Search results for: floor estimation algorithm

Authors: Kaleem Kashif, Tayyaba Anam, Yizhi Wu

Abstract:

This paper introduces a methodology for advancing Italian speech vowels landmark detection within the distinctive feature-based speech recognition domain. Leveraging the legacy tool 'xkl' by integrating combined convolutional neural networks (CNNs) and recurrent neural networks (RNNs), the study presents a comprehensive enhancement to the 'xkl' legacy software. This integration incorporates re-assigned spectrogram methodologies, enabling meticulous acoustic analysis. Simultaneously, our proposed model, integrating combined CNNs and RNNs, demonstrates unprecedented precision and robustness in landmark detection. The augmentation of re-assigned spectrogram fusion within the 'xkl' software signifies a meticulous advancement, particularly enhancing precision related to vowel formant estimation. This augmentation catalyzes unparalleled accuracy in landmark detection, resulting in a substantial performance leap compared to conventional methods. The proposed model emerges as a state-of-the-art solution in the distinctive feature-based speech recognition systems domain. In the realm of deep learning, a synergistic integration of combined CNNs and RNNs is introduced, endowed with specialized temporal embeddings, harnessing self-attention mechanisms, and positional embeddings. The proposed model allows it to excel in capturing intricate dependencies within Italian speech vowels, rendering it highly adaptable and sophisticated in the distinctive feature domain. Furthermore, our advanced temporal modeling approach employs Bayesian temporal encoding, refining the measurement of inter-landmark intervals. Comparative analysis against state-of-the-art models reveals a substantial improvement in accuracy, highlighting the robustness and efficacy of the proposed methodology. Upon rigorous testing on a database (LaMIT) speech recorded in a silent room by four Italian native speakers, the landmark detector demonstrates exceptional performance, achieving a 95% true detection rate and a 10% false detection rate. A majority of missed landmarks were observed in proximity to reduced vowels. These promising results underscore the robust identifiability of landmarks within the speech waveform, establishing the feasibility of employing a landmark detector as a front end in a speech recognition system. The synergistic integration of re-assigned spectrogram fusion, CNNs, RNNs, and Bayesian temporal encoding not only signifies a significant advancement in Italian speech vowels landmark detection but also positions the proposed model as a leader in the field. The model offers distinct advantages, including unparalleled accuracy, adaptability, and sophistication, marking a milestone in the intersection of deep learning and distinctive feature-based speech recognition. This work contributes to the broader scientific community by presenting a methodologically rigorous framework for enhancing landmark detection accuracy in Italian speech vowels. The integration of cutting-edge techniques establishes a foundation for future advancements in speech signal processing, emphasizing the potential of the proposed model in practical applications across various domains requiring robust speech recognition systems.

Keywords: landmark detection, acoustic analysis, convolutional neural network, recurrent neural network

Procedia PDF Downloads 35

Authors: Agnieszka Krzemińska, Katarzyna Świderek, Vicente Molinier, Piotr Paneth

Abstract:

In order to understand in details the interactions between ligands and the enzyme isotope effects were studied between clinically used drugs that bind in the active site of Human Immunodeficiency Virus Reverse Transcriptase, HIV-1 RT, as well as triazole-based inhibitor that binds in the allosteric pocket of this enzyme. The magnitudes and origins of the resulting binding isotope effects were analyzed. Subsequently, binding isotope effect of the same triazole-based inhibitor bound in the active site were analyzed and compared. Together, these results show differences in binding origins in two sites of the enzyme and allow to analyze binding mode and place of newly synthesized inhibitors. Typical protocol is described below on the example of triazole ligand in the allosteric pocket. Triazole was docked into allosteric cavity of HIV-1 RT with Glide using extra-precision mode as implemented in Schroedinger software. The structure of HIV-1 RT was obtained from Protein Data Bank as structure of PDB ID 2RKI. The pKa for titratable amino acids was calculated using PROPKA software, and in order to neutralize the system 15 Cl- were added using tLEaP package implemented in AMBERTools ver.1.5. Also N-terminals and C-terminals were build using tLEaP. The system was placed in 144x160x144Å3 orthorhombic box of water molecules using NAMD program. Missing parameters for triazole were obtained at the AM1 level using Antechamber software implemented in AMBERTools. The energy minimizations were carried out by means of a conjugate gradient algorithm using NAMD. Then system was heated from 0 to 300 K with temperature increment 0.001 K. Subsequently 2 ns Langevin−Verlet (NVT) MM MD simulation with AMBER force field implemented in NAMD was carried out. Periodic Boundary Conditions and cut-offs for the nonbonding interactions, range radius from 14.5 to 16 Å, are used. After 2 ns relaxation 200 ps of QM/MM MD at 300 K were simulated. The triazole was treated quantum mechanically at the AM1 level, protein was described using AMBER and water molecules were described using TIP3P, as implemented in fDynamo library. Molecules 20 Å apart from the triazole were kept frozen, with cut-offs established on range radius from 14.5 to 16 Å. In order to describe interactions between triazole and RT free energy of binding using Free Energy Perturbation method was done. The change in frequencies from ligand in solution to ligand bounded in enzyme was used to calculate binding isotope effects.

Keywords: binding isotope effects, molecular dynamics, HIV, reverse transcriptase

Procedia PDF Downloads 412

Authors: Ekta Yadav, Sukanta Bhattacharya, Brijesh Yadav, Ariel Hus, Jagjit Yadav

Abstract:

Background and Significance: Respiratory exposure to toxicants (chemicals or particulates) causes disruption of lung homeostasis leading to lung toxicity/injury manifested as pulmonary inflammation, edema, and/or other effects depending on the type and extent of exposure. This emphasizes the need for investigating toxicant type-specific mechanisms to understand therapeutic targets. Aquaporins, aka water channels, are known to play a role in lung homeostasis. Particularly, the two major lung aquaporins AQP5 and AQP1 expressed in alveolar epithelial and vasculature endothelia respectively allow for movement of the fluid between the alveolar air space and the associated vasculature. In view of this, the current study is focused on understanding the regulation of lung aquaporins and other targets during inhalation exposure to toxic chemicals (Cigarette smoke chemicals) versus toxic particles (Carbon nanoparticles) or co-exposures to understand their relevance as markers of injury and intervention. Methodologies: C57BL/6 mice (5-7 weeks old) were used in this study following an approved protocol by the University of Cincinnati Institutional Animal Care and Use Committee (IACUC). The mice were exposed via oropharyngeal aspiration to multiwall carbon nanotube (MWCNT) particles suspension once (33 ugs/mouse) followed by housing for four weeks or to Cigarette smoke Extract (CSE) using a daily dose of 30µl/mouse for four weeks, or to co-exposure using the combined regime. Control groups received vehicles following the same dosing schedule. Lung toxicity/injury was assessed in terms of homeostasis changes in the lung tissue and lumen. Exposed lungs were analyzed for transcriptional expression of specific targets (AQPs, surfactant protein A, Mucin 5b) in relation to tissue homeostasis. Total RNA from lungs extracted using TRIreagent kit was analyzed using qRT-PCR based on gene-specific primers. Total protein in bronchoalveolar lavage (BAL) fluid was determined by the DC protein estimation kit (BioRad). GraphPad Prism 5.0 (La Jolla, CA, USA) was used for all analyses. Major findings: CNT exposure alone or as co-exposure with CSE increased the total protein content in the BAL fluid (lung lumen rinse), implying compromised membrane integrity and cellular infiltration in the lung alveoli. In contrast, CSE showed no significant effect. AQP1, required for water transport across membranes of endothelial cells in lungs, was significantly upregulated in CNT exposure but downregulated in CSE exposure and showed an intermediate level of expression for the co-exposure group. Both CNT and CSE exposures had significant downregulating effects on Muc5b, and SP-A expression and the co-exposure showed either no significant effect (Muc5b) or significant downregulating effect (SP-A), suggesting an increased propensity for infection in the exposed lungs. Conclusions: The current study based on the lung toxicity mouse model showed that both toxicant types, particles (CNT) versus chemicals (CSE), cause similar downregulation of lung innate defense targets (SP-A, Muc5b) and mostly a summative effect when presented as co-exposure. However, the two toxicant types show differential induction of aquaporin-1 coinciding with the corresponding differential damage to alveolar integrity (vascular permeability). Interestingly, this implies the potential of AQP1 as a differential marker of toxicant type-specific lung injury.

Keywords: aquaporin, gene expression, lung injury, toxicant exposure

Procedia PDF Downloads 157

Authors: Dipo Dunsin, Mohamed C. Ghanem, Karim Ouazzane

Abstract:

Digital investigators often have a hard time spotting evidence in digital information. It has become hard to determine which source of proof relates to a specific investigation. A growing concern is that the various processes, technology, and specific procedures used in the digital investigation are not keeping up with criminal developments. Therefore, criminals are taking advantage of these weaknesses to commit further crimes. In digital forensics investigations, artificial intelligence is invaluable in identifying crime. It has been observed that an algorithm based on artificial intelligence (AI) is highly effective in detecting risks, preventing criminal activity, and forecasting illegal activity. Providing objective data and conducting an assessment is the goal of digital forensics and digital investigation, which will assist in developing a plausible theory that can be presented as evidence in court. Researchers and other authorities have used the available data as evidence in court to convict a person. This research paper aims at developing a multiagent framework for digital investigations using specific intelligent software agents (ISA). The agents communicate to address particular tasks jointly and keep the same objectives in mind during each task. The rules and knowledge contained within each agent are dependent on the investigation type. A criminal investigation is classified quickly and efficiently using the case-based reasoning (CBR) technique. The MADIK is implemented using the Java Agent Development Framework and implemented using Eclipse, Postgres repository, and a rule engine for agent reasoning. The proposed framework was tested using the Lone Wolf image files and datasets. Experiments were conducted using various sets of ISA and VMs. There was a significant reduction in the time taken for the Hash Set Agent to execute. As a result of loading the agents, 5 percent of the time was lost, as the File Path Agent prescribed deleting 1,510, while the Timeline Agent found multiple executable files. In comparison, the integrity check carried out on the Lone Wolf image file using a digital forensic tool kit took approximately 48 minutes (2,880 ms), whereas the MADIK framework accomplished this in 16 minutes (960 ms). The framework is integrated with Python, allowing for further integration of other digital forensic tools, such as AccessData Forensic Toolkit (FTK), Wireshark, Volatility, and Scapy.

Keywords: artificial intelligence, computer science, criminal investigation, digital forensics

Procedia PDF Downloads 188

Authors: Stephanie Chen, Mitja Echim, Christof Büskens

Abstract:

To describe and propagate the behavior of a system mathematical models are formulated. Parameter identification is used to adapt the coefficients of the underlying laws of science. For complex systems this approach can be incomplete and hence imprecise and moreover too slow to be computed efficiently. Therefore, these models might be not applicable for the numerical optimization of real systems, since these techniques require numerous evaluations of the models. Moreover not all quantities necessary for the identification might be available and hence the system must be adapted manually. Therefore, an approach is described that generates models that overcome the before mentioned limitations by not focusing on physical laws, but on measured (sensor) data of real systems. The approach is more general since it generates models for every system detached from the scientific background. Additionally, this approach can be used in a more general sense, since it is able to automatically identify correlations in the data. The method can be classified as a multivariate data regression analysis. In contrast to many other data regression methods this variant is also able to identify correlations of products of variables and not only of single variables. This enables a far more precise and better representation of causal correlations. The basis and the explanation of this method come from an analytical background: the series expansion. Another advantage of this technique is the possibility of real-time adaptation of the generated models during operation. Herewith system changes due to aging, wear or perturbations from the environment can be taken into account, which is indispensable for realistic scenarios. Since these data driven models can be evaluated very efficiently and with high precision, they can be used in mathematical optimization algorithms that minimize a cost function, e.g. time, energy consumption, operational costs or a mixture of them, subject to additional constraints. The proposed method has successfully been tested in several complex applications and with strong industrial requirements. The generated models were able to simulate the given systems with an error in precision less than one percent. Moreover the automatic identification of the correlations was able to discover so far unknown relationships. To summarize the above mentioned approach is able to efficiently compute high precise and real-time-adaptive data-based models in different fields of industry. Combined with an effective mathematical optimization algorithm like WORHP (We Optimize Really Huge Problems) several complex systems can now be represented by a high precision model to be optimized within the user wishes. The proposed methods will be illustrated with different examples.

Keywords: adaptive modeling, automatic identification of correlations, data based modeling, optimization

Procedia PDF Downloads 381

Authors: Eun-Taek Sin, Hyun-Ju Jang, Chang Geun Song, Yong-Sik Han

Abstract:

Due to the climate change, the coastal urban area repeatedly suffers from the loss of property and life by flooding. There are three main causes of inland submergence. First, when heavy rain with high intensity occurs, the water quantity in inland cannot be drained into rivers by increase in impervious surface of the land development and defect of the pump, storm sewer. Second, river inundation occurs then water surface level surpasses the top of levee. Finally, Coastal inundation occurs due to rising sea water. However, previous studies ignored the complex mechanism of flooding, and showed discrepancy and inadequacy due to linear summation of each analysis result. In this study, inland flooding and river inundation were analyzed together by HDM-2D model. Petrov-Galerkin stabilizing method and flux-blocking algorithm were applied to simulate the inland flooding. In addition, sink/source terms with exponentially growth rate attribute were added to the shallow water equations to include the inland flooding analysis module. The applications of developed model gave satisfactory results, and provided accurate prediction in comprehensive flooding analysis. The applications of developed model gave satisfactory results, and provided accurate prediction in comprehensive flooding analysis. To consider the coastal surge, another module was developed by adding seawater to the existing Inland Flooding-River Inundation binding module for comprehensive flooding analysis. Based on the combined modules, the Coastal Inundation – Inland & River Flow Interface was simulated by inputting the flow rate and depth data in artificial flume. Accordingly, it was able to analyze the flood patterns of coastal cities over time. This study is expected to help identify the complex causes of flooding in coastal areas where complex flooding occurs, and assist in analyzing damage to coastal cities. Acknowledgements—This research was supported by a grant ‘Development of the Evaluation Technology for Complex Causes of Inundation Vulnerability and the Response Plans in Coastal Urban Areas for Adaptation to Climate Change’ [MPSS-NH-2015-77] from the Natural Hazard Mitigation Research Group, Ministry of Public Safety and Security of Korea.

Keywords: flooding analysis, river inundation, inland flooding, 2D hydrodynamic model

Procedia PDF Downloads 337

Authors: Vinicius G. Goecks, Gregory M. Gremillion, William D. Nothwang

Abstract:

Modern approaches to train intelligent agents rely on prolonged training sessions, high amounts of input data, and multiple interactions with the environment. This restricts the application of these learning algorithms in robotics and real-world applications, in which there is low tolerance to inadequate actions, interactions are expensive, and real-time processing and action are required. This paper addresses this issue introducing CyberSteer, a novel approach to efficiently design intrinsic reward functions based on human intention to guide deep reinforcement learning agents with no environment-dependent rewards. CyberSteer uses non-expert human operators for initial demonstration of a given task or desired behavior. The trajectories collected are used to train a behavior cloning deep neural network that asynchronously runs in the background and suggests actions to the deep reinforcement learning module. An intrinsic reward is computed based on the similarity between actions suggested and taken by the deep reinforcement learning algorithm commanding the agent. This intrinsic reward can also be reshaped through additional human demonstration or critique. This approach removes the need for environment-dependent or hand-engineered rewards while still being able to safely shape the behavior of autonomous robotic agents, in this case, based on human intention. CyberSteer is tested in a high-fidelity unmanned aerial vehicle simulation environment, the Microsoft AirSim. The simulated aerial robot performs collision avoidance through a clustered forest environment using forward-looking depth sensing and roll, pitch, and yaw references angle commands to the flight controller. This approach shows that the behavior of robotic systems can be shaped in a reduced amount of time when guided by a non-expert human, who is only aware of the high-level goals of the task. Decreasing the amount of training time required and increasing safety during training maneuvers will allow for faster deployment of intelligent robotic agents in dynamic real-world applications.

Keywords: human-robot interaction, intelligent robots, robot learning, semisupervised learning, unmanned aerial vehicles

Procedia PDF Downloads 245

Authors: Atabaev I. G., Fayziev Sh. A., Irmatova Sh. K.

Abstract:

Raw materials with high content of Kalium oxide widely used in ceramic technology for prevention or decreasing of deformation of ceramic goods during drying process and under thermal annealing. Becouse to low melting temperature it is also used to decreasing of the temperature of thermal annealing during fabrication of ceramic goods [1,2]. So called “Porceline or China stones” - quartz-sericite (muscovite) minerals is also can be used for prevention of deformation as the content of Kalium oxide in muscovite is rather high (SiO2, + KAl2[AlSi3O10](OH)2). [3] . To estimation of possibility of use of this mineral for ceramic manufacture, in the presented article the influence of thermal processing on phase and a chemical content of this raw material is investigated. As well as to other ceramic raw materials (kaoline, white burning clays) the basic requirements of the industry to quality of "a porcelain stone» are following: small size of particles, relative high uniformity of disrtribution of components and phase, white color after burning, small content of colorant oxides or chromophores (Fe2O3, FeO, TiO2, etc) [4,5]. In the presented work natural minerale from the Boynaksay deposit (Uzbekistan) is investigated. The samples was mechanically polished for investigation by Scanning Electron Microscope. Powder with size of particle up to 63 μm was used to X-ray diffractometry and chemical analysis. The annealing of samples was performed at 900, 1120, 1350oC during 1 hour. Chemical composition of Boynaksay raw material according to chemical analysis presented in the table 1. For comparison the composition of raw materials from Russia and USA are also presented. In the Boynaksay quartz – sericite the average parity of quartz and sericite makes 55-60 and 30-35 % accordingly. The distribution of quartz and sericite phases in raw material was investigated using electron probe scanning electronic microscope «JEOL» JXA-8800R. In the figure 1 the scanning electron microscope (SEM) micrograps of the surface and the distributions of Al, Si and K atoms in the sample are presented. As it seen small granular, white and dense mineral includes quartz, sericite and small content of impurity minerals. Basically, crystals of quartz have the sizes from 80 up to 500 μm. Between quartz crystals the sericite inclusions having a tablet form with radiant structure are located. The size of sericite crystals is ~ 40-250 μm. Using data on interplanar distance [6,7] and ASTM Powder X-ray Diffraction Data it is shown that natural «a porcelain stone» quartz – sericite consists the quartz SiO2, sericite (muscovite type) KAl2[AlSi3O10](OH)2 and kaolinite Al203SiO22Н2О (See Figure 2 and Table 2). As it seen in the figure 3 and table 3a after annealing at 900oC the quartz – sericite contains quartz – SiO2 and muscovite - KAl2[AlSi3O10](OH)2, the peaks related with Kaolinite are absent. After annealing at 1120oC the full disintegration of muscovite and formation of mullite phase Al203 SiO2 is observed (the weak peaks of mullite appears in fig 3b and table 3b). After annealing at 1350oC the samples contains crystal phase of quartz and mullite (figure 3c and table 3с). Well known Mullite gives to ceramics high density, abrasive and chemical stability. Thus the obtained experimental data on formation of various phases during thermal annealing can be used for development of fabrication technology of advanced materials. Conclusion: The influence of thermal annealing in the interval 900-1350oC on phase composition and structure of quartz-sericite minerale is investigated. It is shown that during annealing the phase content of raw material is changed. After annealing at 1350oC the samples contains crystal phase of quartz and mullite (which gives gives to ceramics high density, abrasive and chemical stability).

Keywords: quartz-sericite, kaolinite, mullite, thermal processing

Procedia PDF Downloads 391

Authors: Kumar Shashvat, Amol P. Bhondekar

Abstract:

In the midst of the five senses, odor is the most reminiscent and least understood. Odor testing has been mysterious and odor data fabled to most practitioners. The delinquent of recognition and classification of odor is important to achieve. The facility to smell and predict whether the artifact is of further use or it has become undesirable for consumption; the imitation of this problem hooked on a model is of consideration. The general industrial standard for this classification is color based anyhow; odor can be improved classifier than color based classification and if incorporated in machine will be awfully constructive. For cataloging of odor for peas, trees and cashews various discriminative approaches have been used Discriminative approaches offer good prognostic performance and have been widely used in many applications but are incapable to make effectual use of the unlabeled information. In such scenarios, generative approaches have better applicability, as they are able to knob glitches, such as in set-ups where variability in the series of possible input vectors is enormous. Generative models are integrated in machine learning for either modeling data directly or as a transitional step to form an indeterminate probability density function. The algorithms or models Linear Discriminant Analysis and Naive Bayes Classifier have been used for classification of the odor of cashews. Linear Discriminant Analysis is a method used in data classification, pattern recognition, and machine learning to discover a linear combination of features that typifies or divides two or more classes of objects or procedures. The Naive Bayes algorithm is a classification approach base on Bayes rule and a set of qualified independence theory. Naive Bayes classifiers are highly scalable, requiring a number of restraints linear in the number of variables (features/predictors) in a learning predicament. The main recompenses of using the generative models are generally a Generative Models make stronger assumptions about the data, specifically, about the distribution of predictors given the response variables. The Electronic instrument which is used for artificial odor sensing and classification is an electronic nose. This device is designed to imitate the anthropological sense of odor by providing an analysis of individual chemicals or chemical mixtures. The experimental results have been evaluated in the form of the performance measures i.e. are accuracy, precision and recall. The investigational results have proven that the overall performance of the Linear Discriminant Analysis was better in assessment to the Naive Bayes Classifier on cashew dataset.

Keywords: odor classification, generative models, naive bayes, linear discriminant analysis

Procedia PDF Downloads 358

Authors: Mohammad Ghahramani, Fahimeh Saei Manesh

Abstract:

Winning a soccer game is based on thorough and deep analysis of the ongoing match. On the other hand, giant gambling companies are in vital need of such analysis to reduce their loss against their customers. In this research work, we perform deep, real-time analysis on every soccer match around the world that distinguishes our work from others by focusing on particular seasons, teams and partial analytics. Our contributions are presented in the platform called “Analyst Masters.” First, we introduce various sources of information available for soccer analysis for teams around the world that helped us record live statistical data and information from more than 50,000 soccer matches a year. Our second and main contribution is to introduce our proposed in-play performance evaluation. The third contribution is developing new features from stable soccer matches. The statistics of soccer matches and their odds before and in-play are considered in the image format versus time including the halftime. Local Binary patterns, (LBP) is then employed to extract features from the image. Our analyses reveal incredibly interesting features and rules if a soccer match has reached enough stability. For example, our “8-minute rule” implies if 'Team A' scores a goal and can maintain the result for at least 8 minutes then the match would end in their favor in a stable match. We could also make accurate predictions before the match of scoring less/more than 2.5 goals. We benefit from the Gradient Boosting Trees, GBT, to extract highly related features. Once the features are selected from this pool of data, the Decision trees decide if the match is stable. A stable match is then passed to a post-processing stage to check its properties such as betters’ and punters’ behavior and its statistical data to issue the prediction. The proposed method was trained using 140,000 soccer matches and tested on more than 100,000 samples achieving 98% accuracy to select stable matches. Our database from 240,000 matches shows that one can get over 20% betting profit per month using Analyst Masters. Such consistent profit outperforms human experts and shows the inefficiency of the betting market. Top soccer tipsters achieve 50% accuracy and 8% monthly profit in average only on regional matches. Both our collected database of more than 240,000 soccer matches from 2012 and our algorithm would greatly benefit coaches and punters to get accurate analysis.

Keywords: soccer, analytics, machine learning, database

Procedia PDF Downloads 215

Authors: Alvaro Figueira, Bruno Cabral

Abstract:

Nowadays, students are used to be assessed through an online platform. Educators have stepped up from a period in which they endured the transition from paper to digital. The use of a diversified set of question types that range from quizzes to open questions is currently common in most university courses. In many courses, today, the evaluation methodology also fosters the students’ online participation in forums, the download, and upload of modified files, or even the participation in group activities. At the same time, new pedagogy theories that promote the active participation of students in the learning process, and the systematic use of problem-based learning, are being adopted using an eLearning system for that purpose. However, although there can be a lot of feedback from these activities to student’s, usually it is restricted to the assessments of online well-defined tasks. In this article, we propose an automatic system that informs students of abnormal deviations of a 'correct' learning path in the course. Our approach is based on the fact that by obtaining this information earlier in the semester, may provide students and educators an opportunity to resolve an eventual problem regarding the student’s current online actions towards the course. Our goal is to prevent situations that have a significant probability to lead to a poor grade and, eventually, to failing. In the major learning management systems (LMS) currently available, the interaction between the students and the system itself is registered in log files in the form of registers that mark beginning of actions performed by the user. Our proposed system uses that logged information to derive new one: the time each student spends on each activity, the time and order of the resources used by the student and, finally, the online resource usage pattern. Then, using the grades assigned to the students in previous years, we built a learning dataset that is used to feed a machine learning meta classifier. The produced classification model is then used to predict the grades a learning path is heading to, in the current year. Not only this approach serves the teacher, but also the student to receive automatic feedback on her current situation, having past years as a perspective. Our system can be applied to online courses that integrate the use of an online platform that stores user actions in a log file, and that has access to other student’s evaluations. The system is based on a data mining process on the log files and on a self-feedback machine learning algorithm that works paired with the Moodle LMS.

Keywords: data mining, e-learning, grade prediction, machine learning, student learning path

Procedia PDF Downloads 104

Authors: Utkarsh A. Mishra, Ankit Bansal

Abstract:

At high temperature, radiative heat transfer is the dominant mode of heat transfer. It is governed by various phenomena such as photon emission, absorption, and scattering. The solution of the governing integrodifferential equation of radiative transfer is a complex process, more when the effect of participating medium and wavelength properties are taken into consideration. Although a generic formulation of such radiative transport problem can be modeled for a wide variety of problems with non-gray, non-diffusive surfaces, there is always a trade-off between simplicity and accuracy of the problem. Recently, solutions of complicated mathematical problems with statistical methods based on randomization of naturally occurring phenomena have gained significant importance. Photon bundles with discrete energy can be replicated with random numbers describing the emission, absorption, and scattering processes. Photon Monte Carlo (PMC) is a simple, yet powerful technique, to solve radiative transfer problems in complicated geometries with arbitrary participating medium. The method, on the one hand, increases the accuracy of estimation, and on the other hand, increases the computational cost. The participating media -generally a gas, such as CO₂, CO, and H₂O- present complex emission and absorption spectra. To model the emission/absorption accurately with random numbers requires a weighted sampling as different sections of the spectrum carries different importance. Importance sampling (IS) was implemented to sample random photon of arbitrary wavelength, and the sampled data provided unbiased training of MC estimators for better results. A better replacement to uniform random numbers is using deterministic, quasi-random sequences. Halton, Sobol, and Faure Low-Discrepancy Sequences are used in this study. They possess better space-filling performance than the uniform random number generator and gives rise to a low variance, stable Quasi-Monte Carlo (QMC) estimators with faster convergence. An optimal supervised learning scheme was further considered to reduce the computation costs of the PMC simulation. A one-dimensional plane-parallel slab problem with participating media was formulated. The history of some randomly sampled photon bundles is recorded to train an Artificial Neural Network (ANN), back-propagation model. The flux was calculated using the standard quasi PMC and was considered to be the training target. Results obtained with the proposed model for the one-dimensional problem are compared with the exact analytical and PMC model with the Line by Line (LBL) spectral model. The approximate variance obtained was around 3.14%. Results were analyzed with respect to time and the total flux in both cases. A significant reduction in variance as well a faster rate of convergence was observed in the case of the QMC method over the standard PMC method. However, the results obtained with the ANN method resulted in greater variance (around 25-28%) as compared to the other cases. There is a great scope of machine learning models to help in further reduction of computation cost once trained successfully. Multiple ways of selecting the input data as well as various architectures will be tried such that the concerned environment can be fully addressed to the ANN model. Better results can be achieved in this unexplored domain.

Keywords: radiative heat transfer, Monte Carlo Method, pseudo-random numbers, low discrepancy sequences, artificial neural networks

Procedia PDF Downloads 200

Authors: Mariya V. Yanyukina, Michael A. Bolotov

Abstract:

Assembling accuracy is the degree of accordance between the actual values of the parameters obtained during assembly, and the values specified in the assembly drawings and technical specifications. However, the assembling accuracy depends not only on the quality of the production process but also on the correctness of the assembly process. Therefore, preliminary calculations of assembly stages are carried out to verify the correspondence of real geometric parameters to their acceptable values. In the aviation industry, most calculations involve interacting dimensional chains. This greatly complicates the task. Solving such problems requires a special approach. The purpose of this article is to carry out the problem of improving the technology of assembly of aviation units by use of computer calculations. One of the actual examples of the assembly unit, in which there is an interacting dimensional chain, is the turbine wheel of gas turbine engine. Dimensional chain of turbine wheel is formed by geometric parameters of disk and set of blades. The interaction of the dimensional chain consists in the formation of two chains. The first chain is formed by the dimensions that determine the location of the grooves for the installation of the blades, and the dimensions of the blade roots. The second dimensional chain is formed by the dimensions of the airfoil shroud platform. The interaction of the dimensional chain of the turbine wheel is the interdependence of the first and second chains by means of power circuits formed by a plurality of middle parts of the turbine blades. The timeliness of the calculation of the dimensional chain of the turbine wheel is the need to improve the technology of assembly of this unit. The task at hand contains geometric and mathematical components; therefore, its solution can be implemented following the algorithm: 1) research and analysis of production errors by geometric parameters; 2) development of a parametric model in the CAD system; 3) creation of set of CAD-models of details taking into account actual or generalized distributions of errors of geometrical parameters; 4) calculation model in the CAE-system, loading of various combinations of models of parts; 5) the accumulation of statistics and analysis. The main task is to pre-simulate the assembly process by calculating the interacting dimensional chains. The article describes the approach to the solution from the point of view of mathematical statistics, implemented in the software package Matlab. Within the framework of the study, there are data on the measurement of the components of the turbine wheel-blades and disks, as a result of which it is expected that the assembly process of the unit will be optimized by solving dimensional chains.

Keywords: accuracy, assembly, interacting dimension chains, turbine

Procedia PDF Downloads 355

Authors: Maris Eugênia Dela Rosa, Ana Luiza Menegatti Pavan, Marcela De Oliveira, Diana Rodrigues De Pina, Luis Carlos Vulcano

Abstract:

In medical veterinary as well as in human medicine, radiological study is essential for a safe diagnosis in clinical practice. Thus, the quality of radiographic image is crucial. In last year’s there has been an increasing substitution of image acquisition screen-film systems for computed radiology equipment (CR) without technical charts adequacy. Furthermore, to carry out a radiographic examination in veterinary patient is required human assistance for restraint this, which can compromise image quality by generating dose increasing to the animal, for Occupationally Exposed and also the increased cost to the institution. The image optimization procedure and construction of radiographic techniques are performed with the use of homogeneous phantoms. In this study, we sought to develop a homogeneous phantom of canine chest to be applied to the optimization of these images for the CR system. In carrying out the simulator was created a database with retrospectives chest images of computed tomography (CT) of the Veterinary Hospital of the Faculty of Veterinary Medicine and Animal Science - UNESP (FMVZ / Botucatu). Images were divided into four groups according to the animal weight employing classification by sizes proposed by Hoskins & Goldston. The thickness of biological tissues were quantified in a 80 animals, separated in groups of 20 animals according to their weights: (S) Small - equal to or less than 9.0 kg, (M) Medium - between 9.0 and 23.0 kg, (L) Large – between 23.1 and 40.0kg and (G) Giant – over 40.1 kg. Mean weight for group (S) was 6.5±2.0 kg, (M) 15.0±5.0 kg, (L) 32.0±5.5 kg and (G) 50.0 ±12.0 kg. An algorithm was developed in Matlab in order to classify and quantify biological tissues present in CT images and convert them in simulator materials. To classify tissues presents, the membership functions were created from the retrospective CT scans according to the type of tissue (adipose, muscle, bone trabecular or cortical and lung tissue). After conversion of the biologic tissue thickness in equivalent material thicknesses (acrylic simulating soft tissues, bone tissues simulated by aluminum and air to the lung) were obtained four different homogeneous phantoms, with (S) 5 cm of acrylic, 0,14 cm of aluminum and 1,8 cm of air; (M) 8,7 cm of acrylic, 0,2 cm of aluminum and 2,4 cm of air; (L) 10,6 cm of acrylic, 0,27 cm of aluminum and 3,1 cm of air and (G) 14,8 cm of acrylic, 0,33 cm of aluminum and 3,8 cm of air. The developed canine homogeneous phantom is a practical tool, which will be employed in future, works to optimize veterinary X-ray procedures.

Keywords: radiation protection, phantom, veterinary radiology, computed radiography

Procedia PDF Downloads 400

Authors: Sunghun Jung, Wonkook Kim

Abstract:

Due to the difficulty of the utilization of satellite and an aircraft, conventional ocean color remote sensing has a disadvantage in that it is difficult to obtain images of desired places at desired times. These disadvantages make it difficult to capture the anomalies such as the occurrence of the red tide which requires immediate observation. It is also difficult to understand the phenomena such as the resuspension-precipitation process of suspended solids and the spread of low-salinity water originating in the coastal areas. For the remote sensing reflectance of seawater, above-water radiometers (AWR) have been used either by carrying portable AWRs on a ship or installing those at fixed observation points on the Ieodo ocean research station, Socheongcho base, and etc. In particular, however, it requires the high cost to measure the remote reflectance in various seawater environments at various times and it is even not possible to measure it at the desired frequency in the desired sea area at the desired time. Also, in case of the stationary observation, it is advantageous that observation data is continuously obtained, but there is the disadvantage that data of various sea areas cannot be obtained. It is possible to instantly capture various marine phenomena occurring on the coast using the unmanned aerial system (UAS) including vertical takeoff and landing (VTOL) type unmanned aerial vehicles (UAV) since it could move and hover at the one location and acquire data of the desired form at a high resolution. To remotely estimate seawater constituents, it is necessary to install an ultra-spectral sensor. Also, to calculate reflected light from the surface of the sea in consideration of the sun’s incident light, a total of three sensors need to be installed on the UAV. The remote sensing reflectance of seawater is the most basic optical property for remotely estimating color components in seawater and we could remotely estimate the chlorophyll concentration, the suspended solids concentration, and the dissolved organic amount. Estimating seawater physics from the remote sensing reflectance requires the algorithm development using the accumulation data of seawater reflectivity under various seawater and atmospheric conditions. The UAS with three AWRs is developed for the remote reflection sensing on the surface of the sea. Throughout the paper, we explain the details of each UAS component, system operation scenarios, and simulation and experiment results. The UAS consists of a UAV, a solar tracker, a transmitter, a ground control station (GCS), three AWRs, and two gimbals.

Keywords: above-water radiometers (AWR), ground control station (GCS), unmanned aerial system (UAS), unmanned aerial vehicle (UAV)

Procedia PDF Downloads 144

Authors: Garry Gorman, Nigel McKelvey, James Connolly

Abstract:

This study will evaluate a growth mindset intervention with Junior Cycle Coding and Senior Cycle Computer Science students in Ireland, where gamification will be used to incentivise growth mindset behaviour. An artificial intelligence (AI) driven personalised learning system will be developed to present computer programming learning tasks in a manner that is best suited to the individuals’ own learning preferences while incentivising and rewarding growth mindset behaviour of persistence, mastery response to challenge, and challenge seeking. This research endeavours to measure mindset with before and after surveys (conducted nationally) and by recording growth mindset behaviour whilst playing a digital game. This study will harness the capabilities of AI and aims to determine how a personalised learning (PL) experience can impact the mindset of a broad range of students. The focus of this study will be to determine how personalising the learning experience influences female and disadvantaged students' sense of belonging in the computer science classroom when tasks are presented in a manner that is best suited to the individual. Whole Brain Learning will underpin this research and will be used as a framework to guide the research in identifying key areas such as thinking and learning styles, cognitive potential, motivators and fears, and emotional intelligence. This research will be conducted in multiple school types over one academic year. Digital games will be played multiple times over this period, and the data gathered will be used to inform the AI algorithm. The three data sets are described as follows: (i) Before and after survey data to determine the grit scores and mindsets of the participants, (ii) The Growth Mind-Set data from the game, which will measure multiple growth mindset behaviours, such as persistence, response to challenge and use of strategy, (iii) The AI data to guide PL. This study will highlight the effectiveness of an AI-driven personalised learning experience. The data will position AI within the Irish educational landscape, with a specific focus on the teaching of CS. These findings will benefit coding and computer science teachers by providing a clear pedagogy for the effective delivery of personalised learning strategies for computer science education. This pedagogy will help prevent students from developing a fixed mindset while helping pupils to exhibit persistence of effort, use of strategy, and a mastery response to challenges.

Keywords: computer science education, artificial intelligence, growth mindset, pedagogy

Procedia PDF Downloads 71

Authors: Dhekra Khazri, Hakim Gabtni

Abstract:

Because of Sidi Bouzid water table overexploitation, this study aims at integrating geophysical methods to determinate aquifers geometry assessing their geological situation and geophysical characteristics. However in highly tectonic zones controlled by Atlassic structural features with NE-SW major directions (central Tunisia), Bouguer gravimetric responses of some areas can be as much dominated by the regional structural tendency, as being non-identified or either defectively interpreted such as the case of Sidi Bouzid basin. This issue required a residual gravity anomaly elaboration isolating the Sidi Bouzid basin gravity response ranging between -8 and -14 mGal and crucial for its aquifers geometry characterization. Several gravity techniques helped constructing the Sidi Bouzid basin's residual gravity anomaly, such as Upwards continuation compared to polynomial regression trends and power spectrum analysis detecting deep basement sources at (3km), intermediate (2km) and shallow sources (1km). A 3D Euler Deconvolution was also performed detecting deepest accidents trending NE-SW, N-S and E-W with depth values reaching 5500 m and delineating the main outcropping structures of the study area. Further gravity treatments highlighted the subsurface geometry and structural features of Sidi Bouzid basin over Horizontal and vertical gradient, and also filters based on them such as Tilt angle and Source Edge detector locating rooted edges or peaks from potential field data detecting a new E-W lineament compartmentalizing the Sidi Bouzid gutter into two unequally residual anomaly and subsiding domains. This subsurface morphology is also detected by the used 2D seismic reflection sections defining the Sidi Bouzid basin as a deep gutter within a tectonic set of negative flower structures, and collapsed and tilted blocks. Furthermore, these structural features were confirmed by forward gravity modeling process over several modeled residual gravity profiles crossing the main area. Sidi Bouzid basin (central Tunisia) is also of a big interest cause of the unknown total thickness and the undefined substratum of its siliciclastic Tertiary package, and its aquifers unbounded structural subsurface features and deep accidents. The Combination of geological, hydrogeological and geophysical methods is then of an ultimate need. Therefore, a geophysical methods integration based on gravity survey supporting available seismic data through forward gravity modeling, enhanced lateral and vertical extent definition of the basin's complex sedimentary fill via 3D gravity models, improved depth estimation by a depth to basement modeling approach, and provided 3D isochronous seismic mapping visualization of the basin's Tertiary complex refining its geostructural schema. A subsurface basin geomorphology mapping, over an ultimate matching between the basin's residual gravity map and the calculated theoretical signature map, was also displayed over the modeled residual gravity profiles. An ultimate multidisciplinary geophysical study of the Sidi Bouzid basin aquifers can be accomplished via an aeromagnetic survey and a 4D Microgravity reservoir monitoring offering temporal tracking of the target aquifer's subsurface fluid dynamics enhancing and rationalizing future groundwater exploitation in this arid area of central Tunisia.

Keywords: aquifer geometry, geophysics, 3D gravity modeling, improved depths, source edge detector

Procedia PDF Downloads 262

Authors: Imed Feki, Faouzi Msahli

Abstract:

Selection of relevant parameters from a high dimensional process operation setting space is a problem frequently encountered in industrial process modelling. This paper presents a method for selecting the most relevant fabric physical parameters for each sensory quality feature. The proposed relevancy criterion has been developed using two approaches. The first utilizes a fuzzy sensitivity criterion by exploiting from experimental data the relationship between physical parameters and all the sensory quality features for each evaluator. Next an OWA aggregation procedure is applied to aggregate the ranking lists provided by different evaluators. In the second approach, another panel of experts provides their ranking lists of physical features according to their professional knowledge. Also by applying OWA and a fuzzy aggregation model, the data sensitivity-based ranking list and the knowledge-based ranking list are combined using our proposed percolation technique, to determine the final ranking list. The key issue of the proposed percolation technique is to filter automatically and objectively the relevant features by creating a gap between scores of relevant and irrelevant parameters. It permits to automatically generate threshold that can effectively reduce human subjectivity and arbitrariness when manually choosing thresholds. For a specific sensory descriptor, the threshold is defined systematically by iteratively aggregating (n times) the ranking lists generated by OWA and fuzzy models, according to a specific algorithm. Having applied the percolation technique on a real example, of a well known finished textile product especially the stonewashed denims, usually considered as the most important quality criteria in jeans’ evaluation, we separate the relevant physical features from irrelevant ones for each sensory descriptor. The originality and performance of the proposed relevant feature selection method can be shown by the variability in the number of physical features in the set of selected relevant parameters. Instead of selecting identical numbers of features with a predefined threshold, the proposed method can be adapted to the specific natures of the complex relations between sensory descriptors and physical features, in order to propose lists of relevant features of different sizes for different descriptors. In order to obtain more reliable results for selection of relevant physical features, the percolation technique has been applied for combining the fuzzy global relevancy and OWA global relevancy criteria in order to clearly distinguish scores of the relevant physical features from those of irrelevant ones.

Keywords: data sensitivity, feature selection, fuzzy logic, OWA operators, percolation technique

Procedia PDF Downloads 582

Authors: Ghada Abo-Zaid

Abstract:

The main objective of this study was to examine the association between the elevated level of CRP and incidence of hypertension before and after adjusting by age, BMI, gender, SES, smoking, diabetes, cholesterol LDL and cholesterol HDL and to determine whether the association were differ by race. Method: Cross sectional data for participations from age 17 to age 74 years who included in The National Health and Nutrition Examination Survey (NHANES) from 1999 to 2010 were analysed. CRP level was classified into three categories ( > 3mg/L, between 1mg/LL and 3mg/L, and < 3 mg/L). Blood pressure categorization was done using JNC 7 algorithm Hypertension defined as either systolic blood pressure (SBP) of 140 mmHg or more and disystolic blood pressure (DBP) of 90mmHg or greater, otherwise a self-reported prior diagnosis by a physician. Pre-hypertension was defined as (139 > SBP > 120 or 89 > DPB > 80). Multinominal regression model was undertaken to measure the association between CRP level and hypertension. Results: In univariable models, CRP concentrations > 3 mg/L were associated with a 73% greater risk of incident hypertension compared with CRP concentrations < 1 mg/L (Hypertension: odds ratio [OR] = 1.73; 95% confidence interval [CI], 1.50-1.99). Ethnic comparisons showed that American Mexican had the highest risk of incident hypertension (odds ratio [OR] = 2.39; 95% confidence interval [CI], 2.21-2.58).This risk was statistically insignificant, however, either after controlling by other variables (Hypertension: OR = 0.75; 95% CI, 0.52-1.08,), or categorized by race [American Mexican: odds ratio [OR] = 1.58; 95% confidence interval [CI], 0,58-4.26, Other Hispanic: odds ratio [OR] = 0.87; 95% confidence interval [CI], 0.19-4.42, Non-Hispanic white: odds ratio [OR] = 0.90; 95% confidence interval [CI], 0.50-1.59, Non-Hispanic Black: odds ratio [OR] = 0.44; 95% confidence interval [CI], 0.22-0,87]. The same results were found for pre-hypertension, and the Non-Hispanic black showed the highest significant risk for Pre-Hypertension (odds ratio [OR] = 1.60; 95% confidence interval [CI], 1.26-2.03). When CRP concentrations were between 1.0-3.0 mg/L, in an unadjusted models prehypertension was associated with higher likelihood of elevated CRP (OR = 1.37; 95% CI, 1.15-1.62). The same relationship was maintained in Non-Hispanic white, Non-Hispanic black, and other race (Non-Hispanic white: OR = 1.24; 95% CI, 1.03-1.48, Non-Hispanic black: OR = 1.60; 95% CI, 1.27-2.03, other race: OR = 2.50; 95% CI, 1.32-4.74) while the association was insignificant with American Mexican and other Hispanic. In the adjusted model, the relationship between CRP and prehypertension were no longer available. In contrary, Hypertension was not independently associated with elevated CRP, and the results were the same after grouped by race or adjusted by the confounder variables. The same results were obtained when SBP or DBP were on a continuous measure. Conclusions: This study confirmed the existence of an association between hypertension, prehypertension and elevated level of CRP, however this association was no longer available after adjusting by other variables. Ethic group differences were statistically significant at the univariable models, while it disappeared after controlling by other variables.

Keywords: CRP, hypertension, ethnicity, NHANES, blood pressure

Procedia PDF Downloads 392

Authors: Gulfam Haider, sana danish

Abstract:

Convolutional Neural Networks (CNNs) have emerged as powerful tools for image classification, and the choice of optimizers profoundly affects their performance. The study of optimizers and their adaptations remains a topic of significant importance in machine learning research. While numerous studies have explored and advocated for various optimizers, the efficacy of these optimization techniques is still subject to scrutiny. This work aims to address the challenges surrounding the effectiveness of optimizers by conducting a comprehensive analysis and evaluation. The primary focus of this investigation lies in examining the performance of different optimizers when employed in conjunction with the popular activation function, Rectified Linear Unit (ReLU). By incorporating ReLU, known for its favorable properties in prior research, the aim is to bolster the effectiveness of the optimizers under scrutiny. Specifically, we evaluate the adjustment of these optimizers with both the original Softmax activation function and the modified ReLU activation function, carefully assessing their impact on overall performance. To achieve this, a series of experiments are conducted using a well-established benchmark dataset for image classification tasks, namely the Canadian Institute for Advanced Research dataset (CIFAR-10). The selected optimizers for investigation encompass a range of prominent algorithms, including Adam, Root Mean Squared Propagation (RMSprop), Adaptive Learning Rate Method (Adadelta), Adaptive Gradient Algorithm (Adagrad), and Stochastic Gradient Descent (SGD). The performance analysis encompasses a comprehensive evaluation of the classification accuracy, convergence speed, and robustness of the CNN models trained with each optimizer. Through rigorous experimentation and meticulous assessment, we discern the strengths and weaknesses of the different optimization techniques, providing valuable insights into their suitability for image classification tasks. By conducting this in-depth study, we contribute to the existing body of knowledge surrounding optimizers in CNNs, shedding light on their performance characteristics for image classification. The findings gleaned from this research serve to guide researchers and practitioners in making informed decisions when selecting optimizers and activation functions, thus advancing the state-of-the-art in the field of image classification with convolutional neural networks.

Keywords: deep neural network, optimizers, RMsprop, ReLU, stochastic gradient descent

Procedia PDF Downloads 94

Authors: Damtew Samson Zerihun

Abstract:

This paper primarily describes the concept of a hierarchical safety control (HSC) in discrete manufacturing to up-hold productivity with human intervention and machine failures using a systematic approach, through increasing the system availability and using additional knowledge on machines so as to improve the human machine collaboration (HMC). It also highlights the implemented PLC safety algorithm, in applying this generic concept to a concrete pro-duction line using a lab demonstrator called FATIE (Factory Automation Test and Integration Environment). Furthermore, the paper describes a model and provide a systematic representation of human-machine collabora-tion in discrete manufacturing and to this end, the Hierarchical Safety Control concept is proposed. This offers a ge-neric description of human-machine collaboration based on Finite State Machines (FSM) that can be applied to vari-ous discrete manufacturing lines instead of using ad-hoc solutions for each line. With its reusability, flexibility, and extendibility, the Hierarchical Safety Control scheme allows upholding productivity while maintaining safety with reduced engineering effort compared to existing solutions. The approach to the solution begins with a successful partitioning of different zones around the Integrated Manufacturing System (IMS), which are defined by operator tasks and the risk assessment, used to describe the location of the human operator and thus to identify the related po-tential hazards and trigger the corresponding safety functions to mitigate it. This includes selective reduced speed zones and stop zones, and in addition with the hierarchical safety control scheme and advanced safety functions such as safe standstill and safe reduced speed are used to achieve the main goals in improving the safe Human Ma-chine Collaboration and increasing the productivity. In a sample scenarios, It is shown that an increase of productivity in the order of 2.5% is already possible with a hi-erarchical safety control, which consequently under a given assumptions, a total sum of 213 € could be saved for each intervention, compared to a protective stop reaction. Thereby the loss is reduced by 22.8%, if occasional haz-ard can be refined in a hierarchical way. Furthermore, production downtime due to temporary unavailability of safety devices can be avoided with safety failover that can save millions per year. Moreover, the paper highlights the proof of the development, implementation and application of the concept on the lab demonstrator (FATIE), where it is realized on the new safety PLCs, Drive Units, HMI as well as Safety devices in addition to the main components of the IMS.

Keywords: discrete automation, hierarchical safety controller, human machine collaboration, programmable logical controller

Procedia PDF Downloads 352

Authors: Sakshi Gupta, Anupam Agrawal, Ekta Singla

Abstract:

An exoskeleton design for rehabilitation purpose encounters many challenges, including ergonomically acceptable wearing technology, architectural design human-motion compatibility, actuation type, human-robot interaction, etc. In this paper, a passive architecture for upper limb exoskeleton is proposed for assisting in rehabilitation tasks. Kinematic modelling is detailed for task-based kinematic synthesis of the wearable exoskeleton for self-feeding tasks. The exoskeleton architecture possesses expansion and torsional springs which are able to store and redistribute energy over the human arm joints. The elastic characteristics of the springs have been optimized to minimize the mechanical work of the human arm joints. The concept of hybrid combination of a 4-bar parallelogram linkage and a serial linkage were chosen, where the 4-bar parallelogram linkage with expansion spring acts as a rigid structure which is used to provide the rotational degree-of-freedom (DOF) required for lowering and raising of the arm. The single linkage with torsional spring allows for the rotational DOF required for elbow movement. The focus of the paper is kinematic modelling, analysis and task-based synthesis framework for the proposed architecture, keeping in considerations the essential tasks of self-feeding and self-exercising during rehabilitation of partially healthy person. Rehabilitation of primary functional movements (activities of daily life, i.e., ADL) is routine activities that people tend to every day such as cleaning, dressing, feeding. We are focusing on the feeding process to make people independent in respect of the feeding tasks. The tasks are focused to post-surgery patients under rehabilitation with less than 40% weakness. The challenges addressed in work are ensuring to emulate the natural movement of the human arm. Human motion data is extracted through motion-sensors for targeted tasks of feeding and specific exercises. Task-based synthesis procedure framework will be discussed for the proposed architecture. The results include the simulation of the architectural concept for tracking the human-arm movements while displaying the kinematic and static study parameters for standard human weight. D-H parameters are used for kinematic modelling of the hybrid-mechanism, and the model is used while performing task-based optimal synthesis utilizing evolutionary algorithm.

Keywords: passive mechanism, task-based synthesis, emulating human-motion, exoskeleton

Procedia PDF Downloads 119

Authors: Justin Fu, Thomas Mazzuchi, Shahram Sarkani

Abstract:

A key factor in technological immaturity in defense weapons acquisition is lack of understanding critical integrations at the subsystem and component level. To address this shortfall, recent research in integration readiness level (IRL) combines with technology readiness level (TRL) to form a system readiness level (SRL). SRL can be enriched with more robust quantitative methods to provide the program manager a useful tool prior to committing to major weapons acquisition programs. This research harnesses previous mathematical models based on graph theory, Petri nets, and tropical algebra and proposes a modification of the desirable SRL mathematical properties such that a tightly integrated (multitude of interfaces) subsystem can display a lower SRL than an inherently less coupled subsystem. The synthesis of these methods informs an improved decision tool for the program manager to commit to expensive technology development. This research ties the separately developed manufacturing readiness level (MRL) into the network representation of the system and addresses shortfalls in previous frameworks, including the lack of integration weighting and the over-importance of a single extremely immature component. Tropical algebra (based on the minimum of a set of TRLs or IRLs) allows one low IRL or TRL value to diminish the SRL of the entire system, which may not be reflective of actuality if that component is not critical or tightly coupled. Integration connections can be weighted according to importance and readiness levels are modified to be a cardinal scale (based on an analytic hierarchy process). Integration arcs’ importance are dependent on the connected nodes and the additional integrations arcs connected to those nodes. Lack of integration is not represented by zero, but by a perfect integration maturity value. Naturally, the importance (or weight) of such an arc would be zero. To further explore the impact of grouping subsystems, a multi-objective genetic algorithm is then used to find various clusters or communities that can be optimized for the most representative subsystem SRL. This novel calculation is then benchmarked through simulation and using past defense acquisition program data, focusing on the newly introduced Middle Tier of Acquisition (rapidly field prototypes). The model remains a relatively simple, accessible tool, but at higher fidelity and validated with past data for the program manager to decide major defense acquisition program milestones.

Keywords: readiness, maturity, system, integration

Procedia PDF Downloads 66

Authors: Changdong Oh

Abstract:

With the arrival of post-industrial society, social scientists have been giving attention to issues of which factors shape cleavage of political parties. Especially, there is a heated controversy over whether and how social and cultural values influence the identities of parties and voting behavior. Drawing from Moral Foundations Theory (MFT), which approached similar issues by considering the effect of five moral foundations on political decision-making of people, this study investigates the role of moral rhetoric in the evolution of Korean political parties. Researcher collected official announcements released by the major two parties (Democratic Party of Korea, Saenuri Party) from 2007 to 2016, and analyzed the data by using Word2Vec algorithm and Moral Foundations Dictionary. Five moral decision modules of MFT, composed of care, fairness (individualistic morality), loyalty, authority and sanctity (group-based, Durkheimian morality), can be represented in vector spaces consisted of party announcements data. By comparing the party vector and the five morality vectors, researcher can see how the political parties have actively used each of the five moral foundations to express themselves and the opposition. Results report that the conservative party tends to actively draw on collective morality such as loyalty, authority, purity to differentiate itself. Notably, such moral differentiation strategy is prevalent when they criticize an opposition party. In contrast, the liberal party tends to concern with individualistic morality such as fairness. This result indicates that moral cleavage does exist between parties in South Korea. Furthermore, individualistic moral gaps of the two political parties are eased over time, which seems to be due to the discussion of economic democratization of conservative party that emerged after 2012, but the community-related moral gaps widened. These results imply that past political cleavages related to economic interests are diminishing and replaced by cultural and social values associated with communitarian morality. However, since the conservative party’s differentiation strategy is largely related to negative campaigns, it is doubtful whether such moral differentiation among political parties can contribute to the long-term party identification of the voters, thus further research is needed to determine it is sustainable. Despite the limitations, this study makes it possible to track and identify the moral changes of party system through automated text analysis. More generally, this study could contribute to the analysis of various texts associated with the moral foundation and finding a distributed representation of moral, ethical values.

Keywords: moral foundations theory, moral politics, party system, Word2Vec

Procedia PDF Downloads 348

Authors: Joanna Grzybowska, Magdalena Igras, Mariusz Ziółko

Abstract:

Association rule learning is an approach for discovering interesting relationships in large databases. The analysis of relations, invisible at first glance, is a source of new knowledge which can be subsequently used for prediction. We used this data mining technique (which is an automatic and objective method) to learn about interesting affects associations in a corpus of emergency phone calls. We also made an attempt to match revealed rules with their possible situational context. The corpus was collected and subjectively annotated by two researchers. Each of 3306 recordings contains information on emotion: (1) type (sadness, weariness, anxiety, surprise, stress, anger, frustration, calm, relief, compassion, contentment, amusement, joy) (2) valence (negative, neutral, or positive) (3) intensity (low, typical, alternating, high). Also, additional information, that is a clue to speaker’s emotional state, was annotated: speech rate (slow, normal, fast), characteristic vocabulary (filled pauses, repeated words) and conversation style (normal, chaotic). Exponentially many rules can be extracted from a set of items (an item is a previously annotated single information). To generate the rules in the form of an implication X → Y (where X and Y are frequent k-itemsets) the Apriori algorithm was used - it avoids performing needless computations. Then, two basic measures (Support and Confidence) and several additional symmetric and asymmetric objective measures (e.g. Laplace, Conviction, Interest Factor, Cosine, correlation coefficient) were calculated for each rule. Each applied interestingness measure revealed different rules - we selected some top rules for each measure. Owing to the specificity of the corpus (emergency situations), most of the strong rules contain only negative emotions. There are though strong rules including neutral or even positive emotions. Three examples of the strongest rules are: {sadness} → {anxiety}; {sadness, weariness, stress, frustration} → {anger}; {compassion} → {sadness}. Association rule learning revealed the strongest configurations of affects (as well as configurations of affects with affect-related information) in our emergency phone calls corpus. The acquired knowledge can be used for prediction to fulfill the emotional profile of a new caller. Furthermore, a rule-related possible context analysis may be a clue to the situation a caller is in.

Keywords: data mining, emergency phone calls, emotional profiles, rules

Procedia PDF Downloads 392

Authors: Taylan Kabbani, Ekrem Duman

Abstract:

The design of adaptive systems that take advantage of financial markets while reducing the risk can bring more stagnant wealth into the global market. However, most efforts made to generate successful deals in trading financial assets rely on Supervised Learning (SL), which suffered from various limitations. Deep Reinforcement Learning (DRL) offers to solve these drawbacks of SL approaches by combining the financial assets price "prediction" step and the "allocation" step of the portfolio in one unified process to produce fully autonomous systems capable of interacting with its environment to make optimal decisions through trial and error. In this paper, a continuous action space approach is adopted to give the trading agent the ability to gradually adjust the portfolio's positions with each time step (dynamically re-allocate investments), resulting in better agent-environment interaction and faster convergence of the learning process. In addition, the approach supports the managing of a portfolio with several assets instead of a single one. This work represents a novel DRL model to generate profitable trades in the stock market, effectively overcoming the limitations of supervised learning approaches. We formulate the trading problem, or what is referred to as The Agent Environment as Partially observed Markov Decision Process (POMDP) model, considering the constraints imposed by the stock market, such as liquidity and transaction costs. More specifically, we design an environment that simulates the real-world trading process by augmenting the state representation with ten different technical indicators and sentiment analysis of news articles for each stock. We then solve the formulated POMDP problem using the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm, which can learn policies in high-dimensional and continuous action spaces like those typically found in the stock market environment. From the point of view of stock market forecasting and the intelligent decision-making mechanism, this paper demonstrates the superiority of deep reinforcement learning in financial markets over other types of machine learning such as supervised learning and proves its credibility and advantages of strategic decision-making.

Keywords: the stock market, deep reinforcement learning, MDP, twin delayed deep deterministic policy gradient, sentiment analysis, technical indicators, autonomous agent

Procedia PDF Downloads 158

Authors: Rajan Goyal, S. Lamba, S. Annapoorni

Abstract:

The indirect exchange coupled system consists of two ferromagnetic layers separated by non-magnetic spacer layer. The type of exchange coupling may be either ferro or anti-ferro depending on the thickness of the spacer layer. In the present work, the strength of exchange coupling in FePt/Al₂O₃/Fe₃Pt has been investigated by varying the thickness of the spacer layer Al₂O₃. The FePt/Al₂O₃/Fe₃Pt trilayer structure is fabricated on Si <100> single crystal substrate using sputtering technique. The thickness of FePt and Fe₃Pt is fixed at 60 nm and 2 nm respectively. The thickness of spacer layer Al₂O₃ was varied from 0 to 16 nm. The normalized hysteresis loops recorded at room temperature both in the in-plane and out of plane configuration reveals that the orientation of easy axis lies along the plane of the film. It is observed that the hysteresis loop for ts=0 nm does not exhibit any knee around H=0 indicating that the hard FePt layer and soft Fe₃Pt layer are strongly exchange coupled. However, the insertion of Al₂O₃ spacer layer of thickness ts = 0.7 nm results in appearance of a minor knee around H=0 suggesting the weakening of exchange coupling between FePt and Fe₃Pt. The disappearance of knee in hysteresis loop with further increase in thickness of the spacer layer up to 8 nm predicts the co-existence of ferromagnetic (FM) and antiferromagnetic (AFM) exchange interaction between FePt and Fe₃Pt. In addition to this, the out of plane hysteresis loop also shows an asymmetry around H=0. The exchange field Hex = (Hc↑-HC↓)/2, where Hc↑ and Hc↓ are the coercivity estimated from lower and upper branch of hysteresis loop, increases from ~ 150 Oe to ~ 700 Oe respectively. This behavior may be attributed to the uncompensated moments in the hard FePt layer and soft Fe₃Pt layer at the interface. A better insight into the variation in indirect exchange coupling has been investigated using recoil curves. It is observed that the almost closed recoil curves are obtained for ts= 0 nm up to a reverse field of ~ 5 kOe. On the other hand, the appearance of appreciable open recoil curves at lower reverse field ~ 4 kOe for ts = 0.7 nm indicates that uncoupled soft phase undergoes irreversible magnetization reversal at lower reverse field suggesting the weakening of exchange coupling. The openness of recoil curves decreases with increase in thickness of the spacer layer up to 8 nm. This behavior may be attributed to the competition between FM and AFM exchange interactions. The FM exchange coupling between FePt and Fe₃Pt due to porous nature of Al₂O₃ decreases much slower than the weak AFM coupling due to interaction between Fe ions of FePt and Fe₃Pt via O ions of Al₂O₃. The hysteresis loop has been simulated using Monte Carlo based on Metropolis algorithm to investigate the variation in strength of exchange coupling in FePt/Al₂O₃/Fe₃Pt trilayer system.

Keywords: indirect exchange coupling, MH loop, Monte Carlo simulation, recoil curve

Procedia PDF Downloads 171

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 115

Authors: Anders Schou Simonsen, Kim Sorensen, Thomas Condra

Abstract:

Wet scrubbers are used for flue gas desulfurization by injecting water directly into the flue gas stream from a set of sprayers. The water droplets will flow freely inside the scrubber, and flow down along the scrubber walls as a thin wall film while reacting with the gas phase to remove SO₂. This complex multiphase phenomenon can be divided into three main contributions: the continuous gas phase, the liquid droplet phase, and the liquid wall film phase. This study proposes a complete model, where all three main contributions are taken into account and resolved using OpenFOAM for the continuous gas phase, and MATLAB for the liquid droplet and wall film phases. The 3D continuous gas phase is composed of five species: CO₂, H₂O, O₂, SO₂, and N₂, which are resolved along with momentum, energy, and turbulence. Source terms are present for four species, energy and momentum, which are affecting the steady-state solution. The liquid droplet phase experiences breakup, collisions, dynamics, internal chemistry, evaporation and condensation, species mass transfer, energy transfer and wall film interactions. Numerous sub-models have been implemented and coupled to realise the above-mentioned phenomena. The liquid wall film experiences impingement, acceleration, atomization, separation, internal chemistry, evaporation and condensation, species mass transfer, and energy transfer, which have all been resolved using numerous sub-models as well. The continuous gas phase has been coupled with the liquid phases using source terms by an approach, where the two software packages are couples using a link-structure. The complete CFD model has been verified using 16 experimental tests from an existing scrubber installation, where a gradient-based pattern search optimization algorithm has been used to tune numerous model parameters to match the experimental results. The CFD model needed to be fast for evaluation in order to apply this optimization routine, where approximately 1000 simulations were needed. The results show that the complex multiphase phenomena governing wet scrubbers can be resolved in a single model. The optimization routine was able to tune the model to accurately predict the performance of an existing installation. Furthermore, the study shows that a coupling between OpenFOAM and MATLAB is realizable, where the data and source term exchange increases the computational requirements by approximately 5%. This allows for exploiting the benefits of both software programs.

Keywords: desulfurization, discrete phase, scrubber, wall film

Procedia PDF Downloads 235

Authors: Priyanka Chaudhary, M. Rizwan

Abstract:

This paper presents a grid synchronization technique based on adaptive notch filter for SPV (Solar Photovoltaic) system along with MPPT (Maximum Power Point Tracking) techniques. An efficient grid synchronization technique offers proficient detection of various components of grid signal like phase and frequency. It also acts as a barrier for harmonics and other disturbances in grid signal. A reference phase signal synchronized with the grid voltage is provided by the grid synchronization technique to standardize the system with grid codes and power quality standards. Hence, grid synchronization unit plays important role for grid connected SPV systems. As the output of the PV array is fluctuating in nature with the meteorological parameters like irradiance, temperature, wind etc. In order to maintain a constant DC voltage at VSC (Voltage Source Converter) input, MPPT control is required to track the maximum power point from PV array. In this work, a variable step size P & O (Perturb and Observe) MPPT technique with DC/DC boost converter has been used at first stage of the system. This algorithm divides the dPpv/dVpv curve of PV panel into three separate zones i.e. zone 0, zone 1 and zone 2. A fine value of tracking step size is used in zone 0 while zone 1 and zone 2 requires a large value of step size in order to obtain a high tracking speed. Further, adaptive notch filter based control technique is proposed for VSC in PV generation system. Adaptive notch filter (ANF) approach is used to synchronize the interfaced PV system with grid to maintain the amplitude, phase and frequency parameters as well as power quality improvement. This technique offers the compensation of harmonics current and reactive power with both linear and nonlinear loads. To maintain constant DC link voltage a PI controller is also implemented and presented in this paper. The complete system has been designed, developed and simulated using SimPower System and Simulink toolbox of MATLAB. The performance analysis of three phase grid connected solar photovoltaic system has been carried out on the basis of various parameters like PV output power, PV voltage, PV current, DC link voltage, PCC (Point of Common Coupling) voltage, grid voltage, grid current, voltage source converter current, power supplied by the voltage source converter etc. The results obtained from the proposed system are found satisfactory.

Keywords: solar photovoltaic systems, MPPT, voltage source converter, grid synchronization technique

Procedia PDF Downloads 573