Search results for: deep neural models

Authors: Harish Kumar Sharma, Garima Sengar

Abstract:

Rice bran oil is considered comparatively nutritionally superior than different fats/oils. Therefore, model blends prepared from pure rice bran oil (RBO) and sunflower oil (SFO) were explored for changes in the different physicochemical parameters. Repeated deep fat frying process was carried out by using dried potato in order to study the thermal behaviour of pure rice bran oil, sunflower oil and their model blends. Pure rice bran oil and sunflower oil had shown good thermal stability during the repeated deep fat frying cycles. Although, the model blends constituting 60% RBO + 40% SFO showed better suitability during repeated deep fat frying than the remaining blended oils. The quantification of pure rice bran oil in the blended oils, physically refined rice bran oil (PRBO): SnF (sunflower oil) was carried by different methods. The study revealed that regression equations based on the oryzanol content, palmitic acid composition and iodine value can be used for the quantification. The rice bran oil can easily be quantified in the blended oils based on the oryzanol content by HPLC even at 1% level. The palmitic acid content in blended oils can also be used as an indicator to quantify rice bran oil at or above 20% level in blended oils whereas the method based on ultrasonic velocity, acoustic impedance and relative association showed initial promise in the quantification.

Keywords: rice bran oil, sunflower oil, frying, quantification

Procedia PDF Downloads 308

Authors: Szymon Rutkowski, Tomasz Korbak

Abstract:

We present a reinforcement-learning-based method of training neural machine translation models without parallel corpora. The standard encoder-decoder approach to machine translation suffers from two problems we aim to address. First, it needs parallel corpora, which are scarce, especially for low-resource languages. Second, it lacks psychological plausibility of learning procedure: learning a foreign language is about learning to communicate useful information, not merely learning to transduce from one language’s 'encoding' to another. We instead pose the problem of learning to translate as learning a policy in a communication game between two agents: the translator and the classifier. The classifier is trained beforehand on a natural language inference task (determining the entailment relation between a premise and a hypothesis) in the target language. The translator produces a sequence of actions that correspond to generating translations of both the hypothesis and premise, which are then passed to the classifier. The translator is rewarded for classifier’s performance on determining entailment between sentences translated by the translator to disciple’s native language. Translator’s performance thus reflects its ability to communicate useful information to the classifier. In effect, we train a machine translation model without the need for parallel corpora altogether. While similar reinforcement learning formulations for zero-shot translation were proposed before, there is a number of improvements we introduce. While prior research aimed at grounding the translation task in the physical world by evaluating agents on an image captioning task, we found that using a linguistic task is more sample-efficient. Natural language inference (also known as recognizing textual entailment) captures semantic properties of sentence pairs that are poorly correlated with semantic similarity, thus enforcing basic understanding of the role played by compositionality. It has been shown that models trained recognizing textual entailment produce high-quality general-purpose sentence embeddings transferrable to other tasks. We use stanford natural language inference (SNLI) dataset as well as its analogous datasets for French (XNLI) and Polish (CDSCorpus). Textual entailment corpora can be obtained relatively easily for any language, which makes our approach more extensible to low-resource languages than traditional approaches based on parallel corpora. We evaluated a number of reinforcement learning algorithms (including policy gradients and actor-critic) to solve the problem of translator’s policy optimization and found that our attempts yield some promising improvements over previous approaches to reinforcement-learning based zero-shot machine translation.

Keywords: agent-based language learning, low-resource translation, natural language inference, neural machine translation, reinforcement learning

Procedia PDF Downloads 128

Authors: Jui-Chen Huang, Shou-Hsiung Cheng

Abstract:

This study examined their willingness to use telecare for people who have had experience falling in the last three months in Taiwan. This study adopted convenience sampling and a structural questionnaire to collect data. It was based on the definition and the constructs related to the Health Belief Model (HBM). HBM is comprised of seven constructs: perceived benefits (PBs), perceived disease threat (PDT), perceived barriers of taking action (PBTA), external cues to action (ECUE), internal cues to action (ICUE), attitude toward using (ATT), and behavioral intention to use (BI). This study adopted Fuzzy Neural Network (FNN) to put forward an effective method. It shows the dependence of ATT on PB, PDT, PBTA, ECUE, and ICUE. The training and testing data RMSE (root mean square error) are 0.028 and 0.166 in the FNN, respectively. The training and testing data RMSE are 0.828 and 0.578 in the regression model, respectively. On the other hand, as to the dependence of ATT on BI, as presented in the FNN, the training and testing data RMSE are 0.050 and 0.109, respectively. The training and testing data RMSE are 0.529 and 0.571 in the regression model, respectively. The results show that the FNN method is better than the regression analysis. It is an effective and viable good way.

Keywords: fall, fuzzy neural network, health belief model, telecare, willingness

Procedia PDF Downloads 201

Authors: Jaiden Xuan Schraut, Leon Liu, Yiqiao Yin

Abstract:

Computer vision in medical diagnosis has achieved a high level of success in diagnosing diseases with high accuracy. However, conventional classifiers that produce an image to-label result provides insufficient information for medical professionals to judge and raise concerns over the trust and reliability of a model with results that cannot be explained. In order to gain local insight into cancerous regions, separate tasks such as imaging segmentation need to be implemented to aid the doctors in treating patients, which doubles the training time and costs which renders the diagnosis system inefficient and difficult to be accepted by the public. To tackle this issue and drive AI-first medical solutions further, this paper proposes a multi-output network that follows a U-Net architecture for image segmentation output and features an additional convolutional neural networks (CNN) module for auxiliary classification output. Class activation maps are a method of providing insight into a convolutional neural network’s feature maps that leads to its classification but in the case of lung diseases, the region of interest is enhanced by U-net-assisted Class Activation Map (CAM) visualization. Therefore, our proposed model combines image segmentation models and classifiers to crop out only the lung region of a chest X-ray’s class activation map to provide a visualization that improves the explainability and is able to generate classification results simultaneously which builds trust for AI-led diagnosis systems. The proposed U-Net model achieves 97.61% accuracy and a dice coefficient of 0.97 on testing data from the COVID-QU-Ex Dataset which includes both diseased and healthy lungs.

Keywords: multi-output network model, U-net, class activation map, image classification, medical imaging analysis

Procedia PDF Downloads 203

Authors: K. Padmavathi, K. Sri Ramakrishna

Abstract:

This study describes a procedure for the detection of Left and Right Bundle Branch Block (LBBB and RBBB) ECG patterns using spectral Coherence(SC) technique and LM Neural Network. The Coherence function finds common frequencies between two signals and evaluate the similarity of the two signals. The QT variations of Bundle Blocks are observed in lead V1 of ECG. Spectral Coherence technique uses Welch method for calculating PSD. For the detection of normal and Bundle block beats, SC output values are given as the input features for the LMNN classifier. Overall accuracy of LMNN classifier is 99.5 percent. The data was collected from MIT-BIH Arrhythmia database.

Keywords: bundle block, SC, LMNN classifier, welch method, PSD, MIT-BIH, arrhythmia database

Procedia PDF Downloads 281

Authors: Y. Zheng, M. Busi, A. F. Pedersen, M. A. Beltran, C. Gundlach

Abstract:

X-ray phase contrast imaging has shown to yield a better contrast compared to conventional attenuation X-ray imaging, especially for soft tissues in the medical imaging energy range. This can potentially lead to better diagnosis for patients. However, phase contrast imaging has mainly been performed using highly brilliant Synchrotron radiation, as it requires high coherence X-rays. Many research teams have demonstrated that it is also feasible using a laboratory source, bringing it one step closer to clinical use. Nevertheless, the requirement of fine gratings and high precision stepping motors when using a laboratory source prevents it from being widely used. Recently, a random phase object has been proposed as an analyzer. This method requires a much less robust experimental setup. However, previous studies were done using a particular X-ray source (liquid-metal jet micro-focus source) or high precision motors for stepping. We have been working on a much simpler setup with just small modification of a commercial bench-top micro-CT (computed tomography) scanner, by introducing a piece of sandpaper as the phase analyzer in front of the X-ray source. However, it needs a suitable algorithm for speckle tracking and 3D reconstructions. The precision and sensitivity of speckle tracking algorithm determine the resolution of the system, while the 3D reconstruction algorithm will affect the minimum number of projections required, thus limiting the temporal resolution. As phase contrast imaging methods usually require much longer exposure time than traditional absorption based X-ray imaging technologies, a dynamic phase contrast micro-CT with a high temporal resolution is particularly challenging. Different reconstruction methods, including neural network based techniques, will be evaluated in this project to increase the temporal resolution of the phase contrast micro-CT. A Monte Carlo ray tracing simulation (McXtrace) was used to generate a large dataset to train the neural network, in order to address the issue that neural networks require large amount of training data to get high-quality reconstructions.

Keywords: micro-ct, neural networks, reconstruction, speckle-based x-ray phase contrast

Procedia PDF Downloads 257

Authors: Minttu Laukkanen, Janne Huiskonen

Abstract:

This study investigates the sharing economy business models as examples of the sustainable business models. The aim is to contribute to the limited literature on sharing economy in connection with sustainable business models by explaining sharing economy business models value capturing. Specifically, this research answers the following question: How business model choices affect captured sustainable value? A multiple case study approach is applied in this study. Twenty different successful sharing economy business models focusing on consumer business and covering four main areas, accommodation, mobility, food, and consumer goods, are selected for analysis. The secondary data available on companies’ websites, previous research, reports, and other public documents are used. All twenty cases are analyzed through the sharing economy business model framework and sustainable value analysis framework using qualitative data analysis. This study represents general sharing economy business model value attributes and their specifications, i.e. sustainable value propositions for different stakeholders, and further explains the sustainability impacts of different sharing economy business models through captured and uncaptured value. In conclusion, this study represents how business model choices affect sustainable value capturing through eight business model attributes identified in this study. This paper contributes to the research on sustainable business models and sharing economy by examining how business model choices affect captured sustainable value. This study highlights the importance of careful business model and sustainability impacts analyses including the triple bottom line, multiple stakeholders and value captured and uncaptured perspectives as well as sustainability trade-offs. It is not self-evident that sharing economy business models advance sustainability, and business model choices does matter.

Keywords: sharing economy, sustainable business model innovation, sustainable value, value capturing

Procedia PDF Downloads 173

Authors: Manoj Reghu, Prabhu Rajagopal, C. V. Krishnamurthy, Krishnan Balasubramaniam

Abstract:

A thorough understanding of guided ultrasonic wave behavior in structures is essential for the application of existing Non Destructive Evaluation (NDE) technologies, as well as for the development of new methods. However, the analysis of guided wave phenomena is challenging because of their complex dispersive and multimodal nature. Although numerical solution procedures have proven to be very useful in this regard, the increasing complexity of features and defects to be considered, as well as the desire to improve the accuracy of inspection often imposes a large computational cost. Hybrid models that combine numerical solutions for wave scattering with faster alternative methods for wave propagation have long been considered as a solution to this problem. However usually such models require modification of the base code of the solution procedure. Here we aim to develop Generic Hybrid models that can be directly applied to any two different solution procedures. With this goal in mind, a Numerical Hybrid model and an Analytical-Numerical Hybrid model has been developed. The concept and implementation of these Hybrid models are discussed in this paper.

Keywords: guided ultrasonic waves, Finite Element Method (FEM), Hybrid model

Procedia PDF Downloads 465

Authors: Ibrahim Elnour Abdelrahman Eltayeb

Abstract:

Computational modelling is a method used to investigate joint forces during a movement. It can get high accuracy in the joint forces via subject-specific models. However, the construction of subject-specific models remains time-consuming and expensive. The purpose of this paper was to identify what alterations we can make to generic computational models to get a better estimation of the joint forces. It appraised the impact of these alterations on the accuracy of the estimated joint forces. It found different strategies of alterations: joint model, muscle model, and an optimisation problem. All these alterations affected joint contact force accuracy, so showing the potential for improving the model predictions without involving costly and time-consuming medical images.

Keywords: joint force, joint model, optimisation problem, validation

Procedia PDF Downloads 170

Authors: Xiaoqi Jia, Qing Zhu, Hao Zhang, Su Yang

Abstract:

A gender recognition system is able to tell the gender of the given person through a few of frontal facial images. An effective gender recognition approach enables to improve the performance of many other applications, including security monitoring, human-computer interaction, image or video retrieval and so on. In this paper, we present an effective method for gender classification task in frontal facial images based on deep belief networks (DBNs), which can pre-train model and improve accuracy a little bit. Our experiments have shown that the pre-training method with DBNs for gender classification task is feasible and achieves a little improvement of accuracy on FERET and CAS-PEAL-R1 facial datasets.

Keywords: gender recognition, beep belief net-works, semi-supervised learning, greedy-layer wise RBMs

Procedia PDF Downloads 453

Authors: H. Zormati, J. Chebil, J. Bel Hadj Tahar

Abstract:

Next generation wireless transmission technology (5G) is expected to support the development of channel models for higher frequency bands, so clarification of high frequency bands is the most important issue in radio propagation research for 5G, multiple urban microcellular measurements have been carried out at 60 GHz. In this paper, the collected data is uniformly analyzed with focus on the path loss (PL), the objective is to compare simulation results of some studied channel models with the purpose of testing the performance of each one.

Keywords: 5G, channel model, 60GHz channel, millimeter-wave, urban microcell

Procedia PDF Downloads 319

Authors: Zhiguo Yang, Xiangan Yue, Minglu Shao, Yue Yang, Rongjie Yan

Abstract:

It is difficult to realize deep profile control because of the small pore-throats and easy water channeling in low-permeability heterogeneous reservoir, and the traditional polymer microspheres have the contradiction between injection and plugging. In order to solve this contradiction, the controllable self-aggregating colloidal particles (CSA) containing amide groups on the surface of microspheres was prepared based on emulsion polymerization of styrene and acrylamide. The dispersed solution of CSA colloidal particles, whose particle size is much smaller than the diameter of pore-throats, was injected into the reservoir. When the microspheres migrated to the deep part of reservoir, , these CSA colloidal particles could automatically self-aggregate into large particle clusters under the action of the shielding agent and the control agent, so as to realize the plugging of the water channels. In this paper, the morphology, temperature resistance and self-aggregation properties of CSA microspheres were studied by transmission electron microscopy (TEM) and bottle test. The results showed that CSA microspheres exhibited heterogeneous core-shell structure, good dispersion, and outstanding thermal stability. The microspheres remain regular and uniform spheres at 100℃ after aging for 35 days. With the increase of the concentration of the cations, the self-aggregation time of CSA was gradually shortened, and the influence of bivalent cations was greater than that of monovalent cations. Core flooding experiments showed that CSA polymer microspheres have good injection properties, CSA particle clusters can effective plug the water channels and migrate to the deep part of the reservoir for profile control.

Keywords: heterogeneous reservoir, deep profile control, emulsion polymerization, colloidal particles, plugging characteristic

Procedia PDF Downloads 241

Authors: Taylan Kabbani, Ekrem Duman

Abstract:

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

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

Procedia PDF Downloads 178

Authors: Mojtaba Sadeghi, Farshid Baniassadi, Hamed Kashani

Abstract:

One of the challenging issues faced by seismic retrofitting consultants and employers is quick decision-making on the demolition or retrofitting of a structure at the current time or in the future. For this reason, the existing models proposed by researchers have only covered one of the aspects of cost, execution method, and structural vulnerability. Given the effect of each factor on the final decision, it is crucial to devise a new comprehensive model capable of simultaneously covering all the factors. This study attempted to provide an integrated framework that can be utilized to select the most appropriate earthquake risk mitigation solution for buildings. This framework can overcome the limitations of current models by taking into account several factors such as cost, execution method, risk-taking and structural failure. In the newly proposed model, the database and essential information about retrofitting projects are developed based on the historical data on a retrofit project. In the next phase, an analysis is conducted in order to assess the vulnerability of the building under study. Then, artificial neural networks technique is employed to calculate the cost of retrofitting. While calculating the current price of the structure, an economic analysis is conducted to compare demolition versus retrofitting costs. At the next stage, the optimal method is identified. Finally, the implementation of the framework was demonstrated by collecting data concerning 155 previous projects.

Keywords: decision making, demolition, construction management, seismic retrofit

Procedia PDF Downloads 238

Authors: Yun-An Chen, Hung-Jun Lin, Tai-Horng Young, Der-Zen Liu

Abstract:

Decellularized brain extracellular matrix had been shown that it has the ability to influence on cell proliferation, differentiation and associated cell phenotype. However, this scaffold is thought to have poor mechanical properties and rapid degradation, it is hard for cell recellularization. In this study, we used decellularized brain extracellular matrix combined with chitosan, which is naturally occurring polysaccharide and non-cytotoxic polymer, forming a 3-D scaffold for neural stem/precursor cells (NSPCs) regeneration. HE staining and DAPI fluorescence staining confirmed decellularized process could effectively vanish the cellular components from the brain. GAGs and collagen I, collagen IV were be showed a great preservation by Alcain staining and immunofluorescence staining respectively. Decellularized brain extracellular matrix was well mixed in chitosan to form a 3-D scaffold (DB-C scaffold). The pore size was approximately 50±10 μm examined by SEM images. Alamar blue results demonstrated NSPCs had great proliferation ability in DB-C scaffold. NSPCs that were cultured in this complex scaffold differentiated into neurons and astrocytes, as reveled by NSPCs expression of microtubule-associated protein 2 (MAP2) and glial fibrillary acidic protein (GFAP). In conclusion, DB-C scaffold may provide bioinformatics cues for NSPCs generation and aid for CNS injury functional recovery applications.

Keywords: brain, decellularization, chitosan, scaffold, neural stem/precursor cells

Procedia PDF Downloads 320

Authors: Mounir Zekkaoui, Abdelhadi Fennan

Abstract:

The software evolution control requires a deep understanding of the changes and their impact on different system heterogeneous artifacts. And an understanding of descriptive knowledge of the developed software artifacts is a prerequisite condition for the success of the evolutionary process. The implementation of an evolutionary process is to make changes more or less important to many heterogeneous software artifacts such as source code, analysis and design models, unit testing, XML deployment descriptors, user guides, and others. These changes can be a source of degradation in functional, qualitative or behavioral terms of modified software. Hence the need for a unified approach for extraction and representation of different heterogeneous artifacts in order to ensure a unified and detailed description of heterogeneous software artifacts, exploitable by several software tools and allowing to responsible for the evolution of carry out the reasoning change concerned.

Keywords: heterogeneous software artifacts, software evolution control, unified approach, meta model, software architecture

Procedia PDF Downloads 446

Authors: Thrivikraman Aswathi, S. Advaith

Abstract:

As there can be cases where the use of real data is somehow limited, such as when it is hard to get access to a large volume of real data, we need to go for synthetic data generation. This produces high-quality synthetic data while maintaining the statistical properties of a specific dataset. In the present work, a generative adversarial network (GAN) is trained to produce multivariate time series (MTS) data since the MTS is now being gathered more often in various real-world systems. Furthermore, the GAN-generated MTS data is fed into a transformer-based deep learning architecture that carries out the data categorization into predefined classes. Further, the model is evaluated across various distinct domains by generating corresponding MTS data.

Keywords: GAN, transformer, classification, multivariate time series

Procedia PDF Downloads 130

Authors: Thohidul Islam, Md. Hamidul Haque, Robin Kumar Biswas

Abstract:

Floods are one of the deadliest natural disasters which are very complex to model; however, machine learning is opening the door for more reliable and accurate flood prediction. In this research, a multilayer perceptron neural network (MLP) is developed to model the rainfall-water level relation, in a subtropical monsoon climatic region of the Bangladesh-India border. Our experiments show promising empirical results to forecast the water level for 1 day lead time. Our best performing MLP model achieves 98.7% coefficient of determination with lower model complexity which surpasses previously reported results on similar forecasting problems.

Keywords: flood forecasting, machine learning, multilayer perceptron network, regression

Procedia PDF Downloads 172

Authors: Thor Alexis Sazon, Alexander Guzman-Urbina, Yasuhiro Fukushima

Abstract:

This study outlines the method of how to develop a surrogate life cycle model based on fuzzy logic using three fuzzy inference methods: (1) the conventional Fuzzy Inference System (FIS), (2) the hybrid system of Data Analytics and Fuzzy Inference (DAFIS), which uses data clustering for defining the membership functions, and (3) the Adaptive-Neuro Fuzzy Inference System (ANFIS), a combination of fuzzy inference and artificial neural network. These methods were demonstrated with a case study where the Global Warming Potential (GWP) and the Levelized Cost of Energy (LCOE) of solar photovoltaic (PV) were estimated using Solar Irradiation, Module Efficiency, and Performance Ratio as inputs. The effects of using different fuzzy inference types, either Sugeno- or Mamdani-type, and of changing the number of input membership functions to the error between the calibration data and the model-generated outputs were also illustrated. The solution spaces of the three methods were consequently examined with a sensitivity analysis. ANFIS exhibited the lowest error while DAFIS gave slightly lower errors compared to FIS. Increasing the number of input membership functions helped with error reduction in some cases but, at times, resulted in the opposite. Sugeno-type models gave errors that are slightly lower than those of the Mamdani-type. While ANFIS is superior in terms of error minimization, it could generate solutions that are questionable, i.e. the negative GWP values of the Solar PV system when the inputs were all at the upper end of their range. This shows that the applicability of the ANFIS models highly depends on the range of cases at which it was calibrated. FIS and DAFIS generated more intuitive trends in the sensitivity runs. DAFIS demonstrated an optimal design point wherein increasing the input values does not improve the GWP and LCOE anymore. In the absence of data that could be used for calibration, conventional FIS presents a knowledge-based model that could be used for prediction. In the PV case study, conventional FIS generated errors that are just slightly higher than those of DAFIS. The inherent complexity of a Life Cycle study often hinders its widespread use in the industry and policy-making sectors. While the methodology does not guarantee a more accurate result compared to those generated by the Life Cycle Methodology, it does provide a relatively simpler way of generating knowledge- and data-based estimates that could be used during the initial design of a system.

Keywords: solar photovoltaic, fuzzy logic, inference system, artificial neural networks

Procedia PDF Downloads 164

Authors: Yooseo Pang, Jongwoong Choi, Yong Cho, Yongchae Jeong

Abstract:

Climate crisis, such as environmental problems related to energy supply, is getting emerged issues, so the use of renewable energy is essentially required to solve these problems, which are mainly managed by the Paris Agreement, the international treaty on climate change. The government of the Republic of Korea announced that the key long-term goal for a low-carbon strategy is “Carbon neutrality by 2050”. It is focused on the role of the internet data centers (IDC) in which large amounts of data, such as artificial intelligence (AI) and big data as an impact of the 4th industrial revolution, are managed. The demand for the cooling system market for IDC was about 9 billion US dollars in 2020, and 15.6% growth a year is expected in Korea. It is important to control the temperature in IDC with an efficient air conditioning system, so hydrothermal energy is one of the best options for saving energy in the cooling system. In order to save energy and optimize the operating conditions, it has been considered to apply ‘the dam deep water air conditioning system. Deep water at a specific level from the dam can supply constant water temperature year-round. It will be tested & analyzed the amount of energy saving with a pilot plant that has 100RT cooling capacity. Also, a target of this project is 1.2 PUE (Power Usage Effectiveness) which is the key parameter to check the efficiency of the cooling system.

Keywords: hydrothermal energy, HVAC, internet data center, free-cooling

Procedia PDF Downloads 81

Authors: Mai A. Ibrahim, Mohammed El-Beltagy, Motaz Khorshid

Abstract:

Emerging Markets return distributions have shown significance departure from normality were they are characterized by fatter tails relative to the normal distribution and exhibit levels of skewness and kurtosis that constitute a significant departure from normality. Therefore, the classical Markowitz Mean-Variance is not applicable for emerging markets since it assumes normally-distributed returns (with zero skewness and kurtosis) and a quadratic utility function. Moreover, the Markowitz mean-variance analysis can be used in cases of moderate non-normality and it still provides a good approximation of the expected utility, but it may be ineffective under large departure from normality. Higher moments models and median models have been suggested in the literature for asset allocation in this case. Higher moments models have been introduced to account for the insufficiency of the description of a portfolio by only its first two moments while the median model has been introduced as a robust statistic which is less affected by outliers than the mean. Tail risk measures such as Value-at Risk (VaR) and Conditional Value-at-Risk (CVaR) have been introduced instead of Variance to capture the effect of risk. In this research, higher moment models including the Mean-Variance-Skewness (MVS) and Mean-Variance-Skewness-Kurtosis (MVSK) are formulated as single-objective non-linear programming problems (NLP) and median models including the Median-Value at Risk (MedVaR) and Median-Mean Absolute Deviation (MedMAD) are formulated as a single-objective mixed-integer linear programming (MILP) problems. The higher moment models and median models are compared to some benchmark portfolios and tested on real financial data in the Egyptian main Index EGX30. The results show that all the median models outperform the higher moment models were they provide higher final wealth for the investor over the entire period of study. In addition, the results have confirmed the inapplicability of the classical Markowitz Mean-Variance to the Egyptian stock market as it resulted in very low realized profits.

Keywords: Egyptian stock exchange, emerging markets, higher moment models, median models, mixed-integer linear programming, non-linear programming

Procedia PDF Downloads 315

Authors: Celina Pestano-Gabino, Concepcion Gonzalez-Concepcion, M. Candelaria Gil-Fariña

Abstract:

Some estimation methods for VARMA models, and Multivariate Time Series Models in general, rely on the use of a Hankel matrix. It is known that if the data sample is populous enough and the dimension of the Hankel matrix is unnecessarily large, this may result in an unnecessary number of computations as well as in numerical problems. In this sense, the aim of this paper is two-fold. First, we provide some theoretical results for these matrices which translate into a lower dimension for the matrices normally used in the algorithms. This contribution thus serves to improve those methods from a numerical and, presumably, statistical point of view. Second, we have chosen an estimation algorithm to illustrate in practice our improvements. The results we obtained in a simulation of VARMA models show that an increase in the size of the Hankel matrix beyond the theoretical bound proposed as valid does not necessarily lead to improved practical results. Therefore, for future research, we propose conducting similar studies using any of the linear system estimation methods that depend on Hankel matrices.

Keywords: covariances Hankel matrices, Kronecker indices, system identification, VARMA models

Procedia PDF Downloads 243

Authors: Belayneh Matebie, Michael Melese

Abstract:

The study focuses on addressing the challenges in classifying and ensuring the quality of Eragrostis Teff, a small and round grain that is the smallest cereal grain. Employing a traditional classification method is challenging because of its small size and the similarity of its environmental characteristics. To overcome this, this study employs a machine learning approach to develop a source and quality classification system for Teff cereal. Data is collected from various production areas in the Amhara regions, considering two types of cereal (high and low quality) across eight classes. A total of 5,920 images are collected, with 740 images for each class. Image enhancement techniques, including scaling, data augmentation, histogram equalization, and noise removal, are applied to preprocess the data. Convolutional Neural Network (CNN) is then used to extract relevant features and reduce dimensionality. The dataset is split into 80% for training and 20% for testing. Different classifiers, including FVGG16, FINCV3, QSCTC, EMQSCTC, SVM, and RF, are employed for classification, achieving accuracy rates ranging from 86.91% to 97.72%. The ensemble of FVGG16, FINCV3, and QSCTC using the Max-Voting approach outperforms individual algorithms.

Keywords: Teff, ensemble learning, max-voting, CNN, SVM, RF

Procedia PDF Downloads 53

Authors: Reza Bazargan lari, Mohammad H. Fattahi

Abstract:

Natural resources management including water resources requires reliable estimations of time variant environmental parameters. Small improvements in the estimation of environmental parameters would result in grate effects on managing decisions. Noise reduction using wavelet techniques is an effective approach for pre-processing of practical data sets. Predictability enhancement of the river flow time series are assessed using fractal approaches before and after applying wavelet based pre-processing. Time series correlation and persistency, the minimum sufficient length for training the predicting model and the maximum valid length of predictions were also investigated through a fractal assessment.

Keywords: wavelet, de-noising, predictability, time series fractal analysis, valid length, ANN

Procedia PDF Downloads 368

Authors: Gaurav Shinde, Sai Charan Gongiguntla, Prajwal Shirur, Ahmed Hambaba

Abstract:

Health issues are significantly increasing, putting a substantial strain on healthcare services. This has accelerated the integration of machine learning in healthcare, particularly following the COVID-19 pandemic. The utilization of machine learning in healthcare has grown significantly. We introduce DeClEx, a pipeline that ensures that data mirrors real-world settings by incorporating Gaussian noise and blur and employing autoencoders to learn intermediate feature representations. Subsequently, our convolutional neural network, paired with spatial attention, provides comparable accuracy to state-of-the-art pre-trained models while achieving a threefold improvement in training speed. Furthermore, we provide interpretable results using explainable AI techniques. We integrate denoising and deblurring, classification, and explainability in a single pipeline called DeClEx.

Keywords: machine learning, healthcare, classification, explainability

Procedia PDF Downloads 56

Authors: Julia Ive, Lucia Specia

Abstract:

Understanding and detecting toxicity is an important problem to support safer human interactions online. Our work focuses on the important problem of contextual toxicity detection, where automated classifiers are tasked with determining whether a short textual segment (usually a sentence) is toxic within its conversational context. We use “toxicity” as an umbrella term to denote a number of variants commonly named in the literature, including hate, abuse, offence, among others. Detecting toxicity in context is a non-trivial problem and has been addressed by very few previous studies. These previous studies have analysed the influence of conversational context in human perception of toxicity in controlled experiments and concluded that humans rarely change their judgements in the presence of context. They have also evaluated contextual detection models based on state-of-the-art Deep Learning and Natural Language Processing (NLP) techniques. Counterintuitively, they reached the general conclusion that computational models tend to suffer performance degradation in the presence of context. We challenge these empirical observations by devising better contextual predictive models that also rely on NLP data augmentation techniques to create larger and better data. In our study, we start by further analysing the human perception of toxicity in conversational data (i.e., tweets), in the absence versus presence of context, in this case, previous tweets in the same conversational thread. We observed that the conclusions of previous work on human perception are mainly due to data issues: The contextual data available does not provide sufficient evidence that context is indeed important (even for humans). The data problem is common in current toxicity datasets: cases labelled as toxic are either obviously toxic (i.e., overt toxicity with swear, racist, etc. words), and thus context does is not needed for a decision, or are ambiguous, vague or unclear even in the presence of context; in addition, the data contains labeling inconsistencies. To address this problem, we propose to automatically generate contextual samples where toxicity is not obvious (i.e., covert cases) without context or where different contexts can lead to different toxicity judgements for the same tweet. We generate toxic and non-toxic utterances conditioned on the context or on target tweets using a range of techniques for controlled text generation(e.g., Generative Adversarial Networks and steering techniques). On the contextual detection models, we posit that their poor performance is due to limitations on both of the data they are trained on (same problems stated above) and the architectures they use, which are not able to leverage context in effective ways. To improve on that, we propose text classification architectures that take the hierarchy of conversational utterances into account. In experiments benchmarking ours against previous models on existing and automatically generated data, we show that both data and architectural choices are very important. Our model achieves substantial performance improvements as compared to the baselines that are non-contextual or contextual but agnostic of the conversation structure.

Keywords: contextual toxicity detection, data augmentation, hierarchical text classification models, natural language processing

Procedia PDF Downloads 170

Authors: Josua Sihotang

Abstract:

Jakarta Capital Special Region is the province that densely populated with rapidly growing infrastructure but less attention for the environmental condition. This makes some social problem happened like lack of clean water supply. Shallow groundwater and river water condition that has contaminated make the layer of deep water carrier (aquifer) should be done. This research aims to provide the people insight about deep groundwater potential and to determine the depth, location, and quality where the aquifer can be found in Jakarta’s area, particularly Kapuk-Cengkareng’s people. This research was conducted by geophysical method namely Well Logging Analysis. Well Logging is the geophysical method to know the subsurface lithology with the physical characteristic. The observation in this research area was conducted with several well devices that is Spontaneous Potential Log (SP Log), Resistivity Log, and Gamma Ray Log (GR Log). The first devices well is SP log which is work by comprising the electrical potential difference between the electrodes on the surface with the electrodes that is contained in the borehole and rock formations. The second is Resistivity Log, used to determine both the hydrocarbon and water zone based on their porosity and permeability properties. The last is GR Log, work by identifying radioactivity levels of rocks which is containing elements of thorium, uranium, or potassium. The observation result is curve-shaped which describes the type of lithological coating in subsurface. The result from the research can be interpreted that there are four of the deep groundwater layer zone with different quality. The good groundwater layer can be found in layers with good porosity and permeability. By analyzing the curves, it can be known that most of the layers which were found in this wellbore are clay stone with low resistivity and high gamma radiation. The resistivity value of the clay stone layers is about 2-4 ohm-meter with 65-80 Cps gamma radiation. There are several layers with high resistivity value and low gamma radiation (sand stone) that can be potential for being an aquifer. This is reinforced by the sand layer with a right-leaning SP log curve proving that this layer is permeable. These layers have 4-9 ohm-meter resistivity value with 40-65 Cps gamma radiation. These are mostly found as fresh water aquifer.

Keywords: aquifer, deep groundwater potential, well devices, well logging analysis

Procedia PDF Downloads 252

Authors: Qingqing Gai, Junxing Shen, Yu Luo

Abstract:

Due to the hardware architecture characteristics, some operators tend to acquire better performance if the input/output tensor dimensions are aligned to a certain minimum granularity, such as convolution and deconvolution commonly used in deep learning. Furthermore, if the requirements are not met, the general strategy is to pad with 0 to satisfy the requirements, potentially leading to the under-utilization of the hardware resources. Therefore, for the convolution and deconvolution whose input and output channels do not meet the minimum granularity alignment, we propose to transfer the W-dimensional data to the C-dimension for computation (W2C) to enable the C-dimension to meet the hardware requirements. This scheme also reduces the number of computations in the W-dimension. Although this scheme substantially increases computation, the operator’s speed can improve significantly. It achieves remarkable speedups on multiple hardware accelerators, including Nvidia Tensor cores, Qualcomm digital signal processors (DSPs), and Huawei neural processing units (NPUs). All you need to do is modify the network structure and rearrange the operator weights offline without retraining. At the same time, for some operators, such as the Reducemax, we observe that transferring the Cdimensional data to the W-dimension(C2W) and replacing the Reducemax with the Maxpool can accomplish acceleration under certain circumstances.

Keywords: convolution, deconvolution, W2C, C2W, alignment, hardware accelerator

Procedia PDF Downloads 104

Authors: Glemarie C. Hermosa, Sheng-Jie You, Chien Chih Hu

Abstract:

Efficient separation technologies are required for the removal of carbon dioxide from natural gas streams. Membrane-based natural gas separation has emerged as one of the fastest growing technologies, due to the compactness, higher energy efficiency and economic advantages which can be reaped. The removal of Carbon dioxide from gas streams using membrane technology will also give the advantage like environmental friendly process compared to the other technologies used in gas separation. In this study, Polyimide membranes, which are mostly used in the separation of gases, are blended with a new kind of solvent: Deep Eutectic Solvents or simply DES. The three types of DES are used are choline chloride based mixed with three different hydrogen bond donors: Lactic acid, N-methylurea and Urea. The blending of the DESs to Polyimide gave out high permeability performance. The Gas Separation performance for all the membranes involving CO2/CH4 showed low performance while for CO2/N2 surpassed the performance of some studies. Among the three types of DES used the solvent Choline Chloride/Lactic acid exhibited the highest performance for both Gas Separation applications. The values are 10.5 for CO2/CH4 selectivity and 60.5 for CO2/N2. The separation results for CO2/CH4 may be due to the viscosity of the DESs affecting the morphology of the fabricated membrane thus also impacts the performance. DES/blended Polyimide membranes fabricated are novel and have the potential of a low-cost and environmental friendly application for gas separation.

Keywords: deep eutectic solvents, gas separation, polyimide blends, polyimide membranes

Procedia PDF Downloads 310

Authors: I. Sagaama, A. Kechiche, W. Trojet, F. Kamoun

Abstract:

Replacing combustion engine vehicles by electric vehicles (EVs) is a major step in recent years due to their potential benefits. Battery autonomy and charging processes are still a big issue for that kind of vehicles. Therefore, reducing the energy consumption of electric vehicles becomes a necessity. Many researches target introducing recent information and communication technologies in EVs in order to propose reducing energy consumption services. Evaluation of realistic scenarios is a big challenge nowadays. In this paper, we will elaborate a state of the art of different proposed energy consumption models in the literature, then we will present a comparative study of these models, finally, we will extend previous works in order to propose an accurate and realistic energy model for calculating instantaneous power consumption of electric vehicles.

Keywords: electric vehicle, vehicular networks, energy models, traffic simulation

Procedia PDF Downloads 370