor. Tools like Google Colab and PySpark have been explored in this paper. [ABSTRACT FROM AUTHOR]

The main objective of this paper is to improve the accuracy for automatic classification of meal waste from innovative trash data with the help of image processing. There are 2572 images for the classification of meal waste were used for this paper. The images are labeled as "Cardboard", "Plastic", "Paper", "Metal", "Glass", "Trash" and there are 20 number of images have been used for RF classifier taken as first set of machine learning algorithm and is compared with SVM algorithm taken as second set of machine learning algorithm With a g-power value of 80%, the revolutionary garbage data images, a threshold of 0.05%, a confidence interval of 95%, and a standard deviation, these photographs were gathered from various web sources. When compared to the SVM method, which had an accuracy of 61.45%, the proposed system's accuracy was enhanced to 84.81%, with a significant value of 0.001 (p 0.05) with a 95% confidence interval. This study found the meal waste from trash using the RF is significantly better than SVM algorithm. [ABSTRACT FROM AUTHOR]

Medical events are often infrequent, thus becomes hard to predict. In this paper, we focus on predictor that forecasts whether a medical event would occur in the next year, and analyzes the impact of event's frequency and data size via predictor's performance. In the experiment, we made 1572 predictors for medical events using Medical Insurance Claims (MICs) data from 800,000 participants and 205.8 million claims over 8 years. The result revealed that (a) forecasting error will be increased when predicting low-frequency events, and (b) increasing the number of training dataset reduces errors. This result suggests that increasing data size is a key to solve low frequency problems. However, we still need additional methods to cope with sparse and imbalanced data. [ABSTRACT FROM AUTHOR]

The number of digital solutions based on machine learning has increased in recent years. In many industrial sectors, they try to enhance automation in manual or repetitive tasks or provide decision support for complex problems. Data plays an essential role in the selection and implementation of ML algorithms, as it determines the quality of the training and the results. As data drive ML models, selecting the correct data with the suitable ML algorithm for a given use case is crucial but challenging. This paper reviews the application of machine learning in the embodiment design phase addressing the challenge. The work focuses on ML applications in conventional product development and non-conventional additive manufacturing processes. Based on the literature review, the required knowledge to implement the ML algorithms has been derived and presented in a systematic approach. This work highlights the importance of an initial analysis of the existing knowledge in the engineering and additive manufacturing processes in order to implement the proper ML algorithms. [ABSTRACT FROM AUTHOR]

The quick development of technology not only makes life easier but also raises several security concerns, so cyber security has become very important and vital research area, rather an inevitable part of computer system. Still, various research being done on the development of effective intrusion detection system (IDS). An IDS is one of the suspicious network activities. An IDS is used to identify many types of malicious actions that can undermine a computer system's protection and confidence. Recently, ensemble algorithms are applied in IDS in order to identify and classify the security threats. In this paper author intends to do a brief review of various Ensemble learning Algorithms in ML, which are most frequently used in IDS for several applications; with specific interest in dataset and metric. This work provides broad study and investigation on current literature, the gap for improving and creating efficient IDS can be determined. [ABSTRACT FROM AUTHOR]

Hands are the sensors and actuators for many design tasks. While several tools exist to capture human interaction and pose, many are expensive and require intrusive measurement devices to be placed on participants and often takes them out of the natural working environment. This paper reports a novel workflow that combines computer vision, several Machine Learning algorithms, and geometric transformations to provide a low-cost non-intrusive means of spatially tracking hands. A ±3mm position accuracy was attained across a series of 3-dimensional follow the path studies. [ABSTRACT FROM AUTHOR]

Kidney is one of the significant organs in the human body which performs filtering out blood, balances fluid, removes the waste, maintains the level of electrolytes and hormone levels. So, any disorder or dysfunction in kidney needs to be detected on time in order to preserve life. Segmentation on kidney tumor in medical field is a critical task and many conventional methods have been employed for early prediction of kidney abnormalities but with limitations such as high cost, extended time for computation and analysis with huge amount of data. Due to all such problems, the prediction rate and accuracy has reduced considerably. In order to overcome the challenges, Artificial Intelligence (AI) technology has penetrated into the field of medicine particularly in the renal department. The evolution of AI in kidney therapies improve the process of diagnosis through several Machine Learning (ML) and Deep Learning (DL) algorithms. It has the capability of improving and influencing on the status with its capacity of learning from the massive data and apply them accordingly to differentiate on the circumstances. The storage of larger data and segmentation with AI assistance are highly helpful for the analysis of occurrence of the disease. AI algorithms have predicted the severity of tumor stages with effective accuracies. Hence, this paper provides a critical review of different AI based algorithms being used in the kidney tumor prognostication. Its numerous benefits in field of segmentation have been researched from the existing works and provides an insight on the contribution of AI in the kidney disease prediction. [ABSTRACT FROM AUTHOR]

Activity recognition offers a wide range of applications in various industrial processes and healthcare. This work proposes an approach to collect data from a spherical coordinate system using smartphones, then extract the highly efficient features using advanced preprocessing. The paper also proposes an algorithm to recognize activity using various ensemble machine-learning approaches based on extracted features. These approaches are evaluated under various combinations of features to analyze the accuracy, sensitivity, specificity, and training time. Experimental results reveal that weighted KNN performs best among all models by achieving 96.2% accuracy with 12 features. On the other hand, Bagged tree ensemble classifiers perform better than subspace KNN ensemble classifiers with an accuracy of 95.3% using 12 features. [ABSTRACT FROM AUTHOR]

Multi-core processing can decrease the cost of many Machine Learning (ML) tasks by executing them in parallel using multiple cores of a processor. Modern computers are equipped with processors that contain multiple cores that can be leveraged to decrease the execution time of many ML tasks by multiple folds. Especially ensemble of ML algorithms like Random Forest (RF) can take advantage of the multi-core processing ability for improving their performance. In this paper five models using RF algorithm using Gaussian_Quantiles (GQ), Load_Wine (LW), Load_Iris (LI), Load_Breast_Cancer (BC) and Load_Digits (LD) datasets are developed and evaluated using k-fold cross-validation respectively. In general, the execution time of RF algorithm decreases as the number of cores increases. [ABSTRACT FROM AUTHOR]

The community#x2019;s views and inputs have always been the main and most beneficial source for varied range of enterprises. With more widespread community media, it provides a spectacular study and assessment of many fields in which companies used to have faith in peculiar, exhausting and inaccurate ways. This form of analysis is subclass of #x2019;sentence analysis#x2019; area. Sentiment analysis is a broad term that refers to the process of effectively classifying user-generated content into specific polarities. To perform sentiment identification and analysis, a variety of tools and techniques are available, includes supervised techniques for machine-learning that classify the target group after training in data. Hybrid instruments are a blend of machine learning and lexicon-based algorithms, which classify according to annotated dictionary. We employed the SVM with Weka for analyzing sentiments in this paper. Two pre-categorized datasets of tweets are utilized. The performance of SVM is analyzed with the help of analytical metrics. [ABSTRACT FROM AUTHOR]

Defining rules for finding all the uncountable patterns of events of interest, in complex and critical applications, is infeasible, if not impossible. Anomaly Detection (AD) methods are techniques well-suited for tackling this category of problem, namely, finding previously unknown deviant data due to their ability for identifying behavior that significantly went astray from an expected pattern. Isolation Forest (IF) is a state-of-theart technique in the unsupervised Anomaly Detection area, mainly because of efficacy and low computational cost. However, unsupervised Anomaly Detection many times suffers from high false-positive rate and high false-negative rate. Semi-supervised techniques, e.g. Hybrid Isolation Forest (HIF)l can significantly improve the unsupervised algorithms with low human effort, and can also aggregate semantic knowledge into the models. HIF, however, can only aggregate known anomalies in a single class. This paper proposes a Hybrid Isolation Forest-based model capable of aggregating known anomalies in many classes, without loss of performance. Our proposed multiple anomaly classes semi-supervised model showed better performance in some known datasets, as well as showed the ability to significantly improve the underlying unsupervised algorithm. [ABSTRACT FROM AUTHOR]

Recently, there has been increased research on the topic of artificially intelligent programs having the capability of developing advanced systems that are presently used by governments and organizations to analyze highly complex structures across sectors in ways not possible with conventional information technology. While some AI is subject to rigid testing and ethical reviews, other applications raise questions as to what governance structures are in place to control the risks to humanity and long term harmful economic and social consequences. This paper raises awareness about how governments and private industry face an unprecedented challenge in managing these complex systems that include regulators, markets, and special interests that all play a role in influencing the development of AI in different contexts without a full appreciation of the impact of AI on human rights and other consequences. The research focuses on three primary areas: (1) How AI technologies have evolved; (2) What are the major ethical and human rights issues evolving from the use of AI in the public and business environment; and (3) how can we improve our frameworks and governance structure for AI regulation. Through empirical evidence this paper explores the legal implications including the rights and duties of the government and private industry in protecting against unlawful intrusions into people's lives, while at the same time advancing recommendations for accountability frameworks and regulations essential to ensure safety and security in advancing artificially intelligent systems. [ABSTRACT FROM AUTHOR]

To minimize makespan for scheduling independent tasks in cloud computing, an improved estimation of distribution algorithm (IEDA) is proposed to tackle the investigated problem in this paper. Considering that the problem is concerned with multi-dimensional discrete problems, an improved population-based incremental learning (PBIL) algorithm is applied, which the parameter for each component is independent with other components in PBIL. In order to improve the performance of PBIL, on the one hand, the integer encoding scheme is used and the method of probability calculation of PBIL is improved by using the task average processing time; on the other hand, an effective adaptive learning rate function that related to the number of iterations is constructed to trade off the exploration and exploitation of IEDA. In addition, both enhanced Max-Min and Min-Min algorithms are properly introduced to form two initial individuals. In the proposed IEDA, an improved genetic algorithm (IGA) is applied to generate partial initial population by evolving two initial individuals and the rest of initial individuals are generated at random. Finally, the sampling process is divided into two parts including sampling by probabilistic model and IGA respectively. The experiment results show that the proposed IEDA not only gets better solution, but also has faster convergence speed. [ABSTRACT FROM AUTHOR]

Software defect estimation is an approach aiming to make an estimation of which algorithms can predict defects present in the software. This will help in improving the software quality which indeed can satisfy customers. The process is done during the development life cycle so at an early phase which reduces a lot of time and saves cost. Our model is going to be efficient because it can improve the performances and also give better results by estimation results. The main aim of this paper is to evaluate the capability of machine learning algorithms and find the best category while comparing seven machine learning algorithms within the context of NASA datasets obtained from the public PROMISE repository. The best algorithm is the one with highest accuracy rates. [ABSTRACT FROM AUTHOR]

This paper is conducted to develop a system that can aid universities in tracking students' academic progress by accurately determining dropout rates using machine learning techniques. This has been done by collecting and organizing data into machine readable forms, using predictive methods that determines significant patterns and trends, constructing different models involving machine learning techniques: Random Forest, Logistic Regression, and Neural Network, for forecasting and predictions, and finally interpreting the statistical results of the model. After the evaluation, it has been concluded that the Random Forest achieved the highest accuracy of 95.00%, precision of 97.78%, and F1 score of 96.70%. Despite these results, the researchers recommend developing a more interesting and engaging prototype GUI and utilizing a more balanced and diverse dataset that will enable higher accuracy and deeper understanding of the results. [ABSTRACT FROM AUTHOR]

Accurate identification of the QRS complex is critical to analyse heart rate variability (HRV), which is linked to various adverse outcomes in premature infants. Reliable and accurate extraction of HRV characteristics at a large scale in the neonatal context remains a challenge. In this paper, we investigate the capabilities of 15 state-of-the-art QRS complex detection implementations using two real-world preterm neonatal datasets. As an attempt to improve the accuracy and reliability, we introduce a weighted ensemble-based method as an alternative. Obtained results indicate the superiority of the proposed method over the state of the art on both datasets with an F1-score of 0.966 (95% CI 0.962-0.97) and 0.893 (95% CI 0.892-0.894). This motivates the deployment of ensemble-based methods for any HRV-based analysis to ensure robust and accurate QRS complex detection. [ABSTRACT FROM AUTHOR]

Chronic kidney disease is a rising health issue that affects millions of people worldwide. Early detection and characterization of this disease is essential for effective management and control. This disease is associated with several serious health risks, such as cardiovascular disease, increased risk of stroke, and end-stage renal disease, which can be effectively prevented by early detection and treatment. Medical scientists rely on machine learning algorithms to diagnose the disease accurately at its outset. Recently, adding value to healthcare is being accomplished through the integration of machine learning algorithms into mobile health solution. Considering this, this paper proposes a predictive model of three machine learning classifiers, including Support Vector Machine, Decision Tree, and Multilayer Perceptron for chronic kidney disease prediction. The performance of the model was assessed using confusion matrix and executed in popular machine learning software tools such as WEKA and Rapid Minor. The study found that support vector machine yielded the highest accuracy rate of 98% in predicting chronic kidney disease in WEKA among other standard classifiers by using 10-fold cross validation. In addition, the proposed prediction model has been compared with existing models in terms of accuracy, sensitivity, and specificity. The experimental results indicate that the proposed predictive model shows promising results. These findings could integrate with the development of mobile health solution and other innovative approaches to prevent and treat this debilitating condition. [ABSTRACT FROM AUTHOR]

Objective The main objective of this research is the usage and evaluation of Machine Learning algorithms for automatic text codification in statistical production process. Prior Work This is an evolving area of research due to rapid changes in technology as well as the new data ecosystem. The paper will build on previous research done on text classifications techniques. Approach In this paper Machine Learning algorithms will be used and evaluated for text codification. Natural Language Processing and classification algorithms will be implemented in Python. Results Machine Learning is powerful in the process of automation and modernization of statistical production lifecycle. Machine Learning algorithms performance is different for text classification. Data pre-processing and balance on the training data set are important to achieve good results. Implications This study shows that machine learning can be used in automating part of the statistical codification process. The results of this paper will serve the work of Albanian administration and more specifically statistical production. Value This research is a contribution to the usage of Machine Learning for the modernization of the codification process. It will serve as an initial work towards improving the timeliness and lowering statistical production costs. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic has caused millions of infections and deaths worldwide in an ongoing pandemic. With the passage of time, several variants of this virus have surfaced. Machine learning methods and algorithms have been very useful in understanding the virus and its implications so far. In this paper, we have studied a set of novelty detection algorithms and applied it to the problem of detecting COVID-19 variants. Our results show accuracies of 79.64% and 82.43% on the B.1.1.7 and B.1.351 variants respectively on ProtVec unaligned COVID-19 spike protein sequences using One Class SVM with fine-tuned parameters. We believe that a system for automated and timely detection of variants will help countries formulate mitigation measures and study remedies in terms of medicines and vaccines that can protect against the new variants. [ABSTRACT FROM AUTHOR]

Football is a famous sport played around the world that involves various actions from defending to attacking the ball to score goals. Various rules like half-side, No hands, Throw-ins Direct and Indirect kicks, etc. are followed. In a game, players might get yellow-cards and red-cards which serves as warning and elimination from that particular game. There are-many factors like shots, possession, pass accuracy, yellow and red cards, etc. that decides which team wins in a match. In the data that we worked on,the Rating is calculated from the parameters which are mentioned. Machine learning regression algorithms have been applied to predict the Rating feature. The goal of this paper is to know which regression algorithm performs better to predict the Rating variable. Evaluation metrics like Root mean square error (RMSE), Mean absolute error (MAE), and R square (R2) are used to identify the best algorithm among the applied algorithms. [ABSTRACT FROM AUTHOR]

To predict the accuracy percentage of Short Message Services (SMS) spam detection using machine learning classifiers. Two ensemble learning algorithms named random forest algorithm and naive bayes are applied to data. The algorithms have been implemented and tested over a dataset which consists of 5574 records. Ensemble learning methods combined several models trained with a given learning algorithm to improve accuracy. After performing the experiment as result shows mean accuracy of 89.75 % by using random forest and compared naive bayes algorithm mean accuracy is 88.05% for SMS spam detection. There is a statistically significant difference in accuracy for two algorithms is p<0.05 by performing independent samples t-tests. This paper is intended to implement Innovative Classification for prediction of SMS spam detection. The comparison results shows that the naive bayes algorithm has appeared to be better performance than Random Forest Algorithm. [ABSTRACT FROM AUTHOR]

The aim of this paper is mining serendipitous drug usage to validate and generate drug repositioning hypotheses from social media. Materials and Methods: Two machine learning algorithms svm with sample size=12 and adaboost algorithm with sample size=12. Results: The support vector machine algorithm has shown more accuracy of (96. 66%) in reducing the false positive rates when compared with Adaboost algorithm accuracy(84.6%). The pre-test was calculated with a g-power value = 80% and threshold 0. 05% confidence interval of 95% mean and standard deviation by using the G-power tool. t is found that the svm algorithm has more accuracy percentage when compared with the Adaboost algorithm. [ABSTRACT FROM AUTHOR]

In recent years, the internet of things is prevalent and widely used. The new problem with IoT is security, which needs to be considered carefully because of the technology heterogeneity. These threats can affect IoT performance; therefore, it is necessary for effective monitoring. This paper examines several machine learning methods in intrusion detection systems that possibly run on IoT. Random Forests and Decision Tree are employed in this study for performance comparison. The experimental results show that the Random Forest and Decision tree algorithms application produces good performance with a faster response time and possible running on IoT. [ABSTRACT FROM AUTHOR]

Quantum computers are here, and the search for applications and use of these allow us to overcome the limits that today's hardware information processing gives us is constantly going on. Quantum machine learning is one of the many emerging fields that use quantum computers to process information. In this paper, we present a method and a set of experiments where we see the potential and capacity of the Noisy intermediate-scale quantum hardware for the execution of different models having as the basis in some of them the quantum algorithm corresponding to the Quantum Fourier Transform. With this, we demonstrate the effectiveness of how this algorithm can enhance the performance of quantum computations in quantum machine learning models in near-term applications. We used the systems offered by IBM Quantum and the cross-platform Python library for quantum differentiable programming Pennylane by Xanadu Quantum Technologies Inc. [ABSTRACT FROM AUTHOR]

Training models in machine learning often requires big computational capacities, even more when it involves Big Data. The Patient Rule Induction Method (PRIM), introduced as a bump hunting algorithm and a subgroup discovery method, already consumes a lot of computing resources when searching for the rules. In this computational research area, several methods have been developed to handle the problem of computing a big dataset: parallel learning, distributed learning and federated learning. In this paper, we investigate the literature review of these three methods to define each one and the differences between them, and then we determine which of these methods is best suited for the implementation of PRIM in big dataset for out futur works in Big Data. [ABSTRACT FROM AUTHOR]

One of the research emergence as per studied problem on the image processing based computer vision is that object detection in a image with bounding boxes. This complicated processing has to be done with help of machine leaning based algorithms only. In the recent years research has done with machine leaning algorithms like CNN,RCNN, Fast RCNN,FasterRCNN,Yolo algorithm and etc.These algorithms have achieved the proposed concept in different levels. In this paper we presented the comparative study of each algorithm and provided efficiency and weakness contexts. [ABSTRACT FROM AUTHOR]

Megaprojects are continuing to capture worldwide attention from India’s Smart Cities, to the $4.75 billion Hadron Collider at CERN, to the US $150 billion International Space Station, and Europe’s largest railway infrastructure project, Crossrail in London. The Organization for Economic Cooperation and Development (OECD) estimates global investment needs of $6.3 trillion per year from 2016-2030 (Mirabile, et al. 2017). To meet this growing demand, there has been a recent call, within the megaproject scholarship, for a better understanding of “what goes on in megaprojects – how they are managed and organized, from within, by the managers who are tasked with bringing them to fruition.” (Söderlund, et al. 2017). Studies about megaprojects have generally concentrated on the cost conflicts between the stakeholders and cost overrun issues (Adam et al., 2014; Flyvbjerg, 2017), however these have been superseded by more important issues in recent years such as project security, protecting the health and safety of its workers, project sustainability and value creation, and managing the impact of climate change and presently a global pandemic. All of these require a more agile approach to project management and a more sophisticated intelligence that can be generated through Algorithms that are used to generate artificially intelligent systems. Through an analysis of the AI and Project Management literature, ethnographic studies, and semi-structured interviews with project management professionals, this paper explores the growing use of Artificial Intelligence to manage megaprojects including the obligations of private industry, and the government as the guardian of the public interest, while at the same time exploring the technical, managerial and ethical considerations in the deployment of AI. [ABSTRACT FROM AUTHOR]

It is important that multidisciplinary community of scientists first investigate all the myriad potential directions that quantum machine learning might go to discover what it has to offer. Among these avenues, specified algorithms such as Variational Quantum Algorithms (VQAs) have been applied in different industries for simulation, optimization, and prediction purposes. The focus of this paper is to determine the principles associated with the use of VQAs in chemistry, machine leaning, and optimization, amongst many other developed algorithms. The key results indicated that VQAs can successfully be applied in the above-mentioned fields, with the goal to (1) find ground and excited energy states of different molecules in chemistry; (2) maximizing or minimizing an objective function under specified constraints in optimization; (3) developing trainable quantum models for accurate predictions of unknown and unseen datasets. After statistical analysis, it was discovered that the proportion of publications of VQAs for optimization, machine learning, and chemical purposes have been considerably high over the past 6 years. This is justified by the sparking attention that parametrized circuits have offered over the recent years. After comparing different size effects, it was discovered that the fields of application of VQAs are interconnected, since they presented a small effect size, with Cohen’s d value being less than 0.2. From these observations, it can be concluded that the lessons developed from one application can serve as direction for others. The metallurgical industry as a field of chemistry can benefit from the methods developed in VQAs for optimization, simulation, and development of novel materials. [ABSTRACT FROM AUTHOR]

Companies invest vast amounts of time and resources into protecting themselves against outside invasion. However, employees and partners can deal as much damage from the inside. Whether from malice or negligence, the results can be equally as devastating. The focus of this study will be on discovering threats within a predefined synthetic dataset (The Insider Threat Test Dataset by Carnegie Mellon's Software Engineering Institute) using popular machine learning algorithms (Random Forest, K-Nearest Neighbour, and Multi-Layer Perceptron). A comparison of the performance of the three algorithms is provided as a first phase, using a 5-tuple flow-based data. The paper then proposes a novel approach where the detection algorithm includes psychometric analysis (based on the Big Five Personality Traits applied to hacker types) of the network users in addition to traditional features used for detection with machine learning. [ABSTRACT FROM AUTHOR]

Websites have different purposes. Some of which intend legitimate functions in the economy while some of which intend harmful cases towards users. Although various research has been made to address this problem, these detection systems still leave plenty of room for improvement, specifically on its performances. This study was based on the recommended approach for future work by a related work wherein it contains 10 base features of a URL for its classification. The recommended approach states that an extended number of features from the base features increases the detection accuracy. In this paper, it proposes a comparison between the performance of three cases: the base 10 features, an extended feature set, and a set where a feature selection algorithm is applied. The researchers utilized machine learning algorithms to build models in classifying legitimate and malicious URLs. The study showed that there is a directly proportional relationship with a model's number of features and a model's performance. Extending the number of features of the data set leads to an increase with the performance of each model. [ABSTRACT FROM AUTHOR]

Sentiment analysis of text data has become very much popular in past few years, because of the interesting challenges it can offer. Among these challenges, sarcasm is unique. We can address sarcasm as the 'Achilles heel ' of sentiment analysis. Detection of sarcasm from the given data is same time complicated and interesting. It is interesting because, if researchers can find the optimized solutions for finding sarcastic words in the data, it will enhance the sentiment analysis of that data. Even humans also have difficulties in understanding the actual interpretation of a sentence, if it is presented indirectly. This means, a person stated something, but he meant contradictory to the word meaning of the statement. This makes sarcasm detection an interesting topic for researchers. In this paper, we evaluated the performance of several machine learning models like Support vector machine, Naïve Bayes, Decision tree etc., and different ensemble models like Random Forest, XGBoost and AdaBoost etc., with the collaboration of various feature extraction methods such as Term Frequency-Inverse Document Frequency etc. The main evaluation metrics we used to evaluate the performance are accuracy, precision, f-score and recall. Based on the results, we concluded that the models such as XGBoost, LightGBM and Bagging classifiers provide better results in detecting sarcasm with respect to other machine learning models. [ABSTRACT FROM AUTHOR]

The research conducted in this paper is in the field of machine learning. The main object of the research is the learning process of an artificial neural network in order to increase its efficiency. The algorithm based on the analysis of retrospective learning data. The dynamics of changes in the values of the weights of an artificial neural network during training is an important indicator of training efficiency. The algorithm proposed in this work is based on changing the weight gradients values. Changing of the gradients weights makes it possible to understand how actively the network weights change during training. This knowledge helps to diagnose the training process and makes an adjusting the training parameters. The results of the algorithm can be used to train an artificial neural network. The network will help to determine the set of measures (actions) needed to optimize the learning process by the algorithm results. [ABSTRACT FROM AUTHOR]

In this paper, In Recent Year Machine learning plays important role in analyzing the various time of weather condition all around the world especially Indian Subcontinent. Data is available in the government website to analyze the data for few long years. This should be matched with UCI technique for the machine learning and obtain the condition of different level of data repository. Temperature and humidity level condition of different parameters were analyzed and taken as example of weather condition to monitor the weather condition. We need to design a model to fit the different condition and need to extrapolating the needs of information, and optimize the technique using some algorithm and variation should be targeted and value should be analyzed. [ABSTRACT FROM AUTHOR]

As a general rule, for pecuniary and regulatory reasons, commercial banks assiduously manage the credit risk of their loan portfolios. Algorithmic decision-making (ADM) systems may enable lenders to arrive at credit decisions that previously would not have been possible. However, the point at which utilizing ADM is optimized is still open for debate. To help illuminate the issues, a systematic literature review is conducted to investigate the following questions: How does algorithmic decision-making (ADM) contribute to the effectiveness of credit risk assessment (CRA)? And, what, if anything, can be done to improve the contribution of ADM? The review indicates that ADM’s contributions have largely been through enhanced human decision-making under uncertainty. In addition, the review underscores the importance of organizational arrangements to the successful deployment of ADM systems. Furthermore, the technology’s contribution to CRA can be improved by addressing algorithmic bias and transparency issues. [ABSTRACT FROM AUTHOR]

In view of the needs of the vehicle unit, this paper has carried on the research to the sensor distribution question. In the case of fully taking into account the information such as target identity and target priority, the mathematical model of optimal sensor allocation is established based on the maximum information gain criterion, and then the population-based incremental learning algorithm is used to solve. Finally, the validity of the method is verified by solving the problem of sensor assignment problem of the vehicle unit. [ABSTRACT FROM AUTHOR]

In the domain of network security, there is a never-ending quest for cyber-attacks that might disrupt a network. Furthermore, hostile actions in the network are rapidly increasing because of the unanticipated inception and rising use of the Internet. It is important to build a reliable intrusion detection system (IDS) to resist unwanted access to network resources, Information protection, and network intrusion detection. Deep learning has recently acquired popularity because of the possibilities it has for machine learning. As a result, Deep Learning algorithms are being used in a variety of domains, including pattern recognition and categorization. Intrusion detection analysis used data from security event monitoring to assess the network's status. Several traditional machine-learning algorithms for intrusion detection have been proposed, however detection performance and accuracy must be improved. This paper examines the many methods that have been used to classify network traffic in real-time. We selected to experiment with various methods on UNSW-NB15 datasets to find the optimum way for real-time intrusion detection. [ABSTRACT FROM AUTHOR]

In Continuous Integration (CI), regression testing is constrained by the time between commits. This demands for careful selection and/or prioritization of test cases within test suites too large to be run entirely. To this aim, some Machine Learning (ML) techniques have been proposed, as an alternative to deterministic approaches. Two broad strategies for ML-based prioritization are learning-torank and what we call ranking-to-learn (i.e., reinforcement learning). Various ML algorithms can be applied in each strategy. In this paper we introduce ten of such algorithms for adoption in CI practices, and perform a comprehensive study comparing them against each other using subjects from the Apache Commons project.We analyze the influence of several features of the code under test and of the test process. The results allow to draw criteria to support testers in selecting and tuning the technique that best fits their context. [ABSTRACT FROM AUTHOR]

When a program fails to process an input, it need not be the program code that is at fault. It can also be that the input data is faulty, for instance as result of data corruption. To get the data processed, one then has to debug the input data--that is, (1) identify which parts of the input data prevent processing, and (2) recover as much of the (valuable) input data as possible. In this paper, we present a general-purpose algorithm called ddmax that addresses these problems automatically. Through experiments, ddmax maximizes the subset of the input that can still be processed by the program, thus recovering and repairing as much data as possible; the difference between the original failing input and the "maximized" passing input includes all input fragments that could not be processed. To the best of our knowledge, ddmax is the first approach that fixes faults in the input data without requiring program analysis. In our evaluation, ddmax repaired about 69% of input files and recovered about 78% of data within one minute per input. [ABSTRACT FROM AUTHOR]

Firms engaged in electronic commerce increasingly rely on predictive analytics via machine-learning algorithms to drive a wide array of managerial decisions. The tuning of many standard machine learning algorithms can be understood as trading off bias (i.e., accuracy) with variance (i.e., precision) in the algorithm's predictions. The goal of this paper is to understand how competition between firms affects their strategic choice of such algorithms. To this end, we model the interaction of two firms choosing learning algorithms as a game and analyze its equilibria. Absent competition, players care only about the magnitude of predictive error and not its source. In contrast, our main result is that with competition, players prefer to incur error due to variance rather than due to bias, even at the cost of higher total error. In addition, we show that competition can have counterintuitive implications---for example, reducing the error incurred by a firm's algorithm can be harmful to that firm---but we provide conditions under which such phenomena do not occur. In addition to our theoretical analysis, we also validate our insights by applying our metrics to a publicly available data set. [ABSTRACT FROM AUTHOR]

The main objective of this research is to develop an effective intelligent system that can be used by medical practitioners (physicians) to accelerate diagnosis and treatment processes. In this paper, the sparse matrix approach was incorporated in neural network learning algorithm for scalability, minimize higher memory usage/storage capacity, enhancing implementation time and speed up the analysis of the data. The proposed intelligent classification system maximizes the intelligently classification of data and minimizes the number of trends inaccurately identified. For robustness, the proposed method was tested with three different datasets, namely, Hepatitis, SPECT Heart and Cleveland Heart. Therefore, an attempt was made to determine the performance indicators efficacy. Compared to some similar existing methods, the approach presented achieves improved performance. The program used for implementation of the proposed model is MATLAB R2016a (version 9.0) and executed in the 4.0 GB RAM processor of PC Intel Pentium Quad Core N3700. [ABSTRACT FROM AUTHOR]

Machine Learning (ML) is the study and the usage of the mathematical algorithms which can improve their performance without the need for human interaction. These algorithms are considered as a subset of Artificial Intelligence (AI). Machine learning algorithms use past data as input and produce new predicted values as an output. Machine learning algorithms have been used in many areas for solving an innumerable number of tasks. However, the various tasks need applying of different machine learning algorithms for obtaining maximum accuracy of the target results. In this paper, an analysis with consideration of the advantages, disadvantages, and different areas of applications in the real world are made for each of the four ML algorithm groups - supervised, unsupervised, semi-supervised, and reinforcement learning. After the comparative analysis is done, the ensemble methods boosting, stacking, and bagging are introduced, described, and compared. Emphasis is done on defining the accuracy of which ML algorithms can be improved and which ensemble methods can be used for that. Machine Learning algorithms combined with ensemble methods are highly competitive and provide the best results in most cases where they are applicable. [ABSTRACT FROM AUTHOR]

Machine learning is challenging the way we make music. Although research in deep generative models has dramatically improved the capability and fluency of music models, recent work has shown that it can be challenging for humans to partner with this new class of algorithms. In this paper, we present findings on what 13 musician/developer teams, a total of 61 users, needed when co-creating a song with AI, the challenges they faced, and how they leveraged and repurposed existing characteristics of AI to overcome some of these challenges. Many teams adopted modular approaches, such as independently running multiple smaller models that align with the musical building blocks of a song, before re-combining their results. As ML models are not easily steerable, teams also generated massive numbers of samples and curated them post-hoc, or used a range of strategies to direct the generation, or algorithmically ranked the samples. Ultimately, teams not only had to manage the "flare and focus" aspects of the creative process, but also juggle them with a parallel process of exploring and curating multiple ML models and outputs. These findings reflect a need to design machine learning-powered music interfaces that are more decomposable, steerable, interpretable, and adaptive, which in return will enable artists to more effectively explore how AI can extend their personal expression. [ABSTRACT FROM AUTHOR]

Representation learning focused on disentangling the underlying factors of variation in given data has become an important area of research in machine learning. However, most of the studies in this area have relied on datasets from the computer vision domain and thus, have not been readily extended to music. In this paper, we present a new symbolic music dataset that will help researchers working on disentanglement problems demonstrate the efficacy of their algorithms on diverse domains. This will also provide a means for evaluating algorithms specifically designed for music. To this end, we create a dataset comprising of 2-bar monophonic melodies where each melody is the result of a unique combination of nine latent factors that span ordinal, categorical, and binary types. The dataset is large enough (≈ 1.3 million data points) to train and test deep networks for disentanglement learning. In addition, we present benchmarking experiments using popular unsupervised disentanglement algorithms on this dataset and compare the results with those obtained on an image-based dataset. [ABSTRACT FROM AUTHOR]