Search results for: imbalance dataset

Authors: Renato Morbidelli, Carla Saltalippi, Alessia Flammini, Marco Cifrodelli, Corrado Corradini

Abstract:

The trend of magnitude and frequency of extreme rainfalls seems to be different depending on the investigated area of the world. In this work, the impact of climate change on extreme rainfalls in Umbria, an inland region of central Italy, is examined using data recorded during the period 1921-2015 by 10 representative rain gauge stations. The study area is characterized by a complex orography, with altitude ranging from 200 to more than 2000 m asl. The climate is very different from zone to zone, with mean annual rainfall ranging from 650 to 1450 mm and mean annual air temperature from 3.3 to 14.2°C. Over the past 15 years, this region has been affected by four significant droughts as well as by six dangerous flood events, all with very large impact in economic terms. A least-squares linear trend analysis of annual maximums over 60 time series selected considering 6 different durations (1 h, 3 h, 6 h, 12 h, 24 h, 48 h) showed about 50% of positive and 50% of negative cases. For the same time series the non-parametrical Mann-Kendall test with a significance level 0.05 evidenced only 3% of cases characterized by a negative trend and no positive case. Further investigations have also demonstrated that the variance and covariance of each time series can be considered almost stationary. Therefore, the analysis on the magnitude of extreme rainfalls supplies the indication that an evident trend in the change of values in the Umbria region does not exist. However, also the frequency of rainfall events, with particularly high rainfall depths values, occurred during a fixed period has also to be considered. For all selected stations the 2-day rainfall events that exceed 50 mm were counted for each year, starting from the first monitored year to the end of 2015. Also, this analysis did not show predominant trends. Specifically, for all selected rain gauge stations the annual number of 2-day rainfall events that exceed the threshold value (50 mm) was slowly decreasing in time, while the annual cumulated rainfall depths corresponding to the same events evidenced trends that were not statistically significant. Overall, by using a wide available dataset and adopting simple methods, the influence of climate change on the heavy rainfalls in the Umbria region is not detected.

Keywords: climate changes, rainfall extremes, rainfall magnitude and frequency, central Italy

Procedia PDF Downloads 236

Authors: Fazıl Gökgöz, Fahrettin Filiz

Abstract:

Load forecasting has become crucial in recent years and become popular in forecasting area. Many different power forecasting models have been tried out for this purpose. Electricity load forecasting is necessary for energy policies, healthy and reliable grid systems. Effective power forecasting of renewable energy load leads the decision makers to minimize the costs of electric utilities and power plants. Forecasting tools are required that can be used to predict how much renewable energy can be utilized. The purpose of this study is to explore the effectiveness of LSTM-based neural networks for estimating renewable energy loads. In this study, we present models for predicting renewable energy loads based on deep neural networks, especially the Long Term Memory (LSTM) algorithms. Deep learning allows multiple layers of models to learn representation of data. LSTM algorithms are able to store information for long periods of time. Deep learning models have recently been used to forecast the renewable energy sources such as predicting wind and solar energy power. Historical load and weather information represent the most important variables for the inputs within the power forecasting models. The dataset contained power consumption measurements are gathered between January 2016 and December 2017 with one-hour resolution. Models use publicly available data from the Turkish Renewable Energy Resources Support Mechanism. Forecasting studies have been carried out with these data via deep neural networks approach including LSTM technique for Turkish electricity markets. 432 different models are created by changing layers cell count and dropout. The adaptive moment estimation (ADAM) algorithm is used for training as a gradient-based optimizer instead of SGD (stochastic gradient). ADAM performed better than SGD in terms of faster convergence and lower error rates. Models performance is compared according to MAE (Mean Absolute Error) and MSE (Mean Squared Error). Best five MAE results out of 432 tested models are 0.66, 0.74, 0.85 and 1.09. The forecasting performance of the proposed LSTM models gives successful results compared to literature searches.

Keywords: deep learning, long short term memory, energy, renewable energy load forecasting

Procedia PDF Downloads 266

Authors: Rima T. Alfaraj, Murtadha J. Al Tammar, Khaqan Khan, Khalid M. Alruwaili

Abstract:

OBJECTIVE/SCOPE (25-75): Caliper logging data provides critical information about wellbore shape and deformations, such as stress-induced borehole breakouts or washouts. Multiarm mechanical caliper logs are often run using wireline, which can be time-consuming, costly, and/or challenging to run in certain formations. To minimize rig time and improve operational safety, it is valuable to develop analytical solutions that can estimate caliper logs using available Logging-While-Drilling (LWD) data without the need to run wireline caliper logs. As a first step, the objective of this paper is to perform statistical analysis using an extensive datasetto identify important physical parameters that should be considered in developing such analytical solutions. METHODS, PROCEDURES, PROCESS (75-100): Caliper logs and LWD data of eleven wells, with a total of more than 80,000 data points, were obtained and imported into a data analytics software for analysis. Several parameters were selected to test the relationship of the parameters with the measured maximum and minimum caliper logs. These parameters includegamma ray, porosity, shear, and compressional sonic velocities, bulk densities, and azimuthal density. The data of the eleven wells were first visualized and cleaned.Using the analytics software, several analyses were then preformed, including the computation of Pearson’s correlation coefficients to show the statistical relationship between the selected parameters and the caliper logs. RESULTS, OBSERVATIONS, CONCLUSIONS (100-200): The results of this statistical analysis showed that some parameters show good correlation to the caliper log data. For instance, the bulk density and azimuthal directional densities showedPearson’s correlation coefficients in the range of 0.39 and 0.57, which wererelatively high when comparedto the correlation coefficients of caliper data with other parameters. Other parameters such as porosity exhibited extremely low correlation coefficients to the caliper data. Various crossplots and visualizations of the data were also demonstrated to gain further insights from the field data. NOVEL/ADDITIVE INFORMATION (25-75): This study offers a unique and novel look into the relative importance and correlation between different LWD measurements and wireline caliper logs via an extensive dataset. The results pave the way for a more informed development of new analytical solutions for estimating the size and shape of the wellbore in real-time while drilling using LWD data.

Keywords: LWD measurements, caliper log, correlations, analysis

Procedia PDF Downloads 121

Authors: Tesfaye Mengistu

Abstract:

Power cables play a crucial role in the transmission and distribution of electrical energy. As the electricity generation, transmission, distribution, and storage systems become smarter, there is a growing emphasis on incorporating intelligent approaches to ensure the reliability of power cables. Various types of electrical cables are employed for transmitting and distributing electrical energy, with cross-linked polyethylene (XLPE) cables being widely utilized due to their exceptional electrical and mechanical properties. However, insulation defects can occur in XLPE cables due to subpar manufacturing techniques during production and cable joint installation. To address this issue, experts have proposed different methods for monitoring XLPE cables. Some suggest the use of interdigital capacitive (IDC) technology for online monitoring, while others propose employing continuous wave (CW) terahertz (THz) imaging systems to detect internal defects in XLPE plates used for power cable insulation. In this study, we have developed models that employ a custom dataset collected locally to classify the physical safety status of individual power cables. Our models aim to replace physical inspections with computer vision and image processing techniques to classify defective power cables from non-defective ones. The implementation of our project utilized the Python programming language along with the TensorFlow package and a convolutional neural network (CNN). The CNN-based algorithm was specifically chosen for power cable defect classification. The results of our project demonstrate the effectiveness of CNNs in accurately classifying power cable defects. We recommend the utilization of similar or additional datasets to further enhance and refine our models. Additionally, we believe that our models could be used to develop methodologies for detecting power cable defects from live video feeds. We firmly believe that our work makes a significant contribution to the field of power cable inspection and maintenance. Our models offer a more efficient and cost-effective approach to detecting power cable defects, thereby improving the reliability and safety of power grids.

Keywords: artificial intelligence, computer vision, defect detection, convolutional neural net

Procedia PDF Downloads 112

Authors: Basma Yassa

Abstract:

Evidence on cash transfers’ effectiveness in mitigating macro and idiosyncratic shocks’ impacts has been mixed and is mostly concentrated in Latin America, Sub-Saharan Africa, and South Asia with very limited evidence from the MENA region. Yet conditional cash transfers schemes have been continually used, especially in Egypt, as the main social protection tool in response to the recent socioeconomic crises and macro shocks. We use 2 panel datasets and 1 cross-sectional dataset to estimate the effectiveness of cash transfers as a shock-mitigative mechanism in the Egyptian context. In this paper, the results from the different models (Panel Fixed Effects model and the Regression Discontinuity Design (RDD) model) confirm that micro and macro shocks lead to significant decline in several household-level welfare outcomes and that Takaful cash transfers have a significant positive impact in mitigating the negative shock impacts, especially on households’ debt incidence, debt levels, and asset ownership, but not necessarily on food, and non-food expenditure levels. The results indicate large positive significant effects on decreasing household incidence of debt by up to 12.4 percent and lowered the debt size by approximately 18 percent among Takaful beneficiaries compared to non-beneficiaries’. Similar evidence is found on asset ownership levels, as the RDD model shows significant positive effects on total asset ownership and productive asset ownership, but the model failed to detect positive impacts on per capita food and non-food expenditures. Further extensions are still in progress to compare the models’ results with the DID model results when using a nationally representative ELMPS panel data (2018/2024) rounds. Finally, our initial analysis suggests that conditional cash transfers are effective in buffering the negative shock impacts on certain welfare indicators even after successive macro-economic shocks in 2022 and 2023 in the Egyptian Context.

Keywords: cash transfers, fixed effects, household welfare, household debt, micro shocks, regression discontinuity design

Procedia PDF Downloads 44

Authors: Yanjun Chen, Chuanping Hu, Jie Shao, Lin Mei, Chongyang Zhang

Abstract:

Traditional image classification is based on encoding scheme (e.g. Fisher Vector, Vector of Locally Aggregated Descriptor) with low-level image features (e.g. SIFT, HoG). Compared to these low-level local features, deep convolutional features obtained at the mid-level layer of convolutional neural networks (CNN) have richer information but lack of geometric invariance. For scene classification, there are scattered objects with different size, category, layout, number and so on. It is crucial to find the distinctive objects in scene as well as their co-occurrence relationship. In this paper, we propose a method to take advantage of both deep convolutional features and the traditional encoding scheme while taking object-centric and scene-centric information into consideration. First, to exploit the object-centric and scene-centric information, two CNNs that trained on ImageNet and Places dataset separately are used as the pre-trained models to extract deep convolutional features at multiple scales. This produces dense local activations. By analyzing the performance of different CNNs at multiple scales, it is found that each CNN works better in different scale ranges. A scale-wise CNN adaption is reasonable since objects in scene are at its own specific scale. Second, a fisher kernel is applied to aggregate a global representation at each scale and then to merge into a single vector by using a post-processing method called scale-wise normalization. The essence of Fisher Vector lies on the accumulation of the first and second order differences. Hence, the scale-wise normalization followed by average pooling would balance the influence of each scale since different amount of features are extracted. Third, the Fisher vector representation based on the deep convolutional features is followed by a linear Supported Vector Machine, which is a simple yet efficient way to classify the scene categories. Experimental results show that the scale-specific feature extraction and normalization with CNNs trained on object-centric and scene-centric datasets can boost the results from 74.03% up to 79.43% on MIT Indoor67 when only two scales are used (compared to results at single scale). The result is comparable to state-of-art performance which proves that the representation can be applied to other visual recognition tasks.

Keywords: deep convolutional features, Fisher Vector, multiple scales, scale-specific normalization

Procedia PDF Downloads 331

Authors: Pranav Gulati, Isha Sharma

Abstract:

Hypertension and hypotension are known to have repercussions on the health of an individual, with hypertension contributing to an increased probability of risk to cardiovascular diseases and hypotension resulting in syncope. This prompts the development of a non-invasive, non-intrusive, continuous and cuff-less blood pressure monitoring system to detect blood pressure variations and to identify individuals with acute and chronic heart ailments, but due to the unavailability of such devices for practical daily use, it becomes difficult to screen and subsequently regulate blood pressure. The complexities which hamper the steady monitoring of blood pressure comprises of the variations in physical characteristics from individual to individual and the postural differences at the site of monitoring. We propose to develop a continuous, comprehensive cardio-analysis tool, based on reflective photoplethysmography (PPG). The proposed device, in the form of an eyewear captures the PPG signal and estimates the systolic and diastolic blood pressure using a sensor positioned near the temporal artery. This system relies on regression models which are based on extraction of key points from a pair of PPG wavelets. The proposed system provides an edge over the existing wearables considering that it allows for uniform contact and pressure with the temporal site, in addition to minimal disturbance by movement. Additionally, the feature extraction algorithms enhance the integrity and quality of the extracted features by reducing unreliable data sets. We tested the system with 12 subjects of which 6 served as the training dataset. For this, we measured the blood pressure using a cuff based BP monitor (Omron HEM-8712) and at the same time recorded the PPG signal from our cardio-analysis tool. The complete test was conducted by using the cuff based blood pressure monitor on the left arm while the PPG signal was acquired from the temporal site on the left side of the head. This acquisition served as the training input for the regression model on the selected features. The other 6 subjects were used to validate the model by conducting the same test on them. Results show that the developed prototype can robustly acquire the PPG signal and can therefore be used to reliably predict blood pressure levels.

Keywords: blood pressure, photoplethysmograph, eyewear, physiological monitoring

Procedia PDF Downloads 277

Authors: Man-Yun Liu, Emily Chia-Yu Su

Abstract:

Alzheimer's disease (AD) is the public health crisis of the 21st century. AD is a degenerative brain disease and the most common cause of dementia, a costly disease on the healthcare system. Unfortunately, the cause of AD is poorly understood, furthermore; the treatments of AD so far can only alleviate symptoms rather cure or stop the progress of the disease. Currently, there are several ways to diagnose AD; medical imaging can be used to distinguish between AD, other dementias, and early onset AD, and cerebrospinal fluid (CSF). Compared with other diagnostic tools, blood (plasma) test has advantages as an approach to population-based disease screening because it is simpler, less invasive also cost effective. In our study, we used blood biomarkers dataset of The Alzheimer’s disease Neuroimaging Initiative (ADNI) which was funded by National Institutes of Health (NIH) to do data analysis and develop a prediction model. We used independent analysis of datasets to identify plasma protein biomarkers predicting early onset AD. Firstly, to compare the basic demographic statistics between the cohorts, we used SAS Enterprise Guide to do data preprocessing and statistical analysis. Secondly, we used logistic regression, neural network, decision tree to validate biomarkers by SAS Enterprise Miner. This study generated data from ADNI, contained 146 blood biomarkers from 566 participants. Participants include cognitive normal (healthy), mild cognitive impairment (MCI), and patient suffered Alzheimer’s disease (AD). Participants’ samples were separated into two groups, healthy and MCI, healthy and AD, respectively. We used the two groups to compare important biomarkers of AD and MCI. In preprocessing, we used a t-test to filter 41/47 features between the two groups (healthy and AD, healthy and MCI) before using machine learning algorithms. Then we have built model with 4 machine learning methods, the best AUC of two groups separately are 0.991/0.709. We want to stress the importance that the simple, less invasive, common blood (plasma) test may also early diagnose AD. As our opinion, the result will provide evidence that blood-based biomarkers might be an alternative diagnostics tool before further examination with CSF and medical imaging. A comprehensive study on the differences in blood-based biomarkers between AD patients and healthy subjects is warranted. Early detection of AD progression will allow physicians the opportunity for early intervention and treatment.

Keywords: Alzheimer's disease, blood-based biomarkers, diagnostics, early detection, machine learning

Procedia PDF Downloads 322

Authors: Seema Paul, John Mango Magero, Prosun Bhattacharya, Zahra Kalantari, Steve W. Lyon

Abstract:

The purpose of this review paper is to develop an understanding of lake hydrodynamics and the potential climate impact on the Lake Victoria (LV) catchment scale. This paper briefly discusses the main problems of lake hydrodynamics and its’ solutions that are related to quality assessment and climate effect. An empirical methodology in modeling and mapping have considered for understanding lake hydrodynamic and visualizing the long-term observational daily, monthly, and yearly mean dataset results by using geographical information system (GIS) and Comsol techniques. Data were obtained for the whole lake and five different meteorological stations, and several geoprocessing tools with spatial analysis are considered to produce results. The linear regression analyses were developed to build climate scenarios and a linear trend on lake rainfall data for a long period. A potential evapotranspiration rate has been described by the MODIS and the Thornthwaite method. The rainfall effect on lake water level observed by Partial Differential Equations (PDE), and water quality has manifested by a few nutrients parameters. The study revealed monthly and yearly rainfall varies with monthly and yearly maximum and minimum temperatures, and the rainfall is high during cool years and the temperature is high associated with below and average rainfall patterns. Rising temperatures are likely to accelerate evapotranspiration rates and more evapotranspiration is likely to lead to more rainfall, drought is more correlated with temperature and cloud is more correlated with rainfall. There is a trend in lake rainfall and long-time rainfall on the lake water surface has affected the lake level. The onshore and offshore have been concentrated by initial literature nutrients data. The study recommended that further studies should consider fully lake bathymetry development with flow analysis and its’ water balance, hydro-meteorological processes, solute transport, wind hydrodynamics, pollution and eutrophication these are crucial for lake water quality, climate impact assessment, and water sustainability.

Keywords: climograph, climate scenarios, evapotranspiration, linear trend flow, rainfall event on LV, concentration

Procedia PDF Downloads 99

Authors: Raluca Cosmina Budian

Abstract:

The homeless population in Spain, particularly among immigrants, has been a persistent and multifaceted issue. The government has implemented various housing public policies over the years to address homelessness, ranging from shelter programs to initiatives promoting permanent housing solutions. However, understanding the effectiveness of these policies requires insight from the very individuals and professionals directly impacted by or involved in their execution. This research sheds light on national strategies (The 2015-2020 Comprehensive National Strategy for the Homeless and National Strategy to Combat Homelessness in Spain 2023-2030) aimed at tackling homelessness in Spain, with a focus on the evolving landscape of housing public policies and their relationship with the homeless population. We investigate how these strategies have transformed over time and their impact on the inclusion of this vulnerable group. Furthermore, we explore the perspectives of homeless immigrants, distinguishing between those with an extended residency in Spain and those who have more recently arrived (less than 2 years); and distinguishing between women and men. Additionally, we incorporate insights from 13 interviews with professionals dedicated to serving the homeless population. These insights offer a deeper understanding of the intricacies of current homelessness service provision. Our findings reveal the complex dynamics of providing services to homeless individuals, and the importance of aligning these efforts with the broader national strategies for tackling homelessness. Drawing on a comprehensive dataset, we offer a nuanced view of the challenges and successes in implementing inclusive housing policies in the Spanish context. Our research highlights the importance of collaboration between policy makers, service providers and advocates to create a cohesive and effective approach. By fostering such collaboration, we aim to create a more inclusive and comprehensive strategy to address homelessness in Spain and possible affordable housing proposals for this vulnerable group. It´s only underscores the importance of tailored approaches but also contributes to the broader discourse on housing public policies' ability to address homelessness and foster integration. We suggest that a more comprehensive approach, considering the unique needs of immigrants and working in collaboration with professionals in the field, is essential for the development of effective strategies to combat homelessness and ensure the right to adequate housing for all.

Keywords: housing, homeless, public policy, Spain

Procedia PDF Downloads 75

Authors: Kseniia Zviagintseva, Brett Buttliere

Abstract:

Science is fundamentally a problem-solving enterprise, and scientists pay more attention to the negative things, that cause them dissonance and negative affective state of uncertainty or contradiction. While this is agreed upon by philosophers of science, there are few empirical demonstrations. Here we examine the keywords from those papers published by PLoS in 2014 and show with several sentiment analyzers that negative keywords are studied more than positive keywords. Our dataset is the 927,406 keywords of 32,870 scientific articles in all fields published in 2014 by the journal PLOS ONE (collected from Altmetric.com). Counting how often the 47,415 unique keywords are used, we can examine whether those negative topics are studied more than positive. In order to find the sentiment of the keywords, we utilized two sentiment analysis tools, Hu and Liu (2004) and SentiStrength (2014). The results below are for Hu and Liu as these are the less convincing results. The average keyword was utilized 19.56 times, with half of the keywords being utilized only 1 time and the maximum number of uses being 18,589 times. The keywords identified as negative were utilized 37.39 times, on average, with the positive keywords being utilized 14.72 times and the neutral keywords - 19.29, on average. This difference is only marginally significant, with an F value of 2.82, with a p of .05, but one must keep in mind that more than half of the keywords are utilized only 1 time, artificially increasing the variance and driving the effect size down. To examine more closely, we looked at those top 25 most utilized keywords that have a sentiment. Among the top 25, there are only two positive words, ‘care’ and ‘dynamics’, in position numbers 5 and 13 respectively, with all the rest being identified as negative. ‘Diseases’ is the most studied keyword with 8,790 uses, with ‘cancer’ and ‘infectious’ being the second and fourth most utilized sentiment-laden keywords. The sentiment analysis is not perfect though, as the words ‘diseases’ and ‘disease’ are split by taking 1st and 3rd positions. Combining them, they remain as the most common sentiment-laden keyword, being utilized 13,236 times. More than just splitting the words, the sentiment analyzer logs ‘regression’ and ‘rat’ as negative, and these should probably be considered false positives. Despite these potential problems, the effect is apparent, as even the positive keywords like ‘care’ could or should be considered negative, since this word is most commonly utilized as a part of ‘health care’, ‘critical care’ or ‘quality of care’ and generally associated with how to improve it. All in all, the results suggest that negative concepts are studied more, also providing support for the notion that science is most generally a problem-solving enterprise. The results also provide evidence that negativity and contradiction are related to greater productivity and positive outcomes.

Keywords: bibliometrics, keywords analysis, negativity bias, positive and negative words, scientific papers, scientometrics

Procedia PDF Downloads 186

Authors: Ngoc-Hieu Vu

Abstract:

Dairy production in Vietnam is a relatively new agricultural activity and milk production increased remarkably in recent years. Smallholders are still the main drivers for this development, especially in the southern part of the country. However, information on the farming practices is very limited. Therefore, this study aimed to determine factors influencing milk yield and quality (milk fat, total solids, solids-not-fat, total number of bacteria, and somatic cell count) and revenue of dairy farms in Southern Vietnam. The collection of data was at the farm level; individual animal records were unavailable. The 539 studied farms were located in the provinces Lam Dong (N=111 farms), Binh Duong (N=69 farms), Long An (N=174 farms), and Ho Chi Minh city (N=185 farms). The dataset included 9221 monthly test-day records of the farms from January 2013 to May 2015. Seasons were defined as rainy and dry. Farms sizes were classified as small (< 10 milking cows), medium (10 to 19 milking cows) and large (≥ 20 milking cows). The model for each trait contained year-season and farm region-farm size as subclass fixed effects, and individual farm and residual as random effects. Results showed that year-season, region, and farm size were determining sources of variation affecting all studied traits. Milk yield was higher in dry than in rainy seasons (P < 0.05), while it tended to increase from years 2013 to 2015. Large farms had higher yields (445.6 kg/cow) than small (396.7 kg/cow) and medium (428.0 kg/cow) farms (P < 0.05). Small farms, in contrast, were superior to large farms in terms of milk fat, total solids, solids-not-fat, total number of bacteria, and somatic cell count than large farms (P < 0.05). Revenue per cow was higher in large compared with medium and small farms. In conclusion, large farms achieved higher milk yields and revenues per cow, while small farms were superior in milk quality. Overall, milk yields were low and better training, financial support and marketing opportunities for farmers are needed to improve dairy production and increase farm revenues in Southern Vietnam.

Keywords: farm size, milk yield and quality, season, Southern Vietnam

Procedia PDF Downloads 361

Authors: Jonathan Sohn

Abstract:

Deep neural networks (DNNs) have shown state-of-the-art performance for many applications, including computer vision, natural language processing, and speech recognition. Recently, adversarial attacks have been studied in the context of deep neural networks, which aim to alter the results of deep neural networks by modifying the inputs slightly. For example, an adversarial attack on a DNN used for object detection can cause the DNN to miss certain objects. As a result, the reliability of DNNs is undermined by their lack of robustness against adversarial attacks, raising concerns about their use in safety-critical applications such as autonomous driving. In this paper, we focus on studying the adversarial attacks and defenses on DNNs for image classification. There are two types of adversarial attacks studied which are fast gradient sign method (FGSM) attack and projected gradient descent (PGD) attack. A DNN forms decision boundaries that separate the input images into different categories. The adversarial attack slightly alters the image to move over the decision boundary, causing the DNN to misclassify the image. FGSM attack obtains the gradient with respect to the image and updates the image once based on the gradients to cross the decision boundary. PGD attack, instead of taking one big step, repeatedly modifies the input image with multiple small steps. There is also another type of attack called the target attack. This adversarial attack is designed to make the machine classify an image to a class chosen by the attacker. We can defend against adversarial attacks by incorporating adversarial examples in training. Specifically, instead of training the neural network with clean examples, we can explicitly let the neural network learn from the adversarial examples. In our experiments, the digit recognition accuracy on the MNIST dataset drops from 97.81% to 39.50% and 34.01% when the DNN is attacked by FGSM and PGD attacks, respectively. If we utilize FGSM training as a defense method, the classification accuracy greatly improves from 39.50% to 92.31% for FGSM attacks and from 34.01% to 75.63% for PGD attacks. To further improve the classification accuracy under adversarial attacks, we can also use a stronger PGD training method. PGD training improves the accuracy by 2.7% under FGSM attacks and 18.4% under PGD attacks over FGSM training. It is worth mentioning that both FGSM and PGD training do not affect the accuracy of clean images. In summary, we find that PGD attacks can greatly degrade the performance of DNNs, and PGD training is a very effective way to defend against such attacks. PGD attacks and defence are overall significantly more effective than FGSM methods.

Keywords: deep neural network, adversarial attack, adversarial defense, adversarial machine learning

Procedia PDF Downloads 195

Authors: Pranav Gunhal., Krish Jhurani

Abstract:

This study tackles the significant computational challenge of predicting excited state dynamics in organic photovoltaic (OPV) materials—a pivotal factor in the performance of solar energy solutions. Time-dependent density functional theory (TDDFT), though effective, is computationally prohibitive for larger and more complex molecules. As a solution, the research explores the application of transformer neural networks, a type of artificial intelligence (AI) model known for its superior performance in natural language processing, to predict excited state dynamics in OPV materials. The methodology involves a two-fold process. First, the transformer model is trained on an extensive dataset comprising over 10,000 TDDFT calculations of excited state dynamics from a diverse set of OPV materials. Each training example includes a molecular structure and the corresponding TDDFT-calculated excited state lifetimes and key electronic transitions. Second, the trained model is tested on a separate set of molecules, and its predictions are rigorously compared to independent TDDFT calculations. The results indicate a remarkable degree of predictive accuracy. Specifically, for a test set of 1,000 OPV materials, the transformer model predicted excited state lifetimes with a mean absolute error of 0.15 picoseconds, a negligible deviation from TDDFT-calculated values. The model also correctly identified key electronic transitions contributing to the excited state dynamics in 92% of the test cases, signifying a substantial concordance with the results obtained via conventional quantum chemistry calculations. The practical integration of the transformer model with existing quantum chemistry software was also realized, demonstrating its potential as a powerful tool in the arsenal of materials scientists and chemists. The implementation of this AI model is estimated to reduce the computational cost of predicting excited state dynamics by two orders of magnitude compared to conventional TDDFT calculations. The successful utilization of transformer neural networks to accurately predict excited state dynamics provides an efficient computational pathway for the accelerated discovery and design of new OPV materials, potentially catalyzing advancements in the realm of sustainable energy solutions.

Keywords: transformer neural networks, organic photovoltaic materials, excited state dynamics, time-dependent density functional theory, predictive modeling

Procedia PDF Downloads 118

Authors: Anindita Patra, Prasad K. Bhaskaran

Abstract:

The atmosphere-ocean is a dynamically linked system that influences the exchange of energy, mass, and gas at the air-sea interface. The exchange of energy takes place in the form of sensible heat, latent heat, and momentum commonly referred to as fluxes along the atmosphere-ocean boundary. The large scale features such as El Nino and Southern Oscillation (ENSO) is a classic example on the interaction mechanism that occurs along the air-sea interface that deals with the inter-annual variability of the Earth’s Climate System. Most importantly the ocean and atmosphere as a coupled system acts in tandem thereby maintaining the energy balance of the climate system, a manifestation of the coupled air-sea interaction process. The present work is an attempt to understand the long-term variability in atmospheric parameters (from surface to upper levels) and investigate their role in influencing the surface ocean variables. More specifically the influence of atmospheric circulation and its variability influencing the mean Sea Level Pressure (SLP) has been explored. The study reports on a critical examination of both ocean-atmosphere parameters during a monsoon season over the head Bay of Bengal region. A trend analysis has been carried out for several atmospheric parameters such as the air temperature, geo-potential height, and omega (vertical velocity) for different vertical levels in the atmosphere (from surface to the troposphere) covering a period from 1992 to 2012. The Reanalysis 2 dataset from the National Centers for Environmental Prediction-Department of Energy (NCEP-DOE) was used in this study. The study signifies that the variability in air temperature and omega corroborates with the variation noticed in geo-potential height. Further, the study advocates that for the lower atmosphere the geo-potential heights depict a typical east-west contrast exhibiting a zonal dipole behavior over the study domain. In addition, the study clearly brings to light that the variations over different levels in the atmosphere plays a pivotal role in supporting the observed dipole pattern as clearly evidenced from the trends in SLP, associated surface wind speed and significant wave height over the study domain.

Keywords: air temperature, geopotential height, head Bay of Bengal, long-term variability, NCEP reanalysis 2, omega, wind-waves

Procedia PDF Downloads 225

Authors: Habtamu Ayenew Asegie

Abstract:

Wealth, as opposed to income or consumption, implies a more stable and permanent status. Due to natural and human-made difficulties, households' economies will be diminished, and their well-being will fall into trouble. Hence, governments and humanitarian agencies offer considerable resources for poverty and malnutrition reduction efforts. One key factor in the effectiveness of such efforts is the accuracy with which low-income or poor populations can be identified. As a result, this study aims to predict a household’s wealth status using ensemble Machine learning (ML) algorithms. In this study, design science research methodology (DSRM) is employed, and four ML algorithms, Random Forest (RF), Adaptive Boosting (AdaBoost), Light Gradient Boosted Machine (LightGBM), and Extreme Gradient Boosting (XGBoost), have been used to train models. The Ethiopian Demographic and Health Survey (EDHS) dataset is accessed for this purpose from the Central Statistical Agency (CSA)'s database. Various data pre-processing techniques were employed, and the model training has been conducted using the scikit learn Python library functions. Model evaluation is executed using various metrics like Accuracy, Precision, Recall, F1-score, area under curve-the receiver operating characteristics (AUC-ROC), and subjective evaluations of domain experts. An optimal subset of hyper-parameters for the algorithms was selected through the grid search function for the best prediction. The RF model has performed better than the rest of the algorithms by achieving an accuracy of 96.06% and is better suited as a solution model for our purpose. Following RF, LightGBM, XGBoost, and AdaBoost algorithms have an accuracy of 91.53%, 88.44%, and 58.55%, respectively. The findings suggest that some of the features like ‘Age of household head’, ‘Total children ever born’ in a family, ‘Main roof material’ of their house, ‘Region’ they lived in, whether a household uses ‘Electricity’ or not, and ‘Type of toilet facility’ of a household are determinant factors to be a focal point for economic policymakers. The determinant risk factors, extracted rules, and designed artifact achieved 82.28% of the domain expert’s evaluation. Overall, the study shows ML techniques are effective in predicting the wealth status of households.

Keywords: ensemble machine learning, households wealth status, predictive model, wealth status prediction

Procedia PDF Downloads 38

Authors: Dona Shaini Abhilasha Nanayakkara, Kurugamage Jude Pravinda Gregory Perera

Abstract:

E-commerce platforms have revolutionized the shopping experience, offering convenient ways for consumers to make purchases. To improve interactions with customers and optimize marketing strategies, it is essential for businesses to understand user behavior, preferences, and needs on these platforms. This paper focuses on recommending businesses to customize interactions with users based on their behavioral patterns, leveraging data-driven analysis and machine learning techniques. Businesses can improve engagement and boost the adoption of e-commerce platforms by aligning behavioral patterns with user goals of usability and satisfaction. We propose TheAnalyzer, a clustering-based system designed to enhance business productivity by analyzing user-profiles and improving human-computer interaction. The Analyzer seamlessly integrates with business applications, collecting relevant data points based on users' natural interactions without additional burdens such as questionnaires or surveys. It defines five key user analytics as features for its dataset, which are easily captured through users' interactions with e-commerce platforms. This research presents a study demonstrating the successful distinction of users into specific groups based on the five key analytics considered by TheAnalyzer. With the assistance of domain experts, customized business rules can be attached to each group, enabling The Analyzer to influence business applications and provide an enhanced personalized user experience. The outcomes are evaluated quantitatively and qualitatively, demonstrating that utilizing TheAnalyzer’s capabilities can optimize business outcomes, enhance customer satisfaction, and drive sustainable growth. The findings of this research contribute to the advancement of personalized interactions in e-commerce platforms. By leveraging user behavioral patterns and analyzing both new and existing users, businesses can effectively tailor their interactions to improve customer satisfaction, loyalty and ultimately drive sales.

Keywords: data clustering, data standardization, dimensionality reduction, human computer interaction, user profiling

Procedia PDF Downloads 72

Authors: Jitendra Pratap, Jonathan Sivyer

Abstract:

Introduction: Dual energy/Spectral CT scanning permits simultaneous acquisition of two x-ray spectra datasets and can complement radiological diagnosis by allowing tissue characterisation (e.g., uric acid vs. non-uric acid renal stones), enhancing structures (e.g. boost iodine signal to improve contrast resolution), and quantifying substances (e.g. iodine density). However, the latter showed inconsistent results between the 2 main modes of dual energy scanning (i.e. dual source vs. dual layer). Therefore, the present study aimed to determine which technology is more accurate in quantifying iodine density. Methods: Twenty vials with known concentrations of iodine solutions were made using Optiray 350 contrast media diluted in sterile water. The concentration of iodine utilised ranged from 0.1 mg/ml to 1.0mg/ml in 0.1mg/ml increments, 1.5 mg/ml to 4.5 mg/ml in 0.5mg/ml increments followed by further concentrations at 5.0 mg/ml, 7mg/ml, 10 mg/ml and 15mg/ml. The vials were scanned using Dual Energy scan mode on a Siemens Somatom Force at 80kV/Sn150kV and 100kV/Sn150kV kilovoltage pairing. The same vials were scanned using Spectral scan mode on a Philips iQon at 120kVp and 140kVp. The images were reconstructed at 5mm thickness and 5mm increment using Br40 kernel on the Siemens Force and B Filter on Philips iQon. Post-processing of the Dual Energy data was performed on vendor-specific Siemens Syngo VIA (VB40) and Philips Intellispace Portal (Ver. 12) for the Spectral data. For each vial and scan mode, the iodine concentration was measured by placing an ROI in the coronal plane. Intraclass correlation analysis was performed on both datasets. Results: The iodine concentrations were reproduced with a high degree of accuracy for Dual Layer CT scanner. Although the Dual Source images showed a greater degree of deviation in measured iodine density for all vials, the dataset acquired at 80kV/Sn150kV had a higher accuracy. Conclusion: Spectral CT scanning by the dual layer technique has higher accuracy for quantitative measurements of iodine density compared to the dual source technique.

Keywords: CT, iodine density, spectral, dual-energy

Procedia PDF Downloads 119

Authors: Nassima Noufail, Sara Bouhali

Abstract:

In this work, we develop a semi-supervised solution for the purpose of action detection in videos and propose an efficient algorithm for video segmentation. The approach is divided into video segmentation, feature extraction, and classification. In the first part, a video is segmented into clips, and we used the K-means algorithm for this segmentation; our goal is to find groups based on similarity in the video. The application of k-means clustering into all the frames is time-consuming; therefore, we started by the identification of transition frames where the scene in the video changes significantly, and then we applied K-means clustering into these transition frames. We used two image filters, the gaussian filter and the Laplacian of Gaussian. Each filter extracts a set of features from the frames. The Gaussian filter blurs the image and omits the higher frequencies, and the Laplacian of gaussian detects regions of rapid intensity changes; we then used this vector of filter responses as an input to our k-means algorithm. The output is a set of cluster centers. Each video frame pixel is then mapped to the nearest cluster center and painted with a corresponding color to form a visual map. The resulting visual map had similar pixels grouped. We then computed a cluster score indicating how clusters are near each other and plotted a signal representing frame number vs. clustering score. Our hypothesis was that the evolution of the signal would not change if semantically related events were happening in the scene. We marked the breakpoints at which the root mean square level of the signal changes significantly, and each breakpoint is an indication of the beginning of a new video segment. In the second part, for each segment from part 1, we randomly selected a 16-frame clip, then we extracted spatiotemporal features using convolutional 3D network C3D for every 16 frames using a pre-trained model. The C3D final output is a 512-feature vector dimension; hence we used principal component analysis (PCA) for dimensionality reduction. The final part is the classification. The C3D feature vectors are used as input to a multi-class linear support vector machine (SVM) for the training model, and we used a multi-classifier to detect the action. We evaluated our experiment on the UCF101 dataset, which consists of 101 human action categories, and we achieved an accuracy that outperforms the state of art by 1.2%.

Keywords: video segmentation, action detection, classification, Kmeans, C3D

Procedia PDF Downloads 77

Authors: Sahar Elsaleh, Ghada Farhat, Shaikha Al-Derham, Fasih Alam

Abstract:

Background: Cesarean section (CS) delivery is one of the major concerns both in developing and developed countries. The rate of CS deliveries are on the rise globally, and especially in Qatar. Many socio-economic, demographic, clinical and institutional factors play an important role for cesarean sections. This study aims to investigate factors affecting the prevalence of CS among women in Qatar using the UNICEF’s Multiple Indicator Cluster Survey (MICS) 2012 database. Methods: The study has focused on the women’s questionnaire of the MICS, which was successfully distributed to 5699 participants. Following study inclusion and exclusion criteria, a final sample of 761 women aged 19- 49 years who had at least one delivery of giving birth in their lifetime before the survey were included. A number of socio-economic, demographic, clinical and institutional factors, identified through literature review and available in the data, were considered for the analyses. Bivariate and multivariate logistic regression models, along with a multi-level modeling to investigate clustering effect, were undertaken to identify the factors that affect CS prevalence in Qatar. Results: From the bivariate analyses the study has shown that, a number of categorical factors are statistically significantly associated with the dependent variable (CS). When identifying the factors from a multivariate logistic regression, the study found that only three categorical factors -‘age of women’, ‘place at delivery’ and ‘baby weight’ appeared to be significantly affecting the CS among women in Qatar. Although the MICS dataset is based on a cluster survey, an exploratory multi-level analysis did not show any clustering effect, i.e. no significant variation in results at higher level (households), suggesting that all analyses at lower level (individual respondent) are valid without any significant bias in results. Conclusion: The study found a statistically significant association between the dependent variable (CS delivery) and age of women, frequency of TV watching, assistance at birth and place of birth. These results need to be interpreted cautiously; however, it can be used as evidence-base for further research on cesarean section delivery in Qatar.

Keywords: cesarean section, factors, multiple indicator cluster survey, MICS database, Qatar

Procedia PDF Downloads 116

Authors: Ali Hamza

Abstract:

Accurate segmentation of breast ultrasound images is of paramount importance in enhancing the diagnostic capabilities of breast cancer detection. This study presents an approach utilizing the U-Net architecture for segmenting breast ultrasound images aimed at improving the accuracy and reliability of mass identification within the breast tissue. The proposed method encompasses a multi-stage process. Initially, preprocessing techniques are employed to refine image quality and diminish noise interference. Subsequently, the U-Net architecture, a deep learning convolutional neural network (CNN), is employed for pixel-wise segmentation of regions of interest corresponding to potential breast masses. The U-Net's distinctive architecture, characterized by a contracting and expansive pathway, enables accurate boundary delineation and detailed feature extraction. To evaluate the effectiveness of the proposed approach, an extensive dataset of breast ultrasound images is employed, encompassing diverse cases. Quantitative performance metrics such as the Dice coefficient, Jaccard index, sensitivity, specificity, and Hausdorff distance are employed to comprehensively assess the segmentation accuracy. Comparative analyses against traditional segmentation methods showcase the superiority of the U-Net architecture in capturing intricate details and accurately segmenting breast masses. The outcomes of this study emphasize the potential of the U-Net-based segmentation approach in bolstering breast ultrasound image analysis. The method's ability to reliably pinpoint mass boundaries holds promise for aiding radiologists in precise diagnosis and treatment planning. However, further validation and integration within clinical workflows are necessary to ascertain their practical clinical utility and facilitate seamless adoption by healthcare professionals. In conclusion, leveraging the U-Net architecture for breast ultrasound image segmentation showcases a robust framework that can significantly enhance diagnostic accuracy and advance the field of breast cancer detection. This approach represents a pivotal step towards empowering medical professionals with a more potent tool for early and accurate breast cancer diagnosis.

Keywords: mage segmentation, U-Net, deep learning, breast cancer detection, diagnostic accuracy, mass identification, convolutional neural network

Procedia PDF Downloads 84

Authors: Tunde Szecsi, Janka Szilagyi

Abstract:

Students majoring in education are underrepresented in undergraduate scholarship. To enable and encourage teacher candidates to engage in scholarly activities, it is essential to infuse skills such as problem-solving, critical thinking, oral and written communication, collaboration and the utilization of information literacy, into courses in teacher preparation programs. In this empirical study, we examined two teacher education programs – one in New York State and one in Florida – in terms of the approaches of the course-based infusion of skills for undergraduate research, and the effectiveness of this infusion. First, course-related documents such as syllabi, assignment descriptions, and course activities were reviewed and analyzed. The goal of the document analysis was to identify and describe the targeted skills, and the pedagogical approaches and strategies for promoting research skills in teacher candidates. Next, a selection of teacher candidates’ scholarly products from the institution in Florida was used as a data set to examine teacher candidates’ skill development in the context of the identified assignments. This dataset was analyzed both quantitatively and qualitatively to describe the changes that occurred in teacher candidates’ critical thinking, communication, and information literacy skills, and to uncover patterns in the skill development at the two institutions. Descriptive statistics were calculated to explore the changes in these skills of teacher candidates over a period of three years. The findings based on data from the teacher education program in Florida indicated a steady gain in written communication and critical thinking and a modest increase in informational literacy. At the institution in New York, candidates’ submission and success rates on the edTPA, a New York State Teacher Certification exam, was used as a measure of scholarly skills. Overall, although different approaches were used for infusing the development of scholarly skills in the courses, the results suggest that a holistic and well-orchestrated infusion of the skills into most courses in the teacher education program might result in steadily developing scholarly skills. These results offered essential implications for teacher education programs in terms of further improvements in teacher candidates’ skills for engaging in undergraduate research and scholarship. In this presentation, our purpose is to showcase two approaches developed by two teacher education programs to demonstrate how diverse approaches toward the promotion of undergraduate scholarship activities are responsive to the context of the teacher preparation programs.

Keywords: critical thinking, pedagogical strategies, teacher education, undergraduate student research

Procedia PDF Downloads 161

Authors: Haiqin Zhou, Robin Irons

Abstract:

The introduction of hydrogen into local networks may, in many cases, require the initial operation of those systems on natural gas/hydrogen blends, either because of a lack of sufficient hydrogen to allow a 100% conversion or because existing infrastructure imposes limitations on the % hydrogen that can be burned before the end-use technologies are replaced. In many systems, the largest number of end-use technologies are small-scale but numerous appliances used for domestic and industrial heating and cooking. In such a scenario, it is important to understand exactly how much hydrogen can be introduced into these appliances before their performance becomes unacceptable and what imposes that limitation. This study seeks to explore a range of significantly higher hydrogen blends and a broad range of factors that might limit operability or environmental acceptability. We will present tests from a burner designed for space heating and optimized for natural gas as an increasing % of hydrogen blends (increasing from 25%) were burned and explore the range of parameters that might govern the acceptability of operation. These include gaseous emissions (particularly NOx and unburned carbon), temperature, flame length, stability and general operational acceptability. Results will show emissions, Temperature, and flame length as a function of thermal load and percentage of hydrogen in the blend. The relevant application and regulation will ultimately determine the acceptability of these values, so it is important to understand the full operational envelope of the burners in question through the sort of extensive parametric testing we have carried out. The present dataset should represent a useful data source for designers interested in exploring appliance operability. In addition to this, we present data on two factors that may be absolutes in determining allowable hydrogen percentages. The first of these is flame blowback. Our results show that, for our system, the threshold between acceptable and unacceptable performance lies between 60 and 65% mol% hydrogen. Another factor that may limit operation, and which would be important in domestic applications, is the acoustic performance of these burners. We will describe a range of operational conditions in which hydrogen blend burners produce a loud and invasive ‘screech’. It will be important for equipment designers and users to find ways to avoid this or mitigate it if performance is to be deemed acceptable.

Keywords: blends, operational, domestic appliances, future system operation.

Procedia PDF Downloads 23

Authors: Marek Dosbaba

Abstract:

Within the mining sector, SEM-based Automated Mineralogy (AM) has been the standard application for quickly and efficiently handling mineral processing tasks. Over the last decade, the trend has been to analyze larger numbers of samples, often with a higher level of detail. This has necessitated a shift from interactive sample analysis performed by an operator using a SEM, to an increased reliance on offline processing to analyze and report the data. In response to this trend, TESCAN TIMA Mineral Analyzer is designed to quickly create a virtual copy of the studied samples, thereby preserving all the necessary information. Depending on the selected data acquisition mode, TESCAN TIMA can perform hyperspectral mapping and save an X-ray spectrum for each pixel or segment, respectively. This approach allows the user to browse through elemental distribution maps of all elements detectable by means of energy dispersive spectroscopy. Re-evaluation of the existing data for the presence of previously unconsidered elements is possible without the need to repeat the analysis. Additional tiers of data such as a secondary electron or cathodoluminescence images can also be recorded. To take full advantage of these information-rich datasets, TIMA utilizes a new archiving tool introduced by TESCAN. The dataset size can be reduced for long-term storage and all information can be recovered on-demand in case of renewed interest. TESCAN TIMA is optimized for network storage of its datasets because of the larger data storage capacity of servers compared to local drives, which also allows multiple users to access the data remotely. This goes hand in hand with the support of remote control for the entire data acquisition process. TESCAN also brings a newly extended open-source data format that allows other applications to extract, process and report AM data. This offers the ability to link TIMA data to large databases feeding plant performance dashboards or geometallurgical models. The traditional tabular particle-by-particle or grain-by-grain export process is preserved and can be customized with scripts to include user-defined particle/grain properties.

Keywords: Tescan, electron microscopy, mineralogy, SEM, automated mineralogy, database, TESCAN TIMA, open format, archiving, big data

Procedia PDF Downloads 109

Authors: K. Raghu, M. N. Vathhsala, Naveen Aradya, Sharth

Abstract:

River sand is the most preferred fine aggregate for mortar and concrete. River sand is a product of natural weathering of rocks over a period of millions of years and is mined from river beds. Sand mining has disastrous environmental consequences. The excessive mining of river bed is creating an ecological imbalance. This has lead to have restrictions imposed by ministry of environment on sand mining. Driven by the acute need for sand, stone dust or manufactured sand prepared from the crushing and screening of coarse aggregate is being used as sand in the recent past. However manufactured sand is also a natural material and has quarrying and quality issues. To reduce the burden on the environment, alternative materials to be used as fine aggregates are being extensively investigated all over the world. Looking to the quantum of requirements, quality and properties there has been a global consensus on a material – Granulated slags. Granulated slag has been proven as a suitable material for replacing natural sand / crushed fine aggregates. In developed countries, the use of granulated slag as fine aggregate to replace natural sand is well established and is in regular practice. In the present paper Granulated slag has been experimented for usage in mortar. Slags are the main by-products generated during iron and steel production in the steel industry. Over the past decades, the steel production has increased and, consequently, the higher volumes of by-products and residues generated which have driven to the reuse of these materials in an increasingly efficient way. In recent years new technologies have been developed to improve the recovery rates of slags. Increase of slags recovery and use in different fields of applications like cement making, construction and fertilizers help in preserving natural resources. In addition to the environment protection, these practices produced economic benefits, by providing sustainable solutions that can allow the steel industry to achieve its ambitious targets of “zero waste” in coming years. Slags are generated at two different stages of steel production, iron making and steel making known as BF(Blast Furnace) slag and steel slag respectively. The slagging agent or fluxes, such as lime stone, dolomite and quartzite added into BF or steel making furnaces in order to remove impurities from ore, scrap and other ferrous charges during smelting. The slag formation is the result of a complex series of physical and chemical reactions between the non-metallic charge(lime stone, dolomite, fluxes), the energy sources(coal, coke, oxygen, etc.) and refractory materials. Because of the high temperatures (about 15000 C) during their generation, slags do not contain any organic substances. Due to the fact that slags are lighter than the liquid metal, they float and get easily removed. The slags protect the metal bath from atmosphere and maintain temperature through a kind of liquid formation. These slags are in liquid state and solidified in air after dumping in the pit or granulated by impinging water systems. Generally, BF slags are granulated and used in cement making due to its high cementious properties, and steel slags are mostly dumped due to unfavourable physio-chemical conditions. The increasing dump of steel slag not only occupies a plenty of land but also wastes resources and can potentially have an impact on the environment due to water pollution. Since BF slag contains little Fe and can be used directly. BF slag has found a wide application, such as cement production, road construction, Civil Engineering work, fertilizer production, landfill daily cover, soil reclamation, prior to its application outside the iron and steel making process.

Keywords: steel slag, river sand, granulated slag, environmental

Procedia PDF Downloads 244

Authors: Paola Causin, Andrea Aspri, Alessandro Benfenati

Abstract:

Diffuse Optical Tomography (DOT) is an emergent medical imaging technique which employs NIR light to estimate the spatial distribution of optical coefficients in biological tissues for diagnostic purposes, in a noninvasive and non-ionizing manner. DOT reconstruction is a severely ill-conditioned problem due to prevalent scattering of light in the tissue. In this contribution, we present our research in adopting hybrid knowledgedriven/data-driven approaches which exploit the existence of well assessed physical models and build upon them neural networks integrating the availability of data. Namely, since in this context regularization procedures are mandatory to obtain a reasonable reconstruction [1], we explore the use of neural networks as tools to include prior information on the solution. 2. Materials and Methods The idea underlying our approach is to leverage neural networks to solve PDE-constrained inverse problems of the form 𝒒 ∗ = 𝒂𝒓𝒈 𝒎𝒊𝒏𝒒 𝐃(𝒚, 𝒚̃), (1) where D is a loss function which typically contains a discrepancy measure (or data fidelity) term plus other possible ad-hoc designed terms enforcing specific constraints. In the context of inverse problems like (1), one seeks the optimal set of physical parameters q, given the set of observations y. Moreover, 𝑦̃ is the computable approximation of y, which may be as well obtained from a neural network but also in a classic way via the resolution of a PDE with given input coefficients (forward problem, Fig.1 box ). Due to the severe ill conditioning of the reconstruction problem, we adopt a two-fold approach: i) we restrict the solutions (optical coefficients) to lie in a lower-dimensional subspace generated by auto-decoder type networks. This procedure forms priors of the solution (Fig.1 box ); ii) we use regularization procedures of type 𝒒̂ ∗ = 𝒂𝒓𝒈𝒎𝒊𝒏𝒒 𝐃(𝒚, 𝒚̃)+ 𝑹(𝒒), where 𝑹(𝒒) is a regularization functional depending on regularization parameters which can be fixed a-priori or learned via a neural network in a data-driven modality. To further improve the generalizability of the proposed framework, we also infuse physics knowledge via soft penalty constraints (Fig.1 box ) in the overall optimization procedure (Fig.1 box ). 3. Discussion and Conclusion DOT reconstruction is severely hindered by ill-conditioning. The combined use of data-driven and knowledgedriven elements is beneficial and allows to obtain improved results, especially with a restricted dataset and in presence of variable sources of noise.

Keywords: inverse problem in tomography, deep learning, diffuse optical tomography, regularization

Procedia PDF Downloads 74

Authors: Ahmad Odeh, Ahmad Jrade

Abstract:

Energy assessment is an evidently significant factor when evaluating the sustainability of structures especially at the early design stage. Today design practices revolve around the selection of material that reduces the operational energy and yet meets their displinary need. Operational energy represents a substantial part of the building lifecycle energy usage but the fact remains that embodied energy is an important aspect unaccounted for in the carbon footprint. At the moment, little or no consideration is given to embodied energy mainly due to the complexity of calculation and the various factors involved. The equipment used, the fuel needed, and electricity required for each material vary with location and thus the embodied energy will differ for each project. Moreover, the method and the technique used in manufacturing, transporting and putting in place will have a significant influence on the materials’ embodied energy. This anomaly has made it difficult to calculate or even bench mark the usage of such energies. This paper presents a model aimed at helping designers select the construction materials based on their embodied energy. Moreover, this paper presents a systematic approach that uses an efficient method of calculation and ultimately provides new insight into construction material selection. The model is developed in a BIM environment targeting the quantification of embodied energy for construction materials through the three main stages of their life: manufacturing, transportation and placement. The model contains three major databases each of which contains a set of the most commonly used construction materials. The first dataset holds information about the energy required to manufacture any type of materials, the second includes information about the energy required for transporting the materials while the third stores information about the energy required by tools and cranes needed to place an item in its intended location. The model provides designers with sets of all available construction materials and their associated embodied energies to use for the selection during the design process. Through geospatial data and dimensional material analysis, the model will also be able to automatically calculate the distance between the factories and the construction site. To remain within the sustainability criteria set by LEED, a final database is created and used to calculate the overall construction cost based on R.M.S. means cost data and then automatically recalculate the costs for any modifications. Design criteria including both operational and embodied energies will cause designers to revaluate the current material selection for cost, energy, and most importantly sustainability.

Keywords: building information modelling, energy, life cycle analysis, sustainablity

Procedia PDF Downloads 269

Authors: Benoît Leclercq, Ilse Depraetere

Abstract:

The goal of this presentation is to shed new light on the semantics and pragmatics of be able to. It presents the results of a corpus analysis based on data from the BNC (British National Corpus), and discusses these results in light of a specific stance on the semantics-pragmatics interface taking into account recent developments. Be able to is often discussed in relation to can and could, all of which can be used to express ability. Such an onomasiological approach often results in the identification of usage constraints for each expression. In the case of be able to, it is the formal properties of the modal expression (unlike can and could, be able to has non-finite forms) that are in the foreground, and the modal expression is described as the verb that conveys future ability. Be able to is also argued to expressed actualised ability in the past (I was able/could to open the door). This presentation aims to provide a more accurate pragmatic-semantic profile of be able to, based on extensive data analysis and one that is embedded in a very explicit view on the semantics-pragmatics interface. A random sample of 3000 examples (1000 for each modal verb) extracted from the BNC was analysed to account for the following issues. First, the challenge is to identify the exact semantic range of be able to. The results show that, contrary to general assumption, be able to does not only express ability but it shares most of the root meanings usually associated with the possibility modals can and could. The data reveal that what is called opportunity is, in fact, the most frequent meaning of be able to. Second, attention will be given to the notion of actualisation. It is commonly argued that be able to is the preferred form when the residue actualises: (1) The only reason he was able to do that was because of the restriction (BNC, spoken) (2) It is only through my imaginative shuffling of the aces that we are able to stay ahead of the pack. (BNC, written) Although this notion has been studied in detail within formal semantic approaches, empirical data is crucially lacking and it is unclear whether actualisation constitutes a conventional (and distinguishing) property of be able to. The empirical analysis provides solid evidence that actualisation is indeed a conventional feature of the modal. Furthermore, the dataset reveals that be able to expresses actualised 'opportunities' and not actualised 'abilities'. In the final part of this paper, attention will be given to the theoretical implications of the empirical findings, and in particular to the following paradox: how can the same expression encode both modal meaning (non-factual) and actualisation (factual)? It will be argued that this largely depends on one's conception of the semantics-pragmatics interface, and that this need not be an issue when actualisation (unlike modality) is analysed as a generalised conversational implicature and thus is considered part of the conventional pragmatic layer of be able to.

Keywords: Actualisation, Modality, Pragmatics, Semantics

Procedia PDF Downloads 131

Authors: Martín Pratto Burgos

Abstract:

The Engineering School of the University of the Republic in Uruguay offers an Introductory Mathematical Course from the second semester of 2019. This course has been designed to assist students in preparing themselves for math courses that are essential for Engineering Degrees, namely Math1, Math2, and Math3 in this research. The research proposes to build a model that can accurately predict the student's activity and academic progress based on their performance in the three essential Mathematical courses. Additionally, there is a need for a model that can forecast the incidence of the Introductory Mathematical Course in the three essential courses approval during the first academic year. The techniques used are Principal Component Analysis and predictive modelling using the Generalised Linear Model. The dataset includes information from 5135 engineering students and 12 different characteristics based on activity and course performance. Two models are created for a type of data that follows a binomial distribution using the R programming language. Model 1 is based on a variable's p-value being less than 0.05, and Model 2 uses the stepAIC function to remove variables and get the lowest AIC score. After using Principal Component Analysis, the main components represented in the y-axis are the approval of the Introductory Mathematical Course, and the x-axis is the approval of Math1 and Math2 courses as well as student activity three years after taking the Introductory Mathematical Course. Model 2, which considered student’s activity, performed the best with an AUC of 0.81 and an accuracy of 84%. According to Model 2, the student's engagement in school activities will continue for three years after the approval of the Introductory Mathematical Course. This is because they have successfully completed the Math1 and Math2 courses. Passing the Math3 course does not have any effect on the student’s activity. Concerning academic progress, the best fit is Model 1. It has an AUC of 0.56 and an accuracy rate of 91%. The model says that if the student passes the three first-year courses, they will progress according to the timeline set by the curriculum. Both models show that the Introductory Mathematical Course does not directly affect the student’s activity and academic progress. The best model to explain the impact of the Introductory Mathematical Course on the three first-year courses was Model 1. It has an AUC of 0.76 and 98% accuracy. The model shows that if students pass the Introductory Mathematical Course, it will help them to pass Math1 and Math2 courses without affecting their performance on the Math3 course. Matching the three predictive models, if students pass Math1 and Math2 courses, they will stay active for three years after taking the Introductory Mathematical Course, and also, they will continue following the recommended engineering curriculum. Additionally, the Introductory Mathematical Course helps students to pass Math1 and Math2 when they start Engineering School. Models obtained in the research don't consider the time students took to pass the three Math courses, but they can successfully assess courses in the university curriculum.

Keywords: machine-learning, engineering, university, education, computational models

Procedia PDF Downloads 94

Authors: Annamarie Rannou, Horace Bartilow

Abstract:

Global concerns about xenophobic activity are on the rise across developed and developing countries. And yet, social science scholarship has almost exclusively examined xenophobia as a prejudice of advanced western nations. This research argues that the fields of study related to xenophobia must be re-conceptualized within a framework of ethnicity in order to level the playing field for cross-regional inquiry. This study develops a new concept of ethnic xenophobia and integrates existing explanations of anti-migrant expression into theories of ethnic threat. We argue specifically that political elites convert economic, political, and social threats at the national level into ethnic xenophobic activity in order to gain or maintain political advantage among their native selectorate. We expand on Stuart Kaufman’s theory of symbolic politics to underscore the methods of mobilization used against migrants and the power of elite discourse in moments of national crises. An original dataset is used to examine over 35,000 cases of ethnic xenophobic activity targeting refugees. Wordscores software is used to develop a unique measure of anti-migrant elite rhetoric which captures the symbolic discourse of elites in their mobilization of ethnic xenophobic activism. We use a Structural Equation Model (SEM) to test the causal pathways of the theory across seventy-two developed and developing countries from 1990 to 2016. A framework of Most Different Systems Design (MDSD) is also applied to two pairs of developed-developing country cases, including Kenya and the Netherlands and Lebanon and the United States. This study sheds tremendous light on an underrepresented area of comparative research in migration studies. It shows that the causal elements of anti-migrant activity are far more similar than existing research suggests which has major implications for policy makers, practitioners, and academics in fields of migration protection and advocacy. It speaks directly to the mobilization of myths surrounding refugees, in particular, and the nationalization of narratives of migration that may be neutralized by the development of deeper associational relationships between natives and migrants.

Keywords: refugees, ethnicity, symbolic politics, elites, migration, comparative politics

Procedia PDF Downloads 145