Search results for: classifier algorithms

Authors: Salah Abdul Hameed Saleh, Ghada Hasan

Abstract:

The aim of this work is to produce an empirical model for the determination of particulate matter (PM10) concentration in the atmosphere using visible bands of Landsat 8 OLI satellite image over Kirkuk city- IRAQ. The suggested algorithm is established on the aerosol optical reflectance model. The reflectance model is a function of the optical properties of the atmosphere, which can be related to its concentrations. The concentration of PM10 measurements was collected using Particle Mass Profiler and Counter in a Single Handheld Unit (Aerocet 531) meter simultaneously by the Landsat 8 OLI satellite image date. The PM10 measurement locations were defined by a handheld global positioning system (GPS). The obtained reflectance values for visible bands (Coastal aerosol, Blue, Green and blue bands) of landsat 8 OLI image were correlated with in-suite measured PM10. The feasibility of the proposed algorithms was investigated based on the correlation coefficient (R) and root-mean-square error (RMSE) compared with the PM10 ground measurement data. A choice of our proposed multispectral model was founded on the highest value correlation coefficient (R) and lowest value of the root mean square error (RMSE) with PM10 ground data. The outcomes of this research showed that visible bands of Landsat 8 OLI were capable of calculating PM10 concentration with an acceptable level of accuracy.

Keywords: air pollution, PM10 concentration, Lansat8 OLI image, reflectance, multispectral algorithms, Kirkuk area

Procedia PDF Downloads 442

Authors: S. Shankar Bharathi

Abstract:

Inspection of surface defects on metallic components has always been challenging due to its specular property. Occurrences of defects such as scratches, rust, pitting are very common in metallic surfaces during the manufacturing process. These defects if unchecked can hamper the performance and reduce the life time of such component. Many of the conventional image processing algorithms in detecting the surface defects generally involve segmentation techniques, based on thresholding, edge detection, watershed segmentation and textural segmentation. They later employ other suitable algorithms based on morphology, region growing, shape analysis, neural networks for classification purpose. In this paper the work has been focused only towards detecting scratches. Global and other thresholding techniques were used to extract the defects, but it proved to be inaccurate in extracting the defects alone. However, this paper does not focus on comparison of different segmentation techniques, but rather describes a novel approach towards segmentation combined with hausdorff dilation distance. The proposed algorithm is based on the distribution of the intensity levels, that is, whether a certain gray level is concentrated or evenly distributed. The algorithm is based on extraction of such concentrated pixels. Defective images showed higher level of concentration of some gray level, whereas in non-defective image, there seemed to be no concentration, but were evenly distributed. This formed the basis in detecting the defects in the proposed algorithm. Hausdorff dilation distance based on mathematical morphology was used to strengthen the segmentation of the defects.

Keywords: metallic surface, scratches, segmentation, hausdorff dilation distance, machine vision

Procedia PDF Downloads 427

Authors: Sergey Ermolin, Olga Ermolin

Abstract:

A Deep Convolution Neural Network with Triplet Loss is used to identify duplicate images in real-estate advertisements in the presence of image artifacts such as watermarking, cropping, hue/brightness adjustment, and others. The effects of batch normalization, spatial dropout, and various convergence methodologies on the resulting detection accuracy are discussed. For comparative Return-on-Investment study (per industry request), end-2-end performance is benchmarked on both Nvidia Titan GPUs and Intel’s Xeon CPUs. A new real-estate dataset from San Francisco Bay Area is used for this work. Sufficient duplicate detection accuracy is achieved to supplement other database-grounded methods of duplicate removal. The implemented method is used in a Proof-of-Concept project in the real-estate industry.

Keywords: visual recognition, convolutional neural networks, triplet loss, spatial batch normalization with dropout, duplicate removal, advertisement technologies, performance benchmarking

Procedia PDF Downloads 338

Authors: Anum Ali, Kashaf ad Dooja, Asif Saleem

Abstract:

The future smart cities trend will be towards Internet of Things (IoT); IoT creates dynamic connections in a ubiquitous manner. Smart cities offer ease and flexibility for daily life matters. By using small devices that are connected to cloud servers based on IoT, network traffic between these devices is growing exponentially, whose security is a concerned issue, since ratio of cyber attack may make the network traffic vulnerable. This paper discusses the latest machine learning approaches in related work further to tackle the increasing rate of cyber attacks, machine learning algorithm is applied to IoT-based network traffic data. The proposed algorithm train itself on data and identify different sections of devices interaction by using supervised learning which is considered as a classifier related to a specific IoT device class. The simulation results clearly identify the attacks and produce fewer false detections.

Keywords: IoT, traffic security, deep learning, classification

Procedia PDF Downloads 152

Authors: Beier Zhu, Rui Zhang, Qi Song

Abstract:

Text contained within fixed-layout documents can be of great semantic value and so requires a high localization accuracy, such as ID cards, invoices, cheques, and passports. Recently, algorithms based on deep convolutional networks achieve high performance on text detection tasks. However, for text localization in fixed-layout documents, such algorithms detect word bounding boxes individually, which ignores the layout information. This paper presents a novel architecture built on convolutional neural networks (CNNs). A global text localization network and a regional bounding-box regression network are introduced to tackle the problem in a coarse-to-fine manner. The text localization network simultaneously locates word bounding points, which takes the layout information into account. The bounding-box regression network inputs the features pooled from arbitrarily sized RoIs and refine the localizations. These two networks share their convolutional features and are trained jointly. A typical type of fixed-layout documents: ID cards, is selected to evaluate the effectiveness of the proposed system. These networks are trained on data cropped from nature scene images, and synthetic data produced by a synthetic text generation engine. Experiments show that our approach locates high accuracy word bounding boxes and achieves state-of-the-art performance.

Keywords: bounding box regression, convolutional networks, fixed-layout documents, text localization

Procedia PDF Downloads 194

Authors: Yulong Wang, Yuan Yan Tang, Cuiming Zou, Lina Yang

Abstract:

This paper presents a novel sparse representation method for robust pattern classification. Generalized orthogonal matching pursuit (GOMP) is a recently proposed efficient sparse representation technique. However, GOMP adopts the mean square error (MSE) criterion and assign the same weights to all measurements, including both severely and slightly corrupted ones. To reduce the limitation, we propose an information-theoretic GOMP (ITGOMP) method by exploiting the correntropy induced metric. The results show that ITGOMP can adaptively assign small weights on severely contaminated measurements and large weights on clean ones, respectively. An ITGOMP based classifier is further developed for robust pattern classification. The experiments on public real datasets demonstrate the efficacy of the proposed approach.

Keywords: correntropy induced metric, matching pursuit, pattern classification, sparse representation

Procedia PDF Downloads 355

Authors: Ali Ulvi Uzer

Abstract:

Data reduction is an important topic in the field of pattern recognition applications. The basic concept is the reduction of multitudinous amounts of data down to the meaningful parts. The Principal Component Analysis (PCA) method is frequently used for data reduction. The Support Vector Machine (SVM) method is a discriminative classifier formally defined by a separating hyperplane. In other words, given labeled training data, the algorithm outputs an optimal hyperplane which categorizes new examples. This study offers a hybrid approach that uses the PCA for data reduction and Support Vector Machines (SVM) for classification. In order to detect the accuracy of the suggested system, two boreholes taken from the soil sample was used. The classification accuracies for this dataset were obtained through using ten-fold cross-validation method. As the results suggest, this system, which is performed through size reduction, is a feasible system for faster recognition of dataset so our study result appears to be very promising.

Keywords: feature selection, sequential forward selection, support vector machines, soil sample

Procedia PDF Downloads 455

Authors: Michael Zimba

Abstract:

A duplicated image region may be subjected to a number of attacks such as noise addition, compression, reflection, rotation, and scaling with the intention of either merely mating it to its targeted neighborhood or preventing its detection. In this paper, we present an effective and robust method of detecting duplicated regions inclusive of those affected by the various attacks. In order to reduce the dimension of the image, the proposed algorithm firstly performs discrete wavelet transform, DWT, of a suspicious image. However, unlike most existing copy move image forgery (CMIF) detection algorithms operating in the DWT domain which extract only the low frequency sub-band of the DWT of the suspicious image thereby leaving valuable information in the other three sub-bands, the proposed algorithm simultaneously extracts features from all the four sub-bands. The extracted features are not only more accurate representation of image regions but also robust to additive noise, JPEG compression, and affine transformation. Furthermore, principal component analysis-eigenvalue decomposition, PCA-EVD, is applied to reduce the dimension of the features. The extracted features are then sorted using the more computationally efficient Radix Sort algorithm. Finally, same affine transformation selection, SATS, a duplication verification method, is applied to detect duplicated regions. The proposed algorithm is not only fast but also more robust to attacks compared to the related CMIF detection algorithms. The experimental results show high detection rates.

Keywords: affine transformation, discrete wavelet transform, radix sort, SATS

Procedia PDF Downloads 230

Authors: A. Biran, P. Sobhe Bidari, A. Almazroe, V. Lakshminarayanan, K. Raahemifar

Abstract:

Diabetic Retinopathy (DR) is a severe retinal disease which is caused by diabetes mellitus. It leads to blindness when it progress to proliferative level. Early indications of DR are the appearance of microaneurysms, hemorrhages and hard exudates. In this paper, an automatic algorithm for detection of DR has been proposed. The algorithm is based on combination of several image processing techniques including Circular Hough Transform (CHT), Contrast Limited Adaptive Histogram Equalization (CLAHE), Gabor filter and thresholding. Also, Support Vector Machine (SVM) Classifier is used to classify retinal images to normal or abnormal cases including non-proliferative or proliferative DR. The proposed method has been tested on images selected from Structured Analysis of the Retinal (STARE) database using MATLAB code. The method is perfectly able to detect DR. The sensitivity specificity and accuracy of this approach are 90%, 87.5%, and 91.4% respectively.

Keywords: diabetic retinopathy, fundus images, STARE, Gabor filter, support vector machine

Procedia PDF Downloads 294

Authors: O. Sunantha, S. Zhenfeng, S. Phattraporn, A. Zeeshan

Abstract:

The decline of soil organic carbon (SOC) in global agriculture is a critical issue requiring rapid and accurate estimation for informed policymaking. While it is recognized that SOC predictors vary significantly when derived from remote sensing data and environmental variables, identifying the specific parameters most suitable for accurately estimating SOC in diverse agricultural areas remains a challenge. This study utilizes remote sensing data to precisely estimate SOC and identify influential factors in diverse agricultural areas, such as paddy, corn, sugarcane, cassava, and perennial crops. Extreme gradient boosting (XGBoost), random forest (RF), and support vector regression (SVR) models are employed to analyze these factors' impact on SOC estimation. The results show key factors influencing SOC estimation include slope, vegetation indices (EVI), spectral reflectance indices (red index, red edge2), temperature, land use, and surface soil moisture, as indicated by their averaged importance scores across XGBoost, RF, and SVR models. Therefore, using different machine learning algorithms for SOC estimation reveals varying influential factors from remote sensing data and environmental variables. This approach emphasizes feature selection, as different machine learning algorithms identify various key factors from remote sensing data and environmental variables for accurate SOC estimation.

Keywords: factors influencing SOC estimation, remote sensing data, environmental variables, machine learning

Procedia PDF Downloads 34

Authors: Ferzat Anka

Abstract:

Many types of problems can be solved using traditional analytical methods. However, these methods take a long time and cause inefficient use of resources. In particular, different approaches may be required in solving complex and global engineering problems that we frequently encounter in real life. The bigger and more complex a problem, the harder it is to solve. Such problems are called Nondeterministic Polynomial time (NP-hard) in the literature. The main reasons for recommending different metaheuristic algorithms for various problems are the use of simple concepts, the use of simple mathematical equations and structures, the use of non-derivative mechanisms, the avoidance of local optima, and their fast convergence. They are also flexible, as they can be applied to different problems without very specific modifications. Thanks to these features, it can be easily embedded even in many hardware devices. Accordingly, this approach can also be used in trend application areas such as IoT, big data, and parallel structures. Indeed, the metaheuristic approaches are algorithms that return near-optimal results for solving large-scale optimization problems. This study is focused on the new metaheuristic method that has been merged with the chaotic approach. It is based on the chaos theorem and helps relevant algorithms to improve the diversity of the population and fast convergence. This approach is based on Chimp Optimization Algorithm (ChOA), that is a recently introduced metaheuristic algorithm inspired by nature. This algorithm identified four types of chimpanzee groups: attacker, barrier, chaser, and driver, and proposed a suitable mathematical model for them based on the various intelligence and sexual motivations of chimpanzees. However, this algorithm is not more successful in the convergence rate and escaping of the local optimum trap in solving high-dimensional problems. Although it and some of its variants use some strategies to overcome these problems, it is observed that it is not sufficient. Therefore, in this study, a newly expanded variant is described. In the algorithm called Ex-ChOA, hybrid models are proposed for position updates of search agents, and a dynamic switching mechanism is provided for transition phases. This flexible structure solves the slow convergence problem of ChOA and improves its accuracy in multidimensional problems. Therefore, it tries to achieve success in solving global, complex, and constrained problems. The main contribution of this study is 1) It improves the accuracy and solves the slow convergence problem of the ChOA. 2) It proposes new hybrid movement strategy models for position updates of search agents. 3) It provides success in solving global, complex, and constrained problems. 4) It provides a dynamic switching mechanism between phases. The performance of the Ex-ChOA algorithm is analyzed on a total of 8 benchmark functions, as well as a total of 2 classical and constrained engineering problems. The proposed algorithm is compared with the ChoA, and several well-known variants (Weighted-ChoA, Enhanced-ChoA) are used. In addition, an Improved algorithm from the Grey Wolf Optimizer (I-GWO) method is chosen for comparison since the working model is similar. The obtained results depict that the proposed algorithm performs better or equivalently to the compared algorithms.

Keywords: optimization, metaheuristic, chimp optimization algorithm, engineering constrained problems

Procedia PDF Downloads 77

Authors: Zoubida Mahi

Abstract:

A successful satellite mission in space requires a reliable attitude and orbit control system to command, control and position the satellite in appropriate orbits. Several sensors are used for attitude control, such as magnetic sensors, earth sensors, horizon sensors, gyroscopes, and solar sensors. The star tracker is the most accurate sensor compared to other sensors, and it is able to offer high-accuracy attitude control without the need for prior attitude information. There are mainly three approaches in star sensor research: digital simulation, hardware in the loop simulation, and field test of star observation. In the digital simulation approach, all of the processes are done in software, including star image simulation. Hence, it is necessary to develop star image simulation software that could simulate real space environments and various star sensor configurations. In this paper, we present a new stellar image simulation tool that is used to test and validate the stellar sensor algorithms; the developed tool allows to simulate of stellar images with several types of noise, such as background noise, gaussian noise, Poisson noise, multiplicative noise, and several scenarios that exist in space such as the presence of the moon, the presence of optical system problem, illumination and false objects. On the other hand, we present in this paper a new star extraction algorithm based on a new centroid calculation method. We compared our algorithm with other star extraction algorithms from the literature, and the results obtained show the star extraction capability of the proposed algorithm.

Keywords: star tracker, star simulation, star detection, centroid, noise, scenario

Procedia PDF Downloads 96

Authors: Thinh Ngo, Paul Rad, Brian Kelley

Abstract:

Noise estimation is essential in today wireless systems for power control, adaptive modulation, interference suppression and quality of service. Deep learning (DL) has already been applied in the physical layer for modulation and signal classifications. Unacceptably low accuracy of less than 50% is found to undermine traditional application of DL classification for SNR prediction. In this paper, we use divide-and-conquer algorithm and classifier fusion method to simplify SNR classification and therefore enhances DL learning and prediction. Specifically, multiple CNNs are used for classification rather than a single CNN. Each CNN performs a binary classification of a single SNR with two labels: less than, greater than or equal. Together, multiple CNNs are combined to effectively classify over a range of SNR values from −20 ≤ SNR ≤ 32 dB.We use pre-trained CNNs to predict SNR over a wide range of joint channel parameters including multiple Doppler shifts (0, 60, 120 Hz), power-delay profiles, and signal-modulation types (QPSK,16QAM,64-QAM). The approach achieves individual SNR prediction accuracy of 92%, composite accuracy of 70% and prediction convergence one order of magnitude faster than that of traditional estimation.

Keywords: classification, CNN, deep learning, prediction, SNR

Procedia PDF Downloads 133

Authors: Nyeng P. Gyang

Abstract:

Even though past, current and future trends suggest that multicore and cloud computing systems are increasingly prevalent/ubiquitous, this class of parallel systems is nonetheless underutilized, in general, and barely used for research on employing parallel Delaunay triangulation for parallel surface modeling and generation, in particular. The performances, of actual/physical and virtual/cloud multicore systems/machines, at executing various algorithms, which implement various parallelization strategies of the incremental insertion technique of the Delaunay triangulation algorithm, were evaluated. T-tests were run on the data collected, in order to determine whether various performance metrics differences (including execution time, speedup and efficiency) were statistically significant. Results show that the actual machine is approximately twice faster than the virtual machine at executing the same programs for the various parallelization strategies. Results, which furnish the scalability behaviors of the various parallelization strategies, also show that some of the differences between the performances of these systems, during different runs of the algorithms on the systems, were statistically significant. A few pseudo superlinear speedup results, which were computed from the raw data collected, are not true superlinear speedup values. These pseudo superlinear speedup values, which arise as a result of one way of computing speedups, disappear and give way to asymmetric speedups, which are the accurate kind of speedups that occur in the experiments performed.

Keywords: cloud computing systems, multicore systems, parallel Delaunay triangulation, parallel surface modeling and generation

Procedia PDF Downloads 206

Authors: Narges Farokhshad, Milad Molazadeh, Saman Jamalabbasi, Hamed Babaei Giglou, Saeed Bibak

Abstract:

The Persian language is an inflectional subject-object-verb language. This fact makes Persian a more uncertain language. However, using techniques such as Zero-Width Non-Joiner (ZWNJ) recognition, punctuation restoration, and Persian Ezafe construction will lead us to a more understandable and precise language. In most of the works in Persian, these techniques are addressed individually. Despite that, we believe that for text refinement in Persian, all of these tasks are necessary. In this work, we proposed a ViraPart framework that uses embedded ParsBERT in its core for text clarifications. First, used the BERT variant for Persian followed by a classifier layer for classification procedures. Next, we combined models outputs to output cleartext. In the end, the proposed model for ZWNJ recognition, punctuation restoration, and Persian Ezafe construction performs the averaged F1 macro scores of 96.90%, 92.13%, and 98.50%, respectively. Experimental results show that our proposed approach is very effective in text refinement for the Persian language.

Keywords: Persian Ezafe, punctuation, ZWNJ, NLP, ParsBERT, transformers

Procedia PDF Downloads 215

Authors: Yasmine Abu Adla, Racha Soubra, Milana Kasab, Mohamad O. Diab, Aly Chkeir

Abstract:

Over the past several years, researchers have shown a great interest in assessing the mobility of elderly people to measure their functional status. Usually, such an assessment is done by conducting tests that require the subject to walk a certain distance, turn around, and finally sit back down. Consequently, this study aims to provide an at home monitoring system to assess the patient’s status continuously. Thus, we proposed a technique to automatically detect when a subject sits down while walking at home. In this study, we utilized a Doppler radar system to capture the motion of the subjects. More than 20 features were extracted from the radar signals, out of which 11 were chosen based on their intraclass correlation coefficient (ICC > 0.75). Accordingly, the sequential floating forward selection wrapper was applied to further narrow down the final feature vector. Finally, 5 features were introduced to the linear discriminant analysis classifier, and an accuracy of 93.75% was achieved as well as a precision and recall of 95% and 90%, respectively.

Keywords: Doppler radar system, stand-to-sit phase, TUG test, machine learning, classification

Procedia PDF Downloads 161

Authors: António J. Gano, Carmen Rangel

Abstract:

Currently, electric energy storage systems for stationary applications have known an increasing interest, namely with the integration of local renewable energy power sources into energy communities. Li-ion batteries are considered the leading electric storage devices to achieve this integration, and Battery Management Systems (BMS) are decisive for their control and optimum performance. In this work, the advancement of a smart BMS (sBMS) prototype with a modular distributed topology is described. The system, still under development, has a distributed architecture with modular characteristics to operate with different battery pack topologies and charge capacities, integrating adaptive algorithms for functional state real-time monitoring and management of multicellular Li-ion batteries, and is intended for application in the context of a local energy community fed by renewable energy sources. This sBMS system includes different developed hardware units: (1) Cell monitoring units (CMUs) for interfacing with each individual cell or module monitoring within the battery pack; (2) Battery monitoring and switching unit (BMU) for global battery pack monitoring, thermal control and functional operating state switching; (3) Main management and local control unit (MCU) for local sBMS’s management and control, also serving as a communications gateway to external systems and devices. This architecture is fully expandable to battery packs with a large number of cells, or modules, interconnected in series, as the several units have local data acquisition and processing capabilities, communicating over a standard CAN bus and will be able to operate almost autonomously. The CMU units are intended to be used with Li-ion cells but can be used with other cell chemistries, with output voltages within the 2.5 to 5 V range. The different unit’s characteristics and specifications are described, including the different implemented hardware solutions. The developed hardware supports both passive and active methods for charge equalization, considered fundamental functionalities for optimizing the performance and the useful lifetime of a Li-ion battery package. The functional characteristics of the different units of this sBMS system, including different process variables data acquisition using a flexible set of sensors, can support the development of custom algorithms for estimating the parameters defining the functional states of the battery pack (State-of-Charge, State-of-Health, etc.) as well as different charge equalizing strategies and algorithms. This sBMS system is intended to interface with other systems and devices using standard communication protocols, like those used by the Internet of Things. In the future, this sBMS architecture can evolve to a fully decentralized topology, with all the units using Wi-Fi protocols and integrating a mesh network, making unnecessary the MCU unit. The status of the work in progress is reported, leading to conclusions on the system already executed, considering the implemented hardware solution, not only as fully functional advanced and configurable battery management system but also as a platform for developing custom algorithms and optimizing strategies to achieve better performance of electric energy stationary storage devices.

Keywords: Li-ion battery, smart BMS, stationary electric storage, distributed BMS

Procedia PDF Downloads 100

Authors: Mirjana Ruppel, Rajendra Persad, Amit Bahl, Sanja Dogramadzi, Chris Melhuish, Lyndon Smith

Abstract:

Multimodal image registration is a profoundly complex task which is why deep learning has been used widely to address it in recent years. However, two main challenges remain: Firstly, the lack of ground truth data calls for an unsupervised learning approach, which leads to the second challenge of defining a feasible loss function that can compare two images of different modalities to judge their level of alignment. To avoid this issue altogether we implement a generative adversarial network consisting of two registration networks GAB, GBA and two discrimination networks DA, DB connected by spatial transformation layers. GAB learns to generate a deformation field which registers an image of the modality B to an image of the modality A. To do that, it uses the feedback of the discriminator DB which is learning to judge the quality of alignment of the registered image B. GBA and DA learn a mapping from modality A to modality B. Additionally, a cycle-consistency loss is implemented. For this, both registration networks are employed twice, therefore resulting in images ˆA, ˆB which were registered to ˜B, ˜A which were registered to the initial image pair A, B. Thus the resulting and initial images of the same modality can be easily compared. A dataset of liver CT and MRI was used to evaluate the quality of our approach and to compare it against learning and non-learning based registration algorithms. Our approach leads to dice scores of up to 0.80 ± 0.01 and is therefore comparable to and slightly more successful than algorithms like SimpleElastix and VoxelMorph.

Keywords: cycle consistency, deformable multimodal image registration, deep learning, GAN

Procedia PDF Downloads 131

Authors: Aadityaa Singla

Abstract:

The big five personality traits are as follows: openness, conscientiousness, extraversion, agreeableness, and neuroticism. In order to get an insight into their personality, many flocks to these categories, which each have different meanings/characteristics. This information is important not only to individuals but also to career professionals and psychologists who can use this information for candidate assessment or job recruitment. The links between AI and psychology have been well studied in cognitive science, but it is still a rather novel development. It is possible for various AI classification models to accurately predict a personality question via ten input questions. This would contrast with the hundred questions that normal humans have to answer to gain a complete picture of their five personality traits. In order to approach this problem, various AI classification models were used on a dataset to predict what a user may answer. From there, the model's prediction was compared to its actual response. Normally, there are five answer choices (a 20% chance of correct guess), and the models exceed that value to different degrees, proving their significance. By utilizing an MLP classifier, decision tree, linear model, and K-nearest neighbors, they were able to obtain a test accuracy of 86.643, 54.625, 47.875, and 52.125, respectively. These approaches display that there is potential in the future for more nuanced predictions to be made regarding personality.

Keywords: machine learning, personally, big five personality traits, cognitive science

Procedia PDF Downloads 145

Authors: Riadh Zorgati, Thomas Triboulet

Abstract:

In quite diverse application areas such as astronomy, medical imaging, geophysics or nondestructive evaluation, many problems related to calibration, fitting or estimation of a large number of input parameters of a model from a small amount of output noisy data, can be cast as inverse problems. Due to noisy data corruption, insufficient data and model errors, most inverse problems are ill-posed in a Hadamard sense, i.e. existence, uniqueness and stability of the solution are not guaranteed. A wide class of inverse problems in physics relates to the Fredholm equation of the first kind. The ill-posedness of such inverse problem results, after discretization, in a very ill-conditioned linear system of equations, the condition number of the associated matrix can typically range from 109 to 1018. This condition number plays the role of an amplifier of uncertainties on data during inversion and then, renders the inverse problem difficult to handle numerically. Similar problems appear in other areas such as numerical optimization when using interior points algorithms for solving linear programs leads to face ill-conditioned systems of linear equations. Devising efficient solution approaches for such system of equations is therefore of great practical interest. Efficient iterative algorithms are proposed for solving a system of linear equations. The approach is based on a preconditioning of the initial matrix of the system with an approximation of a generalized inverse leading to a stochastic preconditioned matrix. This approach, valid for non-negative matrices, is first extended to hermitian, semi-definite positive matrices and then generalized to any complex rectangular matrices. The main results obtained are as follows: 1) We are able to build a generalized inverse of any complex rectangular matrix which satisfies the convergence condition requested in iterative algorithms for solving a system of linear equations. This completes the (short) list of generalized inverse having this property, after Kaczmarz and Cimmino matrices. Theoretical results on both the characterization of the type of generalized inverse obtained and the convergence are derived. 2) Thanks to its properties, this matrix can be efficiently used in different solving schemes as Richardson-Tanabe or preconditioned conjugate gradients. 3) By using Lp norms, we propose generalized Kaczmarz’s type matrices. We also show how Cimmino's matrix can be considered as a particular case consisting in choosing the Euclidian norm in an asymmetrical structure. 4) Regarding numerical results obtained on some pathological well-known test-cases (Hilbert, Nakasaka, …), some of the proposed algorithms are empirically shown to be more efficient on ill-conditioned problems and more robust to error propagation than the known classical techniques we have tested (Gauss, Moore-Penrose inverse, minimum residue, conjugate gradients, Kaczmarz, Cimmino). We end on a very early prospective application of our approach based on stochastic matrices aiming at computing some parameters (such as the extreme values, the mean, the variance, …) of the solution of a linear system prior to its resolution. Such an approach, if it were to be efficient, would be a source of information on the solution of a system of linear equations.

Keywords: conditioning, generalized inverse, linear system, norms, stochastic matrix

Procedia PDF Downloads 135

Authors: Amin Ranjbar, Amir Arsalan Jalili Zolfaghari, Amir Abolfazl Suratgar, Mehrdad Khajavi

Abstract:

This paper describes two stages of learning-based fault detection procedure in alternators. The procedure consists of three states of machine condition namely shortened brush, high impedance relay and maintaining a healthy condition in the alternator. The fault detection algorithm uses Wigner-Ville distribution as a feature extractor and also appropriate feature classifier. In this work, ANN (Artificial Neural Network) and also SVM (support vector machine) were compared to determine more suitable performance evaluated by the mean squared of errors criteria. Modules work together to detect possible faulty conditions of machines working. To test the method performance, a signal database is prepared by making different conditions on a laboratory setup. Therefore, it seems by implementing this method, satisfactory results are achieved.

Keywords: alternator, artificial neural network, support vector machine, time-frequency analysis, Wigner-Ville distribution

Procedia PDF Downloads 374

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 93

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 48

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 12

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 65

Authors: Rodrigo Arnaldo Scarpel

Abstract:

Trend following is a widely accepted investment strategy that adopts a rule-based trading mechanism that rather than striving to predict market direction or on information gathering to decide when to buy and when to sell a stock. Thus, in trend following one must respond to market’s movements that has recently happen and what is currently happening, rather than on what will happen. Optimally, in trend following strategy, is to catch a bull market at its early stage, ride the trend, and liquidate the position at the first evidence of the subsequent bear market. For applying the trend following strategy one needs to find the trend and identify trade signals. In order to avoid false signals, i.e., identify fluctuations of short, mid and long terms and to separate noise from real changes in the trend, most academic works rely on moving averages and other technical analysis indicators, such as the moving average convergence divergence (MACD) and the relative strength index (RSI) to uncover intelligible stock trading rules following trend following strategy philosophy. Recently, some works has applied machine learning techniques for trade rules discovery. In those works, the process of rule construction is based on evolutionary learning which aims to adapt the rules to the current environment and searches for the global optimum rules in the search space. In this work, instead of focusing on the usage of machine learning techniques for creating trading rules, a time series trend classification employing a semi-supervised approach was used to early identify both the beginning and the end of upward and downward trends. Such classification model can be employed to identify trade signals and the decision-making procedure is that if an up-trend (down-trend) is identified, a buy (sell) signal is generated. Semi-supervised learning is used for model training when only part of the data is labeled and Semi-supervised classification aims to train a classifier from both the labeled and unlabeled data, such that it is better than the supervised classifier trained only on the labeled data. For illustrating the proposed approach, it was employed daily trade information, including the open, high, low and closing values and volume from January 1, 2000 to December 31, 2022, of the São Paulo Exchange Composite index (IBOVESPA). Through this time period it was visually identified consistent changes in price, upwards or downwards, for assigning labels and leaving the rest of the days (when there is not a consistent change in price) unlabeled. For training the classification model, a pseudo-label semi-supervised learning strategy was used employing different technical analysis indicators. In this learning strategy, the core is to use unlabeled data to generate a pseudo-label for supervised training. For evaluating the achieved results, it was considered the annualized return and excess return, the Sortino and the Sharpe indicators. Through the evaluated time period, the obtained results were very consistent and can be considered promising for generating the intended trading signals.

Keywords: evolutionary learning, semi-supervised classification, time series data, trading signals generation

Procedia PDF Downloads 89

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 205

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 158

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 118