Search results for: multiwavelenght dataset

Authors: Alexander Goncharenko, Andrey Denisov, Sergey Alyamkin, Evgeny Terentev

Abstract:

The neural network quantization is highly desired procedure to perform before running neural networks on mobile devices. Quantization without fine-tuning leads to accuracy drop of the model, whereas commonly used training with quantization is done on the full set of the labeled data and therefore is both time- and resource-consuming. Real life applications require simplification and acceleration of quantization procedure that will maintain accuracy of full-precision neural network, especially for modern mobile neural network architectures like Mobilenet-v1, MobileNet-v2 and MNAS. Here we present a method to significantly optimize training with quantization procedure by introducing the trained scale factors for discretization thresholds that are separate for each filter. Using the proposed technique, we quantize the modern mobile architectures of neural networks with the set of train data of only ∼ 10% of the total ImageNet 2012 sample. Such reduction of train dataset size and small number of trainable parameters allow to fine-tune the network for several hours while maintaining the high accuracy of quantized model (accuracy drop was less than 0.5%). Ready-for-use models and code are available in the GitHub repository.

Keywords: distillation, machine learning, neural networks, quantization

Procedia PDF Downloads 309

Authors: Yin Yuanling

Abstract:

A Deep Feedback Recurrent Neural Network (DFRNN) and Bidirectional Long Short-Term Memory (BiLSTM) are designed to address the problem of low accuracy of traditional relationship extraction models. This method combines a deep feedback-based recurrent neural network (DFRNN) with a bi-directional long short-term memory (BiLSTM) approach. The method combines DFRNN, which extracts local features of text based on deep feedback recurrent mechanism, BiLSTM, which better extracts global features of text, and Self-Attention, which extracts semantic information. Experiments show that the method achieves an F1 value of 76.69% on the CEC dataset, which is 0.0652 better than the BiLSTM+Self-ATT model, thus optimizing the performance of the deep learning method in the event relationship extraction task.

Keywords: event relations, deep learning, DFRNN models, bi-directional long and short-term memory networks

Procedia PDF Downloads 124

Authors: Smitha Haridasan, Zelalem Demissie, Atri Dutta, Ajita Rattani

Abstract:

Aided by the wind, all it takes is one ember and a few minutes to create a wildfire. Wildfires are growing in frequency and size due to climate change. Wildfires and its consequences are one of the major environmental concerns. Every year, millions of hectares of forests are destroyed over the world, causing mass destruction and human casualties. Thus early detection of wildfire becomes a critical component to mitigate this threat. Many computer vision-based techniques have been proposed for the early detection of forest fire using video surveillance. Several computer vision-based methods have been proposed to predict and detect forest fires at various spectrums, namely, RGB, HSV, and YCbCr. The aim of this paper is to propose a multi-spectral deep learning model that combines information from different spectrums at intermediate layers for accurate fire detection. A heterogeneous dataset assembled from publicly available datasets is used for model training and evaluation in this study. The experimental results show that multi-spectral deep learning models could obtain an improvement of about 4.68 % over those based on a single spectrum for fire detection.

Keywords: deep learning, forest fire detection, multi-spectral learning, natural hazard detection

Procedia PDF Downloads 228

Authors: Juan Sebastián Hernández

Abstract:

The Nevado del Ruíz volcano, located between the limits of the Departments of Caldas and Tolima in Colombia, presented an unstable behaviour in the course of the year 2020, this volcanic activity led to secondary effects on the crust, which is why the prediction of deformations becomes the task of geoscientists. In the course of this article, the use of tropospheric variables such as evapotranspiration, UV aerosol index, carbon monoxide, nitrogen dioxide, methane, surface temperature, among others, is used to train a set of neural networks that can predict the behaviour of the resulting phase of an unrolled interferogram with the DInSAR technique, whose main objective is to identify and characterise the behaviour of the crust based on the environmental conditions. For this purpose, variables were collected, a generalised linear model was created, and a set of neural networks was created. After the training of the network, validation was carried out with the test data, giving an MSE of 0.17598 and an associated r-squared of approximately 0.88454. The resulting model provided a dataset with good thematic accuracy, reflecting the behaviour of the volcano in 2020, given a set of environmental characteristics.

Keywords: crustal deformation, Tropomi, neural networks (ANN), volcanic activity, DInSAR

Procedia PDF Downloads 87

Authors: Imen Elleuch, Wael Ouarda, Gargouri Bilel

Abstract:

We proposes in this paper a Deep Learning based approach to classify text in order to enrich an Arabic ontology based on the objects classes of Gaston Gross. Those object classes are defined by taking into account the syntactic and semantic features of the treated language. Thus, our proposed approach is a hybrid one. In fact, it is based on the one hand on the object classes that represents a knowledge based-approach on classification of text and in the other hand it uses the deep learning approach that use the word embedding-based-approach to classify text. We have applied our proposed approach on a corpus constructed from an Arabic dictionary. The obtained semantic classification of text will enrich the Arabic objects classes ontology. In fact, new classes can be added to the ontology or an expansion of the features that characterizes each object class can be updated. The obtained results are compared to a similar work that treats the same object with a classical linguistic approach for the semantic classification of text. This comparison highlight our hybrid proposed approach that can be ameliorated by broaden the dataset used in the deep learning process.

Keywords: deep-learning approach, object-classes, semantic classification, Arabic

Procedia PDF Downloads 63

Authors: Graciela Soares Marcelli, Elison Eduardo Jardim Bierhals, Luciane Teresa Salvi, Claudineia Brazil, Rafael Haag

Abstract:

The sun is a source of renewable energy, and its use as both a source of heat and light is one of the most promising energy alternatives for the future. To measure the thermal or photovoltaic systems a solar irradiation database is necessary. Brazil still has a reduced number of meteorological stations that provide frequency tests, as an alternative to the radio data platform, with reanalysis systems, quite significant. ERA-Interim is a global fire reanalysis by the European Center for Medium-Range Weather Forecasts (ECMWF). The data assimilation system used for the production of ERA-Interim is based on a 2006 version of the IFS (Cy31r2). The system includes a 4-dimensional variable analysis (4D-Var) with a 12-hour analysis window. The spatial resolution of the dataset is approximately 80 km at 60 vertical levels from the surface to 0.1 hPa. This work aims to make a comparative analysis between the ERA-Interim data and the data observed in the Solarimmetric Atlas of the State of Rio Grande do Sul, to verify its applicability in the absence of an observed data network. The analysis of the results obtained for a study region as an alternative to the energy potential of a given region.

Keywords: energy potential, reanalyses, renewable energy, solar radiation

Procedia PDF Downloads 151

Authors: Hung Nguyen

Abstract:

With the rise of global supply chain network, the choice of supply chain orientation is critical. The alignment between cultural similarity and supply chain information exchange could help identify appropriate supply chain orientations, which would differentiate the stronger competitors and performers from the weaker ones. Through developing a taxonomy, this study examined whether the choices of action programs and manufacturing performance differ depending on the levels of attainment cultural similarity and information exchange. This study employed statistical tests on a large-scale dataset consisting of 680 manufacturing plants from various cultures and industries. Firms need to align cultural practices with the level of information exchange in order to achieve good overall business performance. There appeared to be consistent three major orientations: the Proactive, the Initiative and the Reactive. Firms are experiencing higher payoffs from various improvements are the ones successful alignment in both information exchange and cultural similarity The findings provide step-by-step decision making for supply chain information exchange and offer guidance especially for global supply chain managers. In including both cultural similarity and information exchange, this paper adds greater comprehensiveness and richness to the supply chain literature.

Keywords: culture, information exchange, supply chain orientation, similarity

Procedia PDF Downloads 347

Authors: Md. Shahriar Sabuz

Abstract:

Bangladesh, a South Asian developing country, has experienced an increasing rate of contraceptive use in the last few decades. But one-third of the pregnancies are still unintended, and the fertility rate surpasses the desired rate of children. It may be because of the discontinuation of the use of contraceptive methods. So, it is necessary to find out the reasons for the discontinuation of the use of contraceptives. Moreover, the rate of contraception discontinuation varies from rural to urban, region to region. In this study, our objectives are to find out the reasons behind the discontinuation of the use of the contraceptive method, and the regional variations of the rate of those reasons. We are using the dataset of Bangladesh Demographic and Health Surveys (BDHS) 2014 for this study and the ever-married women of Bangladesh who have discontinued the use of contraceptive methods aged 15-49. The data was collected from the seven districts of the country. The finding shows that currently there are 23% of women have stopped using their contraception. The most common reasons for stopping using the method are that either they are pregnant or want to be pregnant. A significant number of people are not using the contraceptive method because of the fear of side effects. Though the rate of non-user is higher in rural areas than in urban areas, reasons for method discontinuation are not significantly different between urban and rural areas. However, reasons for discontinuing contraceptive methods significantly vary from region to region.

Keywords: discontinuation of contraceptive, health, pregnant, fertility

Procedia PDF Downloads 86

Authors: Lu Zhao, Nadir Farhi, Zoi Christoforou, Nadia Haddadou

Abstract:

The safety test of deploying intelligent connected vehicles (ICVs) on the road network is a critical challenge. Road traffic network simulation can be used to test the functionality of ICVs, which is not only time-saving and less energy-consuming but also can create scenarios with car collisions. However, the relationship between different human driver behaviors and the car-collision occurrences has been not understood clearly; meanwhile, the procedure of car-collisions generation in the traffic numerical simulators is not fully integrated. In this paper, we propose an approach to identify specific driver profiles from real driven data; then, we replicate them in numerical traffic simulations with the purpose of generating inter-vehicular collisions. We proposed three profiles: (i) 'aggressive': short time-headway, (ii) 'inattentive': long reaction time, and (iii) 'normal' with intermediate values of reaction time and time-headway. These three driver profiles are extracted from the NGSIM dataset and simulated using the intelligent driver model (IDM), with an extension of reaction time. At last, the generation of inter-vehicular collisions is performed by varying the percentages of different profiles.

Keywords: vehicular collisions, human driving behavior, traffic modeling, car-following models, microscopic traffic simulation

Procedia PDF Downloads 164

Authors: Prasanna Ramachandran, Kareem Graham, Helena Kiefel, Sunit Jain, Todd DeSantis

Abstract:

Microbes that reside inside the mammalian GI tract, commonly referred to as the gut microbiome, have been shown to have therapeutic effects in animal models of disease. We hypothesize that specific proteins produced by these microbes are responsible for this activity and may be used directly as therapeutics. To speed up the discovery of these key proteins from the big-data metagenomics, we have applied machine learning techniques. Using amino acid sequences of known epitopes and their corresponding binding partners, protein interaction descriptors (PID) were calculated, making a positive interaction set. A negative interaction dataset was calculated using sequences of proteins known not to interact with these same binding partners. Using Random Forest and positive and negative PID, a machine learning model was trained and used to predict interacting versus non-interacting proteins. Furthermore, the continuous variable, cosine similarity in the interaction descriptors was used to rank bacterial therapeutic candidates. Laboratory binding assays were conducted to test the candidates for their potential as therapeutics. Results from binding assays reveal the accuracy of the machine learning prediction and are subsequently used to further improve the model.

Keywords: protein-interactions, machine-learning, metagenomics, microbiome

Procedia PDF Downloads 363

Authors: Abebe Mulugojam Negash, Yongbin Yu, Ekong Favour, Bekalu Nigus Dawit, Molla Woretaw Teshome, Aynalem Birtukan Yirga

Abstract:

Breast cancer remains a significant health concern, necessitating advancements in diagnostic methodologies. Addressing this, our paper confronts the notable challenges in breast cancer classification, particularly the imbalance in datasets and the constraints in the accuracy and interpretability of prevailing deep learning approaches. We proposed an attention-based residual neural network (ResNet), which effectively combines the robust features of ResNet with an advanced attention mechanism. Enhanced through strategic data augmentation and positive weight adjustments, this approach specifically targets the issue of data imbalance. The proposed model is tested on the BreakHis dataset and achieved accuracies of 99.00%, 99.04%, 98.67%, and 98.08% in different magnifications (40X, 100X, 200X, and 400X), respectively. We evaluated the performance by using different evaluation metrics such as precision, recall, and F1-Score and made comparisons with other state-of-the-art methods. Our experiments demonstrate that the proposed model outperforms existing approaches, achieving higher accuracy in breast cancer classification.

Keywords: residual neural network, attention mechanism, positive weight, data augmentation

Procedia PDF Downloads 77

Authors: Heng Yang, Tao Luo, Yakun Zhang, Kai Wang, Wei Qin, Liang Xie, Ye Yan, Erwei Yin

Abstract:

Recent years have seen that audio-visual recognition has shown great potential in a strong noise environment. The existing method of audio-visual recognition has explored methods with ResNet and feature fusion. However, on the one hand, ResNet always occupies a large amount of memory resources, restricting the application in engineering. On the other hand, the feature merging also brings some interferences in a high noise environment. In order to solve the problems, we proposed an effective framework with bidirectional distillation. At first, in consideration of the good performance in extracting of features, we chose the light model, Efficientnet as our extractor of spatial features. Secondly, self-distillation was applied to learn more information from raw data. Finally, we proposed a bidirectional distillation in decision-level fusion. In more detail, our experimental results are based on a multi-model dataset from 24 volunteers. Eventually, the lipreading accuracy of our framework was increased by 2.3% compared with existing systems, and our framework made progress in audio-visual fusion in a high noise environment compared with the system of audio recognition without visual.

Keywords: lipreading, audio-visual, Efficientnet, distillation

Procedia PDF Downloads 121

Authors: Elisa Coraggio, Dawei Han, Weiru Liu, Theo Tryfonas

Abstract:

Water is one of the most important resources for human society. The world is currently undergoing a wave of urban growth, and pollution problems are of a great impact. Monitoring water quality is a key task for the future of the environment and human species. In recent times, researchers, using Smart Cities technologies are trying to mitigate the problems generated by the population growth in urban areas. The availability of huge amounts of data collected by a pervasive urban IoT can increase the transparency of decision making. Several services have already been implemented in Smart Cities, but more and more services will be involved in the future. Water quality monitoring can successfully be implemented in the urban IoT. The combination of water quality sensors, cloud computing, smart city infrastructure, and IoT technology can lead to a bright future for environmental monitoring. In the past decades, lots of effort has been put on monitoring and predicting water quality using traditional approaches based on manual collection and laboratory-based analysis, which are slow and laborious. The present study proposes a methodology for implementing a water quality prediction model using artificial intelligence techniques and comparing the results obtained with different algorithms. Furthermore, a 3D numerical model will be created using the software D-Water Quality, and simulation results will be used as a training dataset for the artificial intelligence algorithm. This study derives the methodology and demonstrates its implementation based on information and data collected at the floating harbour in the city of Bristol (UK). The city of Bristol is blessed with the Bristol-Is-Open infrastructure that includes Wi-Fi network and virtual machines. It was also named the UK ’s smartest city in 2017.In recent times, researchers, using Smart Cities technologies are trying to mitigate the problems generated by the population growth in urban areas. The availability of huge amounts of data collected by a pervasive urban IoT can increase the transparency of decision making. Several services have already been implemented in Smart Cities, but more and more services will be involved in the future. Water quality monitoring can successfully be implemented in the urban IoT. The combination of water quality sensors, cloud computing, smart city infrastructure, and IoT technology can lead to a bright future for the environment monitoring. In the past decades, lots of effort has been put on monitoring and predicting water quality using traditional approaches based on manual collection and laboratory-based analysis, which are slow and laborious. The present study proposes a new methodology for implementing a water quality prediction model using artificial intelligence techniques and comparing the results obtained with different algorithms. Furthermore, a 3D numerical model will be created using the software D-Water Quality, and simulation results will be used as a training dataset for the Artificial Intelligence algorithm. This study derives the methodology and demonstrate its implementation based on information and data collected at the floating harbour in the city of Bristol (UK). The city of Bristol is blessed with the Bristol-Is-Open infrastructure that includes Wi-Fi network and virtual machines. It was also named the UK ’s smartest city in 2017.

Keywords: artificial intelligence, hydroinformatics, numerical modelling, smart cities, water quality

Procedia PDF Downloads 169

Authors: Joseph Wolff, Jeffrey Eilbott

Abstract:

Autism Spectrum Disorders (ASD) are a spectrum of social disorders characterized by deficits in social communication, verbal ability, and interaction that can vary in severity. In recent years, researchers have used magnetic resonance imaging (MRI) to help detect how neural patterns in individuals with ASD differ from those of neurotypical (NT) controls for classification purposes. This study analyzed the classification of ASD within males and females using functional MRI data. Functional connectivity (FC) correlations among brain regions were used as feature inputs for machine learning algorithms. Analysis was performed on 558 cases from the Autism Brain Imaging Data Exchange (ABIDE) I dataset. When trained specifically on females, the algorithm underperformed in classifying the ASD subset of our testing population. Although the subject size was relatively smaller in the female group, the manual matching of both male and female training groups helps explain the algorithm’s bias, indicating the altered sex abnormalities in functional brain networks compared to typically developing peers. These results highlight the importance of taking sex into account when considering how generalizations of findings on males with ASD apply to females.

Keywords: autism spectrum disorder, machine learning, neuroimaging, sex differences

Procedia PDF Downloads 198

Authors: Kennedy Efosa Ehimwenma, Sujatha Krishnamoorthy, Safiya Al‑Sharji

Abstract:

The probability computation of events is in the interval of [0, 1], which are values that are determined by the number of outcomes of events in a sample space S. The probability Pr(A) that an event A will never occur is 0. The probability Pr(B) that event B will certainly occur is 1. This makes both events A and B a certainty. Furthermore, the sum of probabilities Pr(E₁) + Pr(E₂) + … + Pr(Eₙ) of a finite set of events in a given sample space S equals 1. Conversely, the difference of the sum of two probabilities that will certainly occur is 0. This paper first discusses Bayes, the complement of probability, and the difference of probability for occurrences of learning-events before applying them in the prediction of learning objects in student learning. Given the sum of 1; to make a recommendation for student learning, this paper proposes that the difference of argMaxPr(S) and the probability of student-performance quantifies the weight of learning objects for students. Using a dataset of skill-set, the computational procedure demonstrates i) the probability of skill-set events that have occurred that would lead to higher-level learning; ii) the probability of the events that have not occurred that requires subject-matter relearning; iii) accuracy of the decision tree in the prediction of student performance into class labels and iv) information entropy about skill-set data and its implication on student cognitive performance and recommendation of learning.

Keywords: complement of probability, Bayes’ rule, prediction, pre-assessments, computational education, information theory

Procedia PDF Downloads 146

Authors: Tamanna Goyal, Divya Bansal, Sanjeev Sofat

Abstract:

In world-threatening terrorist attacks, where early detection, distinction, and prediction are effective diagnosis techniques and for functionally accurate and precise analysis of terrorism data, there are so many data mining & statistical approaches to assure accuracy. The computational extraction of derived patterns is a non-trivial task which comprises specific domain discovery by means of sophisticated algorithm design and analysis. This paper proposes an approach for similarity extraction by obtaining the useful attributes from the available datasets of terrorist attacks and then applying feature selection technique based on the statistical impurity measures followed by clustering techniques on the basis of similarity measures. On the basis of degree of participation of attributes in the rules, the associative dependencies between the attacks are analyzed. Consequently, to compute the similarity among the discovered rules, we applied a weighted similarity measure. Finally, the rules are grouped by applying using hierarchical clustering. We have applied it to an open source dataset to determine the usability and efficiency of our technique, and a literature search is also accomplished to support the efficiency and accuracy of our results.

Keywords: association rules, clustering, similarity measure, statistical approaches

Procedia PDF Downloads 308

Authors: Merle Van Den Akker, Neil Stewart, Andrea Isoni

Abstract:

Contactless methods of payment referred to as tap&go, have become increasingly popular globally. However, little is known about the consequences of this payment method on spending, spending habits, personal finance management, and debt accumulation. The literature on other payment methods such as credit cards suggests that, through increased ease and reduced friction, the pain of paying in these methods is reduced, leading to higher and more frequent spending, resulting in higher debt accumulation. Within this research, we use a dataset of 300 million transactions of 165.000 individuals to see whether the onset of using contactless methods of payment increases the occurrence of overdraft fees. Using the R package MatchIt, we find, when matching people on initial overdraft occurrence and salary, that people who do start using contactless incur a significantly higher number of overdraft fees, as compared to those who do not start using contactless in the same year. Having accounted for income, opting-in, and time-of-year effects, these results show that contactless methods of payment fall within the scope of earlier theories on credit cards, such as the pain of paying, meaning that this payment method leads to increasing difficulties managing personal finance.

Keywords: contactless, debt accumulation, overdraft fees, payment methods, spending

Procedia PDF Downloads 112

Authors: Yuntao Liu, Lei Wang, Haoran Xia

Abstract:

Machine learning has shown extensive applications in the development of classification models for autism spectrum disorder (ASD) using neural image data. This paper proposes a fusion multi-modal classification network based on a graph neural network. First, the brain is segmented into 116 regions of interest using a medical segmentation template (AAL, Anatomical Automatic Labeling). The image features of sMRI and the signal features of fMRI are extracted, which build the node and edge embedding representations of the brain map. Then, we construct a dynamically updated brain map neural network and propose a method based on a dynamic brain map adjacency matrix update mechanism and learnable graph to further improve the accuracy of autism diagnosis and recognition results. Based on the Autism Brain Imaging Data Exchange I dataset(ABIDE I), we reached a prediction accuracy of 74% between ASD and TD subjects. Besides, to study the biomarkers that can help doctors analyze diseases and interpretability, we used the features by extracting the top five maximum and minimum ROI weights. This work provides a meaningful way for brain disorder identification.

Keywords: autism spectrum disorder, brain map, supervised machine learning, graph network, multimodal data, model interpretability

Procedia PDF Downloads 45

Authors: Mustofa Mahmud Al Mamun

Abstract:

Government spending, one of the major components of gross domestic product (GDP), is composed of government consumption, investment, and transfer payments. A change in government spending during recessionary periods can generate an increase in GDP greater than the increase in spending. This is called the "multiplier effect". Accurate estimation of government spending multiplier is important because fiscal policy has been used to stimulate a flagging economy. Many recent studies have focused on identifying parts of the economy that responds more to a stimulus under a variety of circumstances. This paper used the US dataset from 1966 to 2020 and local projection method assuming standard identification strategy to estimate the multipliers. The model includes important macroaggregates and controls for forecasted government spending, interest rate, consumer price index (CPI), export, import, and level of public debt. Investment multipliers are found to be positive and larger than the consumption multipliers. Consumption multipliers are either negative or not significantly different than zero. Results do not vary across the business cycle. However, the consumption multiplier estimated from pre-1980 data is positive.

Keywords: business cycle, consumption multipliers, forecasted government spending, investment multipliers, local projection method, zero lower bound

Procedia PDF Downloads 216

Authors: Nihad Karim Chowdhury, Mohammad Sanaullah Chowdhury, Muhammed Jamshed Alam Patwary, Rubel Biswas

Abstract:

In content-based 3D shape retrieval system, achieving high search performance has become an important research problem. A challenging aspect of this problem is to find an effective shape descriptor which can discriminate similar shapes adequately. To address this problem, we propose a new shape descriptor for 3D shape models by combining multi-resolution with multi-block center-symmetric local binary pattern operator. Given an arbitrary 3D shape, we first apply pose normalization, and generate a set of multi-viewed 2D rendered images. Second, we apply Gaussian multi-resolution filter to generate several levels of images from each of 2D rendered image. Then, overlapped sub-images are computed for each image level of a multi-resolution image. Our unique multi-block CS-LBP comes next. It allows the center to be composed of m-by-n rectangular pixels, instead of a single pixel. This process is repeated for all the 2D rendered images, derived from both ‘depth-buffer’ and ‘silhouette’ rendering. Finally, we concatenate all the features vectors into one dimensional histogram as our proposed 3D shape descriptor. Through several experiments, we demonstrate that our proposed 3D shape descriptor outperform the previous methods by using a benchmark dataset.

Keywords: 3D shape retrieval, 3D shape descriptor, CS-LBP, overlapped sub-images

Procedia PDF Downloads 433

Authors: Marjan Golmaryami, Marzieh Behzadi

Abstract:

Recent developments in stock exchange highlight the need for an efficient and accurate method that helps stockholders make better decision. Since stock markets have lots of fluctuations during the time and different effective parameters, it is difficult to make good decisions. The purpose of this study is to employ artificial neural network (ANN) which can deal with time series data and nonlinear relation among variables to forecast next day stock price. Unlike other evolutionary algorithms which were utilized in stock exchange prediction, we trained our proposed neural network with metaheuristic bat algorithm, with fast and powerful convergence and applied it in stock price prediction for the first time. In order to prove the performance of the proposed method, this research selected a 7 year dataset from Parsian Bank stocks and after imposing data preprocessing, used 3 types of ANN (back propagation-ANN, particle swarm optimization-ANN and bat-ANN) to predict the closed price of stocks. Afterwards, this study engaged MATLAB to simulate 3 types of ANN, with the scoring target of mean absolute percentage error (MAPE). The results may be adapted to other companies stocks too.

Keywords: artificial neural network (ANN), bat algorithm, particle swarm optimization algorithm (PSO), stock exchange

Procedia PDF Downloads 534

Authors: Rosdyana Mangir Irawan Kusuma, Wei-Chun Kao, Ho-Thi Trang, Yu-Yen Ou, Kai-Lung Hua

Abstract:

Stock market prediction is still a challenging problem because there are many factors that affect the stock market price such as company news and performance, industry performance, investor sentiment, social media sentiment, and economic factors. This work explores the predictability in the stock market using deep convolutional network and candlestick charts. The outcome is utilized to design a decision support framework that can be used by traders to provide suggested indications of future stock price direction. We perform this work using various types of neural networks like convolutional neural network, residual network and visual geometry group network. From stock market historical data, we converted it to candlestick charts. Finally, these candlestick charts will be feed as input for training a convolutional neural network model. This convolutional neural network model will help us to analyze the patterns inside the candlestick chart and predict the future movements of the stock market. The effectiveness of our method is evaluated in stock market prediction with promising results; 92.2% and 92.1 % accuracy for Taiwan and Indonesian stock market dataset respectively.

Keywords: candlestick chart, deep learning, neural network, stock market prediction

Procedia PDF Downloads 426

Authors: Cuong V Nguyen, Ralph Horne, John Fien, France Cheong

Abstract:

Social vulnerability to climate change is increasingly being acknowledged, and proposals to measure and manage it are emerging. Building upon this work, this paper proposes an approach to social vulnerability assessment using a new mechanism to aggregate and account for causal relationships among components of a Social Vulnerability Index (SVI). To operationalize this index, the authors propose a means to develop an appropriate primary dataset, through application of a specifically-designed household survey questionnaire. The data collection and analysis, including calibration and calculation of the SVI is demonstrated through application in case study city in central coastal Vietnam. The calculation of SVI at the fine-grained local neighbourhood scale provides high resolution in vulnerability assessment, and also obviates the need for secondary data, which may be unavailable or problematic, particularly at the local scale in developing countries. The SVI household survey is underpinned by the results of a Delphi survey, an in-depth interview and focus group discussions with local environmental professionals and community members. The research reveals inherent limitations of existing SVIs but also indicates the potential for their use in assessing social vulnerability and making decisions associated with responding to climate change at the local scale.

Keywords: climate change, local scale, social vulnerability, social vulnerability index

Procedia PDF Downloads 421

Authors: Amartya Hatua, Trung Nguyen, Andrew Sung

Abstract:

In this paper, we study the information diffusion process on Twitter as a multivariate time series problem. Our model concerns three measures (volume, network influence, and sentiment of tweets) based on 10 features, and we collected 27 million tweets to build our information diffusion time series dataset for analysis. Then, different time series clustering techniques with Dynamic Time Warping (DTW) distance were used to identify different patterns of information diffusion. Finally, we built the information diffusion prediction models for new hashtags which comprise two phrases: The first phrase is recognizing the pattern using k-NN with DTW distance; the second phrase is building the forecasting model using the traditional Autoregressive Integrated Moving Average (ARIMA) model and the non-linear recurrent neural network of Long Short-Term Memory (LSTM). Preliminary results of performance evaluation between different forecasting models show that LSTM with clustering information notably outperforms other models. Therefore, our approach can be applied in real-world applications to analyze and predict the information diffusion characteristics of selected topics or memes (hashtags) in Twitter.

Keywords: ARIMA, DTW, information diffusion, LSTM, RNN, time series clustering, time series forecasting, Twitter

Procedia PDF Downloads 380

Authors: Ali Reza Sattarzadeh, Ronny J. Kutadinata, Pubudu N. Pathirana, Van Thanh Huynh

Abstract:

Traffic congestion has become a severe worldwide problem, affecting everyday life, fuel consumption, time, and air pollution. The primary causes of these issues are inadequate transportation infrastructure, poor traffic signal management, and rising population. Traffic flow forecasting is one of the essential and effective methods in urban congestion and traffic management, which has attracted the attention of researchers. With the development of technology, undeniable progress has been achieved in existing methods. However, there is a possibility of improvement in the extraction of temporal and spatial features to determine the importance of traffic flow sequences and extraction features. In the proposed model, we implement the convolutional neural network (CNN) and long short-term memory (LSTM) deep learning models for mining nonlinear correlations and their effectiveness in increasing the accuracy of traffic flow prediction in the real dataset. According to the experiments, the results indicate that implementing Conv-LSTM networks increases the productivity and accuracy of deep learning models for traffic flow prediction.

Keywords: deep learning algorithms, intelligent transportation systems, spatiotemporal features, traffic flow prediction

Procedia PDF Downloads 154

Authors: Sara El Mansouria Beghdadi

Abstract:

Quantitative structure–activity relationship (QSAR) modelling is one of the main computer tools used in medicinal chemistry. Over the past two decades, the incidence of fungal infections has increased due to the development of resistance. In this study, the QSAR was performed on a series of esters of 2-carboxamido-3-(1H-imidazole-1-yl) propanoic acid derivatives. These compounds have showed moderate and very good antifungal activity. The multiple linear regression (MLR) was used to generate the linear 2d-QSAR models. The dataset consists of 115 compounds with their antifungal activity (log MIC) against «Candida albicans» (ATCC SC5314). Descriptors were calculated, and different models were generated using Chemoffice, Avogadro, GaussView software. The selected model was validated. The study suggests that the increase in lipophilicity and the reduction in the electronic character of the substituent in R1, as well as the reduction in the steric hindrance of the substituent in R2 and its aromatic character, supporting the potentiation of the antifungal effect. The results of QSAR could help scientists to propose new compounds with higher antifungal activities intended for immunocompromised patients susceptible to multi-resistant nosocomial infections.

Keywords: quantitative structure–activity relationship, imidazole, antifungal, candida albicans (ATCC SC5314)

Procedia PDF Downloads 67

Authors: Alireza Keshmiri, Shahriar Bagheri, Nan Wu

Abstract:

This research presents an analytical model for the development of an energy harvester with piezoelectric rings stacked at the boundary of the structure based on the Adomian decomposition method. The model is applied to geometrically non-uniform beams to derive the steady-state dynamic response of the structure subjected to base motion excitation and efficiently harvest the subsequent vibrational energy. The in-plane polarization of the piezoelectric rings is employed to enhance the electrical power output. A parametric study for the proposed energy harvester with various design parameters is done to prepare the dataset required for optimization. Finally, simulation-based optimization technique helps to find the optimum structural design with maximum efficiency. To solve the optimization problem, an artificial neural network is first trained to replace the simulation model, and then, a genetic algorithm is employed to find the optimized design variables. Higher geometrical non-uniformity and length of the beam lowers the structure natural frequency and generates a larger power output.

Keywords: piezoelectricity, energy harvesting, simulation-based optimization, artificial neural network, genetic algorithm

Procedia PDF Downloads 110

Authors: Md. Nakib Ul Hasan, Md. Ariful Islam, Md. Sumon Miah, Misbah Ul Hoque, Bulbul Ahmed

Abstract:

The variability of cotton fiber characteristics has always been influenced by origin, weather conditions, method of culturing, and harvesting. Spinners work tirelessly to ensure consistent yarn quality by using the different origins of fibers to maximizes the profit margin. Spinners often fail to select desired raw materials of various origins to achieve an appropriate mixing plan due to the lack of knowledge on the interrelationship among fiber properties. The purpose of this research is to investigate the correlations among dominating fiber properties such as micronaire, strength, breaking elongation, upper half mean length, length uniformity index, short fiber index, maturity, reflectance, and yellowness. For this purpose, fiber samples from 500 Indian cotton bales and 350 Greek cotton bales were collected and tested using the high volume instrument (HVI). The fiber properties dataset was then compiled and analyzed using python 3.7 to determine the correlations matrix. Results show that Indian cotton fiber have highest correlation between strength-mat = 0.84, followed by SFI-Unf =-0.83, and Neps-Unf = -0.72. Greek cotton fiber, in contrast, have highest correlation between SFI-Unf =-0.98, followed by SFI-Mat = 0.89, +b-Len = 0.84, and Str-Mat = 0.74. Overall, the Greek cotton fiber showed a higher correlational matrix than compared to that of Indian cotton fiber.

Keywords: cotton fiber, fiber properties correlation, Greek cotton, HVI, Indian cotton, spinning

Procedia PDF Downloads 148

Authors: Nazmul Shahadat, Anthony S. Maida

Abstract:

In recent years, hypercomplex inspired neural networks improved deep CNN architectures due to their ability to share weights across input channels and thus improve cohesiveness of representations within the layers. The work described herein studies the effect of replacing existing layers in an Axial Attention ResNet with their quaternion variants that use cross-channel weight sharing to assess the effect on image classification. We expect the quaternion enhancements to produce improved feature maps with more interlinked representations. We experiment with the stem of the network, the bottleneck layer, and the fully connected backend by replacing them with quaternion versions. These modifications lead to novel architectures which yield improved accuracy performance on the ImageNet300k classification dataset. Our baseline networks for comparison were the original real-valued ResNet, the original quaternion-valued ResNet, and the Axial Attention ResNet. Since improvement was observed regardless of which part of the network was modified, there is a promise that this technique may be generally useful in improving classification accuracy for a large class of networks.

Keywords: axial attention, representational networks, weight sharing, cross-channel correlations, quaternion-enhanced axial attention, deep networks

Procedia PDF Downloads 68

Authors: Yang Yang, Dan Liu

Abstract:

Anomaly detection models for network flow based on machine learning have poor detection performance under extremely unbalanced training data conditions and also have slow detection speed and large resource consumption when deploying on network edge devices. Embedding multi-teacher knowledge distillation (mKD) in anomaly detection can transfer knowledge from multiple teacher models to a single model. Inspired by this, we proposed a state-of-the-art model, mKDNAD, to improve detection performance. mKDNAD mine and integrate the knowledge of one-dimensional sequence and two-dimensional image implicit in network flow to improve the detection accuracy of small sample classes. The multi-teacher knowledge distillation method guides the train of the student model, thus speeding up the model's detection speed and reducing the number of model parameters. Experiments in the CICIDS2017 dataset verify the improvements of our method in the detection speed and the detection accuracy in dealing with the small sample classes.

Keywords: network flow anomaly detection (NAD), multi-teacher knowledge distillation, machine learning, deep learning

Procedia PDF Downloads 109