Search results for: decision-making algorithms

Authors: Masood Varshosaz, Kamyar Hasanpour

Abstract:

In recent years, we have seen growing interest in recognizing humans in drone images for post-disaster search and rescue operations. Deep learning algorithms have shown great promise in this area, but they often require large amounts of labeled data to train the models. To keep the data acquisition cost low, augmentation techniques can be used to create additional data from existing images. There are many techniques of such that can help generate variations of an original image to improve the performance of deep learning algorithms. While data augmentation is potentially assumed to improve the accuracy and robustness of the models, it is important to ensure that the performance gains are not outweighed by the additional computational cost or complexity of implementing the techniques. To this end, it is important to evaluate the impact of data augmentation on the performance of the deep learning models. In this paper, we evaluated the most currently available 2D data augmentation techniques on a standard convolutional network which was trained for recognizing humans in drone images. The techniques include rotation, scaling, random cropping, flipping, shifting, and their combination. The results showed that the augmented models perform 1-3% better compared to a base network. However, as the augmented images only contain the human parts already visible in the original images, a new data augmentation approach is needed to include the invisible parts of the human body. Thus, we suggest a new method that employs simulated 3D human models to generate new data for training the network.

Keywords: human recognition, deep learning, drones, disaster mitigation

Procedia PDF Downloads 91

Authors: Chevon Fernando, Banuka Athuraliya

Abstract:

The purpose of this research is to provide a solution to the students’ time management, which usually becomes an issue because students must study and manage their personal commitments. "TimeTune," an AI-based study planner that provides an opportunity to maneuver study timeframes by incorporating modern machine learning algorithms with calendar applications, is unveiled as the ideal solution. The research is focused on the development of LSTM models that connect to the Google Calendar API in the process of developing learning paths that would be fit for a unique student's daily life experience and study history. A key finding of this research is the success in building the LSTM model to predict optimal study times, which, integrating with the real-time data of Google Calendar, will generate the timetables automatically in a personalized and customized manner. The methodology encompasses Agile development practices and Object-Oriented Analysis and Design (OOAD) principles, focusing on user-centric design and iterative development. By adopting this method, students can significantly reduce the tension associated with poor study habits and time management. In conclusion, "TimeTune" displays an advanced step in personalized education technology. The fact that its application of ML algorithms and calendar integration is quite innovative is slowly and steadily revolutionizing the lives of students. The excellence of maintaining a balanced academic and personal life is stress reduction, which the applications promise to provide for students when it comes to managing their studies.

Keywords: personalized learning, study planner, time management, calendar integration

Procedia PDF Downloads 46

Authors: Xu Jie

Abstract:

As a key network security method, cryptographic services must fully cope with problems such as the wide variety of cryptographic algorithms, high concurrency requirements, random job crossovers, and instantaneous surges in workloads. Its complexity and dynamics also make it difficult for traditional static security policies to cope with the ever-changing situation. Cyber Threats and Environment. Traditional resource scheduling algorithms are inadequate when facing complex decision-making problems in dynamic environments. A network cryptographic resource allocation algorithm based on reinforcement learning is proposed, aiming to optimize task energy consumption, migration cost, and fitness of differentiated services (including user, data, and task security) by modeling the multi-job collaborative cryptographic service scheduling problem as a multi-objective optimized job flow scheduling problem and using a multi-agent reinforcement learning method, efficient scheduling and optimal configuration of cryptographic service resources are achieved. By introducing reinforcement learning, resource allocation strategies can be adjusted in real-time in a dynamic environment, improving resource utilization and achieving load balancing. Experimental results show that this algorithm has significant advantages in path planning length, system delay and network load balancing and effectively solves the problem of complex resource scheduling in cryptographic services.

Keywords: cloud computing, cryptography on-demand service, reinforcement learning, workflow scheduling

Procedia PDF Downloads 10

Authors: Farhad Salek, Shahaboddin Resalati

Abstract:

The production of electric vehicles (EVs) featuring lithium-ion battery technology has substantially escalated over the past decade, demonstrating a steady and persistent upward trajectory. The imminent retirement of electric vehicle (EV) batteries after approximately eight years underscores the critical need for their redirection towards recycling, a task complicated by the current inadequacy of recycling infrastructures globally. A potential solution for such concerns involves extending the operational lifespan of electric vehicle (EV) batteries through their utilization in stationary energy storage systems during secondary applications. Such adoptions, however, require addressing the safety concerns associated with batteries’ knee points and thermal runaways. This paper develops an accurate mathematical model representative of the second-life battery packs from a cell-to-pack scale using an equivalent circuit model (ECM) methodology. Neural network algorithms are employed to forecast the degradation parameters based on the EV batteries' aging history to develop a degradation model. The degradation model is integrated with the ECM to reflect the impacts of the cycle aging mechanism on battery parameters during operation. The developed model is tested under real-life load profiles to evaluate the life span of the batteries in various operating conditions. The methodology and the algorithms introduced in this paper can be considered the basis for Battery Management System (BMS) design and techno-economic analysis of such technologies.

Keywords: second life battery, electric vehicles, degradation, neural network

Procedia PDF Downloads 64

Authors: Serge Tshibangu, Turgay Celik, Zenzo Ncube

Abstract:

Road crashes kill nearly over a million people every year, and leave millions more injured or permanently disabled. Various annual reports reveal that the percentage of fatal crashes due to fatigue/driver falling asleep comes directly after the percentage of fatal crashes due to intoxicated drivers. This percentage is higher than the combined percentage of fatal crashes due to illegal/Un-Safe U-turn and illegal/Un-Safe reversing. Although a relatively small percentage of police reports on road accidents highlights drowsiness and fatigue, the importance of these factors is greater than we might think, hidden by the undercounting of their events. Some scenarios show that these factors are significant in accidents with killed and injured people. Thus the need for an automatic drivers fatigue detection system in order to considerably reduce the number of accidents owing to fatigue.This research approaches the drivers fatigue detection problem in an innovative way by combining cues collected from both temporal analysis of drivers’ faces and environment. Monotony in driving environment is inter-related with visual symptoms of fatigue on drivers’ faces to achieve fatigue detection. Optical and infrared (IR) sensors are used to analyse the monotony in driving environment and to detect the visual symptoms of fatigue on human face. Internal cues from drivers faces and external cues from environment are combined together using machine learning algorithms to automatically detect fatigue.

Keywords: intelligent transportation systems, bayesian networks, yawning computing, machine learning algorithms

Procedia PDF Downloads 455

Authors: Tsegay Giday Woldu, Haibin Zhang, Xin Zhang, Yemane Hailu Fissuh

Abstract:

It is well known that nonlinear conjugate gradient method is one of the widely used first order methods to solve large scale unconstrained smooth optimization problems. Because of the low memory requirement, attractive theoretical features, practical computational efficiency and nice convergence properties, nonlinear conjugate gradient methods have a special role for solving large scale unconstrained optimization problems. Large scale optimization problems are with important applications in practical and scientific world. However, nonlinear conjugate gradient methods have restricted information about the curvature of the objective function and they are likely less efficient and robust compared to some second order algorithms. To overcome these drawbacks, the new modified nonlinear conjugate gradient method is presented. The noticeable features of our work are that the new search direction possesses the sufficient descent property independent of any line search and it belongs to a trust region. Under mild assumptions and standard Wolfe line search technique, the global convergence property of the proposed algorithm is established. Furthermore, to test the practical computational performance of our new algorithm, numerical experiments are provided and implemented on the set of some large dimensional unconstrained problems. The numerical results show that the proposed algorithm is an efficient and robust compared with other similar algorithms.

Keywords: conjugate gradient method, global convergence, large scale optimization, sufficient descent property

Procedia PDF Downloads 203

Authors: Kayode A. Olaniyi, Olabanji F. Omotoye, Adeola A. Ogunleye

Abstract:

The skew detection and correction form an important part of digital document analysis. This is because uncompensated skew can deteriorate document features and can complicate further document image processing steps. Efficient text document analysis and digitization can rarely be achieved when a document is skewed even at a small angle. Once the documents have been digitized through the scanning system and binarization also achieved, document skew correction is required before further image analysis. Research efforts have been put in this area with algorithms developed to eliminate document skew. Skew angle correction algorithms can be compared based on performance criteria. Most important performance criteria are accuracy of skew angle detection, range of skew angle for detection, speed of processing the image, computational complexity and consequently memory space used. The standard Hough Transform has successfully been implemented for text documentation skew angle estimation application. However, the standard Hough Transform algorithm level of accuracy depends largely on how much fine the step size for the angle used. This consequently consumes more time and memory space for increase accuracy and, especially where number of pixels is considerable large. Whenever the Hough transform is used, there is always a tradeoff between accuracy and speed. So a more efficient solution is needed that optimizes space as well as time. In this paper, an improved Hough transform (HT) technique that optimizes space as well as time to robustly detect document skew is presented. The modified algorithm of Hough Transform presents solution to the contradiction between the memory space, running time and accuracy. Our algorithm starts with the first step of angle estimation accurate up to zero decimal place using the standard Hough Transform algorithm achieving minimal running time and space but lacks relative accuracy. Then to increase accuracy, suppose estimated angle found using the basic Hough algorithm is x degree, we then run again basic algorithm from range between ±x degrees with accuracy of one decimal place. Same process is iterated till level of desired accuracy is achieved. The procedure of our skew estimation and correction algorithm of text images is implemented using MATLAB. The memory space estimation and process time are also tabulated with skew angle assumption of within 00 and 450. The simulation results which is demonstrated in Matlab show the high performance of our algorithms with less computational time and memory space used in detecting document skew for a variety of documents with different levels of complexity.

Keywords: hough-transform, skew-detection, skew-angle, skew-correction, text-document

Procedia PDF Downloads 154

Authors: Uswah Ahmad Khan, Muhammad Moazam Fraz, Humayoon Shafique Satti, Qasim Aziz

Abstract:

Approximately 10-14% of the global population experiences a functional disorder known as irritable bowel syndrome (IBS). The disorder is defined by persistent abdominal pain and an irregular bowel pattern. IBS significantly impairs work productivity and disrupts patients' daily lives and activities. Although IBS is widespread, there is still an incomplete understanding of its underlying pathophysiology. This study aims to help characterize the phenotype of IBS patients by differentiating the comorbidities found in IBS patients from those in non-IBS patients using machine learning algorithms. In this study, we extracted samples coding for IBS from the UK BioBank cohort and randomly selected patients without a code for IBS to create a total sample size of 18,000. We selected the codes for comorbidities of these cases from 2 years before and after their IBS diagnosis and compared them to the comorbidities in the non-IBS cohort. Machine learning models, including Decision Trees, Gradient Boosting, Support Vector Machine (SVM), AdaBoost, Logistic Regression, and XGBoost, were employed to assess their accuracy in predicting IBS. The most accurate model was then chosen to identify the features associated with IBS. In our case, we used XGBoost feature importance as a feature selection method. We applied different models to the top 10% of features, which numbered 50. Gradient Boosting, Logistic Regression and XGBoost algorithms yielded a diagnosis of IBS with an optimal accuracy of 71.08%, 71.427%, and 71.53%, respectively. Among the comorbidities most closely associated with IBS included gut diseases (Haemorrhoids, diverticular diseases), atopic conditions(asthma), and psychiatric comorbidities (depressive episodes or disorder, anxiety). This finding emphasizes the need for a comprehensive approach when evaluating the phenotype of IBS, suggesting the possibility of identifying new subsets of IBS rather than relying solely on the conventional classification based on stool type. Additionally, our study demonstrates the potential of machine learning algorithms in predicting the development of IBS based on comorbidities, which may enhance diagnosis and facilitate better management of modifiable risk factors for IBS. Further research is necessary to confirm our findings and establish cause and effect. Alternative feature selection methods and even larger and more diverse datasets may lead to more accurate classification models. Despite these limitations, our findings highlight the effectiveness of Logistic Regression and XGBoost in predicting IBS diagnosis.

Keywords: comorbidities, disease association, irritable bowel syndrome (IBS), predictive analytics

Procedia PDF Downloads 116

Authors: R. O. Osaseri, A. R. Usiobaifo

Abstract:

The power transformer which is responsible for the voltage transformation is of great relevance in the power system and oil-immerse transformer is widely used all over the world. A prompt and proper maintenance of the transformer is of utmost importance. The dissolved gasses content in power transformer, oil is of enormous importance in detecting incipient fault of the transformer. There is a need for accurate prediction of the incipient fault in transformer oil in order to facilitate the prompt maintenance and reducing the cost and error minimization. Study on fault prediction and diagnostic has been the center of many researchers and many previous works have been reported on the use of artificial intelligence to predict incipient failure of transformer faults. In this study machine learning technique was employed by using gradient decent algorithms and Support Vector Machine (SVM) in predicting incipient fault diagnosis of transformer. The method focuses on creating a system that improves its performance on previous result and historical data. The system design approach is basically in two phases; training and testing phase. The gradient decent algorithm is trained with a training dataset while the learned algorithm is applied to a set of new data. This two dataset is used to prove the accuracy of the proposed model. In this study a transformer fault diagnostic model based on Support Vector Machine (SVM) and gradient decent algorithms has been presented with a satisfactory diagnostic capability with high percentage in predicting incipient failure of transformer faults than existing diagnostic methods.

Keywords: diagnostic model, gradient decent, machine learning, support vector machine (SVM), transformer fault

Procedia PDF Downloads 320

Authors: Omar Farshad Jeelani, Makaire Njie, Viktoriia M. Korzhuk

Abstract:

Recently, the application of AI-powered algorithms in healthcare continues to flourish. Particularly, access to healthcare information, including patient health history, diagnostic data, and PII (Personally Identifiable Information) is paramount in the delivery of efficient patient outcomes. However, as the exchange of healthcare information between patients and healthcare providers through AI-powered solutions increases, protecting a person’s information and their privacy has become even more important. Arguably, the increased adoption of healthcare AI has resulted in a significant concentration on the security risks and protection measures to the security and privacy of healthcare data, leading to escalated analyses and enforcement. Since these challenges are brought by the use of AI-based healthcare solutions to manage healthcare data, AI-based data protection measures are used to resolve the underlying problems. Consequently, this project proposes AI-powered safeguards and policies/laws to protect the privacy of healthcare data. The project presents the best-in-school techniques used to preserve the data privacy of AI-powered healthcare applications. Popular privacy-protecting methods like Federated learning, cryptographic techniques, differential privacy methods, and hybrid methods are discussed together with potential cyber threats, data security concerns, and prospects. Also, the project discusses some of the relevant data security acts/laws that govern the collection, storage, and processing of healthcare data to guarantee owners’ privacy is preserved. This inquiry discusses various gaps and uncertainties associated with healthcare AI data collection procedures and identifies potential correction/mitigation measures.

Keywords: data privacy, artificial intelligence (AI), healthcare AI, data sharing, healthcare organizations (HCOs)

Procedia PDF Downloads 92

Authors: Alaa Sheta

Abstract:

Economic Load Dispatch (ELD) is one of the vital optimization problems in power system planning. Solving the ELD problems mean finding the best mixture of power unit outputs of all members of the power system network such that the total fuel cost is minimized while sustaining operation requirements limits satisfied across the entire dispatch phases. Many optimization techniques were proposed to solve this problem. A famous one is the Quadratic Programming (QP). QP is a very simple and fast method but it still suffer many problem as gradient methods that might trapped at local minimum solutions and cannot handle complex nonlinear functions. Numbers of metaheuristic algorithms were used to solve this problem such as Genetic Algorithms (GAs) and Particle Swarm Optimization (PSO). In this paper, another meta-heuristic search algorithm named Differential Evolution (DE) is used to solve the ELD problem in power systems planning. The practicality of the proposed DE based algorithm is verified for three and six power generator system test cases. The gained results are compared to existing results based on QP, GAs and PSO. The developed results show that differential evolution is superior in obtaining a combination of power loads that fulfill the problem constraints and minimize the total fuel cost. DE found to be fast in converging to the optimal power generation loads and capable of handling the non-linearity of ELD problem. The proposed DE solution is able to minimize the cost of generated power, minimize the total power loss in the transmission and maximize the reliability of the power provided to the customers.

Keywords: economic load dispatch, power systems, optimization, differential evolution

Procedia PDF Downloads 281

Authors: I. Valdez, J. Perdomo, M. Colindres, N. Castro

Abstract:

Today, digital servo systems are extensively used in industrial manufacturing processes, robotic applications, vehicles and other areas. In such control systems, control action is provided by digital controllers with different compensation algorithms, which are designed to meet specific requirements for a given application. Due to the constant search for optimization in industrial processes, it is of interest to design digital controllers that offer ease of realization, improved computational efficiency, affordable return rates, and ease of tuning that ultimately improve the performance of the controlled actuators. There is a vast range of options of compensation algorithms that could be used, although in the industry, most controllers used are based on a PID structure. This research article compares different types of digital compensators implemented in a servo system for DC motor position control. PID compensation is evaluated on its two most common architectures: PID position form (1 DOF), and PID speed form (2 DOF). State feedback algorithms are also evaluated, testing two modern control theory techniques: discrete state observer for non-measurable variables tracking, and a linear quadratic method which allows a compromise between the theoretical optimal control and the realization that most closely matches it. The compared control systems’ performance is evaluated through simulations in the Simulink platform, in which it is attempted to model accurately each of the system’s hardware components. The criteria by which the control systems are compared are reference tracking and disturbance rejection. In this investigation, it is considered that the accurate tracking of the reference signal for a position control system is particularly important because of the frequency and the suddenness in which the control signal could change in position control applications, while disturbance rejection is considered essential because the torque applied to the motor shaft due to sudden load changes can be modeled as a disturbance that must be rejected, ensuring reference tracking. Results show that 2 DOF PID controllers exhibit high performance in terms of the benchmarks mentioned, as long as they are properly tuned. As for controllers based on state feedback, due to the nature and the advantage which state space provides for modelling MIMO, it is expected that such controllers evince ease of tuning for disturbance rejection, assuming that the designer of such controllers is experienced. An in-depth multi-dimensional analysis of preliminary research results indicate that state feedback control method is more satisfactory, but PID control method exhibits easier implementation in most control applications.

Keywords: control, DC motor, discrete PID, discrete state feedback

Procedia PDF Downloads 266

Authors: Evan W. Krieger, Vijayan K. Asari, Saibabu Arigela

Abstract:

Video imagery captured for real-time security and surveillance applications is typically captured in complex lighting conditions. These less than ideal conditions can result in imagery that can have underexposed or overexposed regions. It is also typical that the video is too low in resolution for certain applications. The purpose of security and surveillance video is that we should be able to make accurate conclusions based on the images seen in the video. Therefore, if poor lighting and low resolution conditions occur in the captured video, the ability to make accurate conclusions based on the received information will be reduced. We propose a solution to this problem by using image preprocessing to improve these images before use in a particular application. The proposed algorithm will integrate an intensity enhancement algorithm with a super resolution technique. The intensity enhancement portion consists of a nonlinear inverse sign transformation and an adaptive contrast enhancement. The super resolution section is a single image super resolution technique is a Fourier phase feature based method that uses a machine learning approach with kernel regression. The proposed technique intelligently integrates these algorithms to be able to produce a high quality output while also being more efficient than the sequential use of these algorithms. This integration is accomplished by performing the proposed algorithm on the intensity image produced from the original color image. After enhancement and super resolution, a color restoration technique is employed to obtain an improved visibility color image.

Keywords: dynamic range compression, multi-level Fourier features, nonlinear enhancement, super resolution

Procedia PDF Downloads 553

Authors: Madhuka G. P. D. Udantha, Gihan V. Dias, Surangika Ranathunga

Abstract:

Today Websites contain very interesting applications. But there are only few methodologies to analyze User navigations through the Websites and formulating if the Website is put to correct use. The web logs are only used if some major attack or malfunctioning occurs. Web Logs contain lot interesting dealings on users in the system. Analyzing web logs has become a challenge due to the huge log volume. Finding interesting patterns is not as easy as it is due to size, distribution and importance of minor details of each log. Web logs contain very important data of user and site which are not been put to good use. Retrieving interesting information from logs gives an idea of what the users need, group users according to their various needs and improve site to build an effective and efficient site. The model we built is able to detect attacks or malfunctioning of the system and anomaly detection. Logs will be more complex as volume of traffic and the size and complexity of web site grows. Unsupervised techniques are used in this solution which is fully automated. Expert knowledge is only used in validation. In our approach first clean and purify the logs to bring them to a common platform with a standard format and structure. After cleaning module web session builder is executed. It outputs two files, Web Sessions file and Indexed URLs file. The Indexed URLs file contains the list of URLs accessed and their indices. Web Sessions file lists down the indices of each web session. Then DBSCAN and EM Algorithms are used iteratively and recursively to get the best clustering results of the web sessions. Using homogeneity, completeness, V-measure, intra and inter cluster distance and silhouette coefficient as parameters these algorithms self-evaluate themselves to input better parametric values to run the algorithms. If a cluster is found to be too large then micro-clustering is used. Using Cluster Signature Module the clusters are annotated with a unique signature called finger-print. In this module each cluster is fed to Associative Rule Learning Module. If it outputs confidence and support as value 1 for an access sequence it would be a potential signature for the cluster. Then the access sequence occurrences are checked in other clusters. If it is found to be unique for the cluster considered then the cluster is annotated with the signature. These signatures are used in anomaly detection, prevent cyber attacks, real-time dashboards that visualize users, accessing web pages, predict actions of users and various other applications in Finance, University Websites, News and Media Websites etc.

Keywords: anomaly detection, clustering, pattern recognition, web sessions

Procedia PDF Downloads 286

Authors: Zaid Jamal Saeed Almahmoud

Abstract:

Our increasing reliance on electricity, with inefficient consumption trends, has resulted in several economical and environmental threats. These threats include wasting billions of dollars, draining limited resources, and elevating the impact of climate change. As a solution, the smart grid is emerging as the future power grid, with smart techniques to optimize power consumption and electricity generation. Minimizing the peak power consumption under a fixed delay requirement is a significant problem in the smart grid. In addition, matching demand to supply is a key requirement for the success of the future electricity. In this work, we consider the problem of minimizing the peak demand under appliances constraints by scheduling power jobs with uniform release dates and deadlines. As the problem is known to be NP-Hard, we propose two versions of a heuristic algorithm for solving this problem. Our theoretical analysis and experimental results show that our proposed heuristics outperform existing methods by providing a better approximation to the optimal solution. In addition, we consider dynamic pricing methods to minimize the peak load and match demand to supply in the smart grid. Our contribution is the proposal of generic, as well as customized pricing heuristics to minimize the peak demand and match demand with supply. In addition, we propose optimal pricing algorithms that can be used when the maximum deadline period of the power jobs is relatively small. Finally, we provide theoretical analysis and conduct several experiments to evaluate the performance of the proposed algorithms.

Keywords: heuristics, optimization, smart grid, peak demand, power supply

Procedia PDF Downloads 86

Authors: M. Aghili, S. Tabarestani, C. Freytes, M. Shojaie, M. Cabrerizo, A. Barreto, N. Rishe, R. E. Curiel, D. Loewenstein, R. Duara, M. Adjouadi

Abstract:

A major challenge in medical studies, especially those that are longitudinal, is the problem of missing measurements which hinders the effective application of many machine learning algorithms. Furthermore, recent Alzheimer's Disease studies have focused on the delineation of Early Mild Cognitive Impairment (EMCI) and Late Mild Cognitive Impairment (LMCI) from cognitively normal controls (CN) which is essential for developing effective and early treatment methods. To address the aforementioned challenges, this paper explores the potential of using the eXtreme Gradient Boosting (XGBoost) algorithm in handling missing values in multiclass classification. We seek a generalized classification scheme where all prodromal stages of the disease are considered simultaneously in the classification and decision-making processes. Given the large number of subjects (1631) included in this study and in the presence of almost 28% missing values, we investigated the performance of XGBoost on the classification of the four classes of AD, NC, EMCI, and LMCI. Using 10-fold cross validation technique, XGBoost is shown to outperform other state-of-the-art classification algorithms by 3% in terms of accuracy and F-score. Our model achieved an accuracy of 80.52%, a precision of 80.62% and recall of 80.51%, supporting the more natural and promising multiclass classification.

Keywords: eXtreme gradient boosting, missing data, Alzheimer disease, early mild cognitive impairment, late mild cognitive impair, multiclass classification, ADNI, support vector machine, random forest

Procedia PDF Downloads 186

Authors: Golnush Masghati-Amoli, Paul Chin

Abstract:

Over the past recent years, with the rapid increases in data availability and computing power, Machine Learning (ML) techniques have been called on in a range of different industries for their strong predictive capability. However, the use of Machine Learning in commercial banking has been limited due to a special challenge imposed by numerous regulations that require lenders to be able to explain their analytic models, not only to regulators but often to consumers. In other words, although Machine Leaning techniques enable better prediction with a higher level of accuracy, in comparison with other industries, they are adopted less frequently in commercial banking especially for scoring purposes. This is due to the fact that Machine Learning techniques are often considered as a black box and fail to provide information on why a certain risk score is given to a customer. In order to bridge this gap between the explain-ability and performance of Machine Learning techniques, a Hybrid Model is developed at Dun and Bradstreet that is focused on blending Machine Learning algorithms with traditional approaches such as scorecards. The Hybrid Model maximizes efficiency of traditional scorecards by merging its practical benefits, such as explain-ability and the ability to input domain knowledge, with the deep insights of Machine Learning techniques which can uncover patterns scorecard approaches cannot. First, through development of Machine Learning models, engineered features and latent variables and feature interactions that demonstrate high information value in the prediction of customer risk are identified. Then, these features are employed to introduce observed non-linear relationships between the explanatory and dependent variables into traditional scorecards. Moreover, instead of directly computing the Weight of Evidence (WoE) from good and bad data points, the Hybrid Model tries to match the score distribution generated by a Machine Learning algorithm, which ends up providing an estimate of the WoE for each bin. This capability helps to build powerful scorecards with sparse cases that cannot be achieved with traditional approaches. The proposed Hybrid Model is tested on different portfolios where a significant gap is observed between the performance of traditional scorecards and Machine Learning models. The result of analysis shows that Hybrid Model can improve the performance of traditional scorecards by introducing non-linear relationships between explanatory and target variables from Machine Learning models into traditional scorecards. Also, it is observed that in some scenarios the Hybrid Model can be almost as predictive as the Machine Learning techniques while being as transparent as traditional scorecards. Therefore, it is concluded that, with the use of Hybrid Model, Machine Learning algorithms can be used in the commercial banking industry without being concerned with difficulties in explaining the models for regulatory purposes.

Keywords: machine learning algorithms, scorecard, commercial banking, consumer risk, feature engineering

Procedia PDF Downloads 132

Authors: Omar Alzeley, Sergey Utev

Abstract:

Minimizing a constrained multivariate function is the fundamental of Machine learning, and these algorithms are at the core of data mining and data visualization techniques. The decision function that maps input points to output points is based on the result of optimization. This optimization is the central of learning theory. One approach to complex systems where the dynamics of the system is inferred by a statistical analysis of the fluctuations in time of some associated observable is time series analysis. The purpose of this paper is a mathematical transition from the autoregressive model of classical time series to the matrix formalization of quantum theory. Firstly, we have proposed a quantum time series model (QTS). Although Hamiltonian technique becomes an established tool to detect a deterministic chaos, other approaches emerge. The quantum probabilistic technique is used to motivate the construction of our QTS model. The QTS model resembles the quantum dynamic model which was applied to financial data. Secondly, various statistical methods, including machine learning algorithms such as the Kalman filter algorithm, are applied to estimate and analyses the unknown parameters of the model. Finally, simulation techniques such as Markov chain Monte Carlo have been used to support our investigations. The proposed model has been examined by using real and simulated data. We establish the relation between quantum statistical machine and quantum time series via random matrix theory. It is interesting to note that the primary focus of the application of QTS in the field of quantum chaos was to find a model that explain chaotic behaviour. Maybe this model will reveal another insight into quantum chaos.

Keywords: machine learning, simulation techniques, quantum probability, tensor product, time series

Procedia PDF Downloads 467

Authors: Zhilin Wang, Zhihao Zheng, Linxin Liu

Abstract:

This study proposes a low-code model-building approach that aims to simplify the development and deployment of artificial intelligence (AI) models. With an intuitive way to drag and drop and connect components, users can easily build complex models and integrate multiple algorithms for training. After the training is completed, the system automatically generates a callable model service API. This method not only lowers the technical threshold of AI development and improves development efficiency but also enhances the flexibility of algorithm integration and simplifies the deployment process of models. The core strength of this method lies in its ease of use and efficiency. Users do not need to have a deep programming background and can complete the design and implementation of complex models with a simple drag-and-drop operation. This feature greatly expands the scope of AI technology, allowing more non-technical people to participate in the development of AI models. At the same time, the method performs well in algorithm integration, supporting many different types of algorithms to work together, which further improves the performance and applicability of the model. In the experimental part, we performed several performance tests on the method. The results show that compared with traditional model construction methods, this method can make more efficient use, save computing resources, and greatly shorten the model training time. In addition, the system-generated model service interface has been optimized for high availability and scalability, which can adapt to the needs of different application scenarios.

Keywords: low-code, model building, artificial intelligence, algorithm integration, model deployment

Procedia PDF Downloads 27

Authors: Nadia Belu, Laurenţiu Mihai Ionescu, Agnieszka Misztal

Abstract:

The automotive industry is one of the most important industries in the world that concerns not only the economy, but also the world culture. In the present financial and economic context, this field faces new challenges posed by the current crisis, companies must maintain product quality, deliver on time and at a competitive price in order to achieve customer satisfaction. Two of the most recommended techniques of quality management by specific standards of the automotive industry, in the product development, are Failure Mode and Effects Analysis (FMEA) and Control Plan. FMEA is a methodology for risk management and quality improvement aimed at identifying potential causes of failure of products and processes, their quantification by risk assessment, ranking of the problems identified according to their importance, to the determination and implementation of corrective actions related. The companies use Control Plans realized using the results from FMEA to evaluate a process or product for strengths and weaknesses and to prevent problems before they occur. The Control Plans represent written descriptions of the systems used to control and minimize product and process variation. In addition Control Plans specify the process monitoring and control methods (for example Special Controls) used to control Special Characteristics. In this paper we propose a computer-aided solution with Genetic Algorithms in order to reduce the drafting of reports: FMEA analysis and Control Plan required in the manufacture of the product launch and improved knowledge development teams for future projects. The solution allows to the design team to introduce data entry required to FMEA. The actual analysis is performed using Genetic Algorithms to find optimum between RPN risk factor and cost of production. A feature of Genetic Algorithms is that they are used as a means of finding solutions for multi criteria optimization problems. In our case, along with three specific FMEA risk factors is considered and reduce production cost. Analysis tool will generate final reports for all FMEA processes. The data obtained in FMEA reports are automatically integrated with other entered parameters in Control Plan. Implementation of the solution is in the form of an application running in an intranet on two servers: one containing analysis and plan generation engine and the other containing the database where the initial parameters and results are stored. The results can then be used as starting solutions in the synthesis of other projects. The solution was applied to welding processes, laser cutting and bending to manufacture chassis for buses. Advantages of the solution are efficient elaboration of documents in the current project by automatically generating reports FMEA and Control Plan using multiple criteria optimization of production and build a solid knowledge base for future projects. The solution which we propose is a cheap alternative to other solutions on the market using Open Source tools in implementation.

Keywords: automotive industry, FMEA, control plan, automotive technology

Procedia PDF Downloads 405

Authors: Saad M. Darwish, Adel A. El-Zoghabi, Moustafa F. Ashry

Abstract:

The popularity of the Internet and the availability of powerful computers and high-speed networks as low-cost commodity components are changing the way we use computers today. These technical opportunities have led to the possibility of using geographically distributed and multi-owner resources to solve large-scale problems in science, engineering, and commerce. Recent research on these topics has led to the emergence of a new paradigm known as Grid computing. To achieve the promising potentials of tremendous distributed resources, effective and efficient load balancing algorithms are fundamentally important. Unfortunately, load balancing algorithms in traditional parallel and distributed systems, which usually run on homogeneous and dedicated resources, cannot work well in the new circumstances. In this paper, the concept of a fast fractal transform in heterogeneous grid computing based on R-tree and the domain-range entropy is proposed to improve fault tolerance and load balancing algorithm by improve connectivity, communication delay, network bandwidth, resource availability, and resource unpredictability. A novel two-dimension figure of merit is suggested to describe the network effects on load balance and fault tolerance estimation. Fault tolerance is enhanced by adaptively decrease replication time and message cost while load balance is enhanced by adaptively decrease mean job response time. Experimental results show that the proposed method yields superior performance over other methods.

Keywords: Grid computing, load balancing, fault tolerance, R-tree, heterogeneous systems

Procedia PDF Downloads 488

Authors: Vahid Asadzadeh, Amin Ataee

Abstract:

Since the advent of social media, there has been a wave of optimism among researchers and civic activists about the influence of virtual networks on the democratization process, which has gradually waned. One of the lesser-known concerns is how to increase the possibility of hearing the voices of different minorities. According to the theory of media logic, the media, using their technological capabilities, act as a structure through which events and ideas are interpreted. Social media, through the use of the learning machine and the use of algorithms, has formed a kind of structure in which the voices of minorities and less popular topics are lost among the commotion of the trends. In fact, the recommended systems and algorithms used in social media are designed to help promote trends and make popular content more popular, and content that belongs to minorities is constantly marginalized. As social networks gradually play a more active role in politics, the possibility of freely participating in the reproduction and reinterpretation of structures in general and political structures in particular (as Laclau‎ and Mouffe had in mind‎) can be considered as criteria to democracy in action. The point is that the media logic of virtual networks is shaped by the rule and even the tyranny of the majority, and this logic does not make it possible to design a self-foundation and self-revolutionary model of democracy. In other words, today's social networks, though seemingly full of variety But they are governed by the logic of homogeneity, and they do not have the possibility of multiplicity as is the case in immanent radical democracies (influenced by Gilles Deleuze). However, with the emergence and increasing popularity of Clubhouse as a new social media, there seems to be a shift in the social media space, and that is the diminishing role of algorithms and systems reconditioners as content delivery interfaces. This has led to the fact that in the Clubhouse, the voices of minorities are better heard, and the diversity of political tendencies manifests itself better. The purpose of this article is to show, first, how social networks serve the elimination of minorities in general, and second, to argue that the media logic of social networks must adapt to new interpretations of democracy that give more space to minorities and human rights. Finally, this article will show how the Clubhouse serves the new interpretations of democracy at least in a minimal way. To achieve the mentioned goals, in this article by a descriptive-analytical method, first, the relation between media logic and postmodern democracy will be inquired. The political economy popularity in social media and its conflict with democracy will be discussed. Finally, it will be explored how the Clubhouse provides a new horizon for the concepts embodied in radical democracy, a horizon that more effectively serves the rights of minorities and human rights in general.

Keywords: algorithmic tyranny, Clubhouse, minority rights, radical democracy, social media

Procedia PDF Downloads 145

Authors: Nayer Mofidtabatabaei

Abstract:

Advancements in artificial intelligence (AI) have transformed various fields, including psychology and behavioral sciences. This paper explores the diverse ways in which AI is applied to enhance research, diagnosis, therapy, and understanding of human behavior and mental health. We discuss the potential benefits and challenges associated with AI in these fields, emphasizing the ethical considerations and the need for collaboration between AI researchers and psychological and behavioral science experts. Artificial Intelligence (AI) has gained prominence in recent years, revolutionizing multiple industries, including healthcare, finance, and entertainment. One area where AI holds significant promise is the field of psychology and behavioral sciences. AI applications in this domain range from improving the accuracy of diagnosis and treatment to understanding complex human behavior patterns. This paper aims to provide an overview of the various AI applications in psychological and behavioral sciences, highlighting their potential impact, challenges, and ethical considerations. Mental Health Diagnosis AI-driven tools, such as natural language processing and sentiment analysis, can analyze large datasets of text and speech to detect signs of mental health issues. For example, chatbots and virtual therapists can provide initial assessments and support to individuals suffering from anxiety or depression. Autism Spectrum Disorder (ASD) Diagnosis AI algorithms can assist in early ASD diagnosis by analyzing video and audio recordings of children's behavior. These tools help identify subtle behavioral markers, enabling earlier intervention and treatment. Personalized Therapy AI-based therapy platforms use personalized algorithms to adapt therapeutic interventions based on an individual's progress and needs. These platforms can provide continuous support and resources for patients, making therapy more accessible and effective. Virtual Reality Therapy Virtual reality (VR) combined with AI can create immersive therapeutic environments for treating phobias, PTSD, and social anxiety. AI algorithms can adapt VR scenarios in real-time to suit the patient's progress and comfort level. Data Analysis AI aids researchers in processing vast amounts of data, including survey responses, brain imaging, and genetic information. Privacy Concerns Collecting and analyzing personal data for AI applications in psychology and behavioral sciences raise significant privacy concerns. Researchers must ensure the ethical use and protection of sensitive information. Bias and Fairness AI algorithms can inherit biases present in training data, potentially leading to biased assessments or recommendations. Efforts to mitigate bias and ensure fairness in AI applications are crucial. Transparency and Accountability AI-driven decisions in psychology and behavioral sciences should be transparent and subject to accountability. Patients and practitioners should understand how AI algorithms operate and make decisions. AI applications in psychological and behavioral sciences have the potential to transform the field by enhancing diagnosis, therapy, and research. However, these advancements come with ethical challenges that require careful consideration. Collaboration between AI researchers and psychological and behavioral science experts is essential to harness AI's full potential while upholding ethical standards and privacy protections. The future of AI in psychology and behavioral sciences holds great promise, but it must be navigated with caution and responsibility.

Keywords: artificial intelligence, psychological sciences, behavioral sciences, diagnosis and therapy, ethical considerations

Procedia PDF Downloads 69

Authors: Alanoud Moraya Aldalan, Abdulaziz Almaleh

Abstract:

A crucial component of maintaining a customer-oriented business as in the telecom industry is understanding the reasons and factors that lead to customer churn. Competition between telecom companies has greatly increased in recent years. It has become more important to understand customers’ needs in this strong market of telecom industries, especially for those who are looking to turn over their service providers. So, predictive churn is now a mandatory requirement for retaining those customers. Machine learning can be utilized to accomplish this. Churn Prediction has become a very important topic in terms of machine learning classification in the telecommunications industry. Understanding the factors of customer churn and how they behave is very important to building an effective churn prediction model. This paper aims to predict churn and identify factors of customers’ churn based on their past service usage history. Aiming at this objective, the study makes use of feature selection, normalization, and feature engineering. Then, this study compared the performance of four different machine learning algorithms on the Orange dataset: Logistic Regression, Random Forest, Decision Tree, and Gradient Boosting. Evaluation of the performance was conducted by using the F1 score and ROC-AUC. Comparing the results of this study with existing models has proven to produce better results. The results showed the Gradients Boosting with feature selection technique outperformed in this study by achieving a 99% F1-score and 99% AUC, and all other experiments achieved good results as well.

Keywords: machine learning, gradient boosting, logistic regression, churn, random forest, decision tree, ROC, AUC, F1-score

Procedia PDF Downloads 133

Authors: G. Martino, F. Silva, E. Marchal

Abstract:

The control over delivered iron ore blend characteristics is one of the most important aspects of the mining business. The iron ore price is a function of its composition, which is the outcome of the beneficiation process. So, end-to-end integrated planning of mine operations can reduce risks of penalties on the iron ore price. In a standard iron mining company, the production chain is composed of mining, ore beneficiation, and client supply. When mine planning and client supply decisions are made uncoordinated, the beneficiation plant struggles to deliver the best blend possible. Technological improvements in several fields allowed bridging the gap between departments and boosting integrated decision-making processes. Clusterization and classification algorithms over historical production data generate reasonable previsions for quality and volume of iron ore produced for each pile of run-of-mine (ROM) processed. Mathematical modeling can use those deterministic relations to propose iron ore blends that better-fit specifications within a delivery schedule. Additionally, a model capable of representing the whole production chain can clearly compare the overall impact of different decisions in the process. This study shows how flexibilization combined with a planning optimization model between the mine and the ore beneficiation processes can reduce risks of out of specification deliveries. The model capabilities are illustrated on a hypothetical iron ore mine with magnetic separation process. Finally, this study shows ways of cost reduction or profit increase by optimizing process indicators across the production chain and integrating the different plannings with the sales decisions.

Keywords: clusterization and classification algorithms, integrated planning, mathematical modeling, optimization, penalty minimization

Procedia PDF Downloads 122

Authors: Jiří F. Urbánek, Alena Oulehlová, Hana Malachová, Jiří J. Urbánek Jr.

Abstract:

Crisis situations investigation and modelling are introduced and made within the complex system of energetic critical infrastructure, operating on peril´s environments. Every crisis situations and perils has an origin in the emergency/ crisis event occurrence and they need critical/ crisis interfaces assessment. Here, the emergency events can be expected - then crisis scenarios can be pre-prepared by pertinent organizational crisis management authorities towards their coping; or it may be unexpected - without pre-prepared scenario of event. But the both need operational coping by means of crisis management as well. The operation, forms, characteristics, behaviour and utilization of crisis management have various qualities, depending on real critical infrastructure organization perils, and prevention training processes. An aim is always - better security and continuity of the organization, which successful obtainment needs to find and investigate critical/ crisis zones and functions in critical infrastructure organization models, operating in pertinent perils environment. Our DYVELOP (Dynamic Vector Logistics of Processes) method is disposables for it. Here, it is necessary to derive and create identification algorithm of critical/ crisis interfaces. The locations of critical/ crisis interfaces are the flags of crisis situation in organization of critical infrastructure models. Then, the model of crisis situation will be displayed at real organization of Czech energetic crisis infrastructure subject in real peril environment. These efficient measures are necessary for the infrastructure protection. They will be derived for peril mitigation, crisis situation coping and for environmentally friendly organization survival, continuity and its sustainable development advanced possibilities.

Keywords: algorithms, energetic infrastructure complex system, modelling, peril´s environment

Procedia PDF Downloads 402

Authors: Tiantian Li, Lusheng Wang, Zhaohui Zhan, Daming Zhu

Abstract:

In this paper, we propose two string problems, and study algorithms and complexity of various versions for those problems. Let S = {s₁, s₂, . . . , sₘ} be a set of m strings. A common substring of S is a substring appearing in every string in S. Given a set of m strings S = {s₁, s₂, . . . , sₘ} and a positive integer k, we want to find a set C of k common substrings of S such that the k common substrings in C appear in the same order and have no overlap among the m input strings in S, and the total length of the k common substring in C is maximized. This problem is referred to as the longest total length of k common substrings from m input strings (LCSS(k, m) for short). The other problem we study here is called the longest total length of a set of common substrings with length more than l from m input string (LSCSS(l, m) for short). Given a set of m strings S = {s₁, s₂, . . . , sₘ} and a positive integer l, for LSCSS(l, m), we want to find a set of common substrings of S, each is of length more than l, such that the total length of all the common substrings is maximized. We show that both problems are NP-hard when k and m are variables. We propose dynamic programming algorithms with time complexity O(k n₁n₂) and O(n₁n₂) to solve LCSS(k, 2) and LSCSS(l, 2), respectively, where n1 and n₂ are the lengths of the two input strings. We then design an algorithm for LSCSS(l, m) when every length > l common substring appears once in each of the m − 1 input strings. The running time is O(n₁²m), where n1 is the length of the input string with no restriction on length > l common substrings. Finally, we propose a fixed parameter algorithm for LSCSS(l, m), where each length > l common substring appears m − 1 + c times among the m − 1 input strings (other than s1). In other words, each length > l common substring may repeatedly appear at most c times among the m − 1 input strings {s₂, s₃, . . . , sₘ}. The running time of the proposed algorithm is O((n12ᶜ)²m), where n₁ is the input string with no restriction on repeats. The LSCSS(l, m) is proposed to handle whole chromosome sequence alignment for different strains of the same species, where more than 98% of letters in core regions are identical.

Keywords: dynamic programming, algorithm, common substrings, string

Procedia PDF Downloads 12

Authors: Kholod Mohammad Abualnaja

Abstract:

A new algorithm is presented to find some new iterative methods for solving nonlinear equations F(x)=0 by using the variational iteration method. The efficiency of the considered method is illustrated by example. The results show that the proposed iteration technique, without linearization or small perturbation, is very effective and convenient.

Keywords: variational iteration method, nonlinear equations, Lagrange multiplier, algorithms

Procedia PDF Downloads 540

Authors: S. Jurado, E. Pazmino

Abstract:

Determination of the medial axis of a porous media sample is a non-trivial problem of interest for several disciplines, e.g., hydrology, fluid dynamics, contaminant transport, filtration, oil extraction, etc. However, the computational tools available for researchers are limited and restricted. The primary aim of this work was to develop a series of algorithms to extract porosity, medial axis structure, and pore-throat size distributions from porous media domains. A complementary objective was to provide the algorithms as free computational software available to the academic community comprising researchers and students interested in 3D data processing. The burn algorithm was tested on porous media data obtained from High-Resolution X-Ray Microtomography (HRXMT) and idealized computer-generated domains. The real data and idealized domains were discretized in voxels domains of 550³ elements and binarized to denote solid and void regions to determine porosity. Subsequently, the algorithm identifies the layer of void voxels next to the solid boundaries. An iterative process removes or 'burns' void voxels in sequence of layer by layer until all the void space is characterized. Multiples strategies were tested to optimize the execution time and use of computer memory, i.e., segmentation of the overall domain in subdomains, vectorization of operations, and extraction of single burn layer data during the iterative process. The medial axis determination was conducted identifying regions where burnt layers collide. The final medial axis structure was refined to avoid concave-grain effects and utilized to determine the pore throat size distribution. A graphic user interface software was developed to encompass all these algorithms, including the generation of idealized porous media domains. The software allows input of HRXMT data to calculate porosity, medial axis, and pore-throat size distribution and provide output in tabular and graphical formats. Preliminary tests of the software developed during this study achieved medial axis, pore-throat size distribution and porosity determination of 100³, 320³ and 550³ voxel porous media domains in 2, 22, and 45 minutes, respectively in a personal computer (Intel i7 processor, 16Gb RAM). These results indicate that the software is a practical and accessible tool in postprocessing HRXMT data for the academic community.

Keywords: medial axis, pore-throat distribution, porosity, porous media

Procedia PDF Downloads 114

Authors: Mehrshad Khosraviani, Sepehr Najjarpour

Abstract:

Considering the usage of graph-based approaches in systems and synthetic biology, and the various types of ‎the graphs employed by them, a comprehensive graph library based ‎on the three-tier architecture (3TA) was previously introduced for full representation of the biological systems. Although proposing a 3TA-based graph library, three following reasons motivated us to redesign the graph ‎library based on the service-oriented architecture (SOA): (1) Maintaining the accuracy of the data related to an input graph (including its edges, its ‎vertices, its topology, etc.) without involving the end user:‎ Since, in the case of using 3TA, the library files are available to the end users, they may ‎be utilized incorrectly, and consequently, the invalid graph data will be provided to the ‎computer algorithms. However, considering the usage of the SOA, the operation of the ‎graph registration is specified as a service by encapsulation of the library files. In other words, overall control operations needed for registration of the valid data will be the ‎responsibility of the services. (2) Partitioning of the library product into some different parts: Considering 3TA, a whole library product was provided in general. While here, the product ‎can be divided into smaller ones, such as an AND/OR graph drawing service, and each ‎one can be provided individually. As a result, the end user will be able to select any ‎parts of the library product, instead of all features, to add it to a project. (3) Reduction of the complexities: While using 3TA, several other libraries must be needed to add for connecting to the ‎database, responsibility of the provision of the needed library resources in the SOA-‎based graph library is entrusted with the services by themselves. Therefore, the end user ‎who wants to use the graph library is not involved with its complexity. In the end, in order to ‎make ‎the library easier to control in the system, and to restrict the end user from accessing the files, ‎it was preferred to use the service-oriented ‎architecture ‎‎(SOA) over the three-tier architecture (3TA) and to redevelop the previously proposed graph library based on it‎.

Keywords: Bio-Design Automation, Biological System, Graph Library, Service-Oriented Architecture, Systems and Synthetic Biology

Procedia PDF Downloads 310