Search results for: reduced order models

Authors: Joseph M. Rich, Vinay A. Duddalwar, Assad A. Oberai

Abstract:

Prostate adenocarcinoma is the most common cancer in males, with osseous metastases as the commonest site of metastatic prostate carcinoma (mPC). Treatment monitoring is based on the evaluation and characterization of lesions on multiple imaging studies, including Computed Tomography (CT). Monitoring of the osseous disease burden, including follow-up of lesions and identification and characterization of new lesions, is a laborious task for radiologists. Deep learning algorithms are increasingly used to perform tasks such as identification and segmentation for osseous metastatic disease and provide accurate information regarding metastatic burden. Here, nnUNet was used to produce a model which can segment CT scan images of prostate adenocarcinoma vertebral bone metastatic lesions. nnUNet is an open-source Python package that adds optimizations to deep learning-based UNet architecture but has not been extensively combined with transfer learning techniques due to the absence of a readily available functionality of this method. The IRB-approved study data set includes imaging studies from patients with mPC who were enrolled in clinical trials at the University of Southern California (USC) Health Science Campus and Los Angeles County (LAC)/USC medical center. Manual segmentation of metastatic lesions was completed by an expert radiologist Dr. Vinay Duddalwar (20+ years in radiology and oncologic imaging), to serve as ground truths for the automated segmentation. Despite nnUNet’s success on some medical segmentation tasks, it only produced an average Dice Similarity Coefficient (DSC) of 0.31 on the USC dataset. DSC results fell in a bimodal distribution, with most scores falling either over 0.66 (reasonably accurate) or at 0 (no lesion detected). Applying more aggressive data augmentation techniques dropped the DSC to 0.15, and reducing the number of epochs reduced the DSC to below 0.1. Datasets have been identified for transfer learning, which involve balancing between size and similarity of the dataset. Identified datasets include the Pancreas data from the Medical Segmentation Decathlon, Pelvic Reference Data, and CT volumes with multiple organ segmentations (CT-ORG). Some of the challenges of producing an accurate model from the USC dataset include small dataset size (115 images), 2D data (as nnUNet generally performs better on 3D data), and the limited amount of public data capturing annotated CT images of bone lesions. Optimizations and improvements will be made by applying transfer learning and generative methods, including incorporating generative adversarial networks and diffusion models in order to augment the dataset. Performance with different libraries, including MONAI and custom architectures with Pytorch, will be compared. In the future, molecular correlations will be tracked with radiologic features for the purpose of multimodal composite biomarker identification. Once validated, these models will be incorporated into evaluation workflows to optimize radiologist evaluation. Our work demonstrates the challenges of applying automated image segmentation to small medical datasets and lays a foundation for techniques to improve performance. As machine learning models become increasingly incorporated into the workflow of radiologists, these findings will help improve the speed and accuracy of vertebral metastatic lesions detection.

Keywords: deep learning, image segmentation, medicine, nnUNet, prostate carcinoma, radiomics

Procedia PDF Downloads 90

Authors: Konstantinos Perifanos, Eirini Florou, Dionysis Goutsos

Abstract:

This paper presents and benchmarks a number of end-to-end Deep Learning based models for metaphor detection in Greek. We combine Convolutional Neural Networks and Recurrent Neural Networks with representation learning to bear on the metaphor detection problem for the Greek language. The models presented achieve exceptional accuracy scores, significantly improving the previous state-of-the-art results, which had already achieved accuracy 0.82. Furthermore, no special preprocessing, feature engineering or linguistic knowledge is used in this work. The methods presented achieve accuracy of 0.92 and F-score 0.92 with Convolutional Neural Networks (CNNs) and bidirectional Long Short Term Memory networks (LSTMs). Comparable results of 0.91 accuracy and 0.91 F-score are also achieved with bidirectional Gated Recurrent Units (GRUs) and Convolutional Recurrent Neural Nets (CRNNs). The models are trained and evaluated only on the basis of training tuples, the related sentences and their labels. The outcome is a state-of-the-art collection of metaphor detection models, trained on limited labelled resources, which can be extended to other languages and similar tasks.

Keywords: metaphor detection, deep learning, representation learning, embeddings

Procedia PDF Downloads 145

Authors: Ye Wang, Ickjai Lee

Abstract:

Due to the widespread of mobile sensing, there is a strong need to handle trails of moving objects, trajectories. This paper proposes three visual analytic approaches for higher order information of trajectory data sets based on the higher order Voronoi diagram data structure. Proposed approaches reveal geometrical information, topological, and directional information. Experimental results demonstrate the applicability and usefulness of proposed three approaches.

Keywords: visual analytics, higher order information, trajectory datasets, spatio-temporal data

Procedia PDF Downloads 400

Authors: Elimhan N. Mahmudov

Abstract:

The paper concerns the necessary and sufficient conditions of optimality for Cauchy problem of fourth order discrete (PD) and discrete-approximate (PDA) inclusions. The main problem is formulation of the fourth order adjoint discrete and discrete-approximate inclusions and transversality conditions, which are peculiar to problems including fourth order derivatives and approximate derivatives. Thus the necessary and sufficient conditions of optimality are obtained incorporating the Euler-Lagrange and Hamiltonian forms of inclusions. Derivation of optimality conditions are based on the apparatus of locally adjoint mapping (LAM). Moreover in the application of these results we consider the fourth order linear discrete and discrete-approximate inclusions.

Keywords: difference, optimization, fourth, approximation, transversality

Procedia PDF Downloads 367

Authors: Fatima Babar, Majid I. Khan, Malik Najmus Saqib, Muhammad Tahir

Abstract:

This article provides range based improvements over a well-known single-hop range free localization scheme, Approximate Point in Triangulation (APIT) by proposing an energy efficient Barycentric coordinate based Point-In-Triangulation (PIT) test along with PIT based trilateration. These improvements result in energy efficiency, reduced localization error and improved localization coverage compared to APIT and its variants. Moreover, we propose to embed Received signal strength indication (RSSI) based distance estimation in DV-Hop which is a multi-hop localization scheme. The proposed localization algorithm achieves energy efficiency and reduced localization error compared to DV-Hop and its available improvements. Furthermore, a hybrid multi-hop localization scheme is also proposed that utilize Barycentric coordinate based PIT test and both range based (Received signal strength indicator) and range free (hop count) techniques for distance estimation. Our experimental results provide evidence that proposed hybrid multi-hop localization scheme results in two to five times reduction in the localization error compare to DV-Hop and its variants, at reduced energy requirements.

Keywords: Localization, Trilateration, Triangulation, Wireless Sensor Networks

Procedia PDF Downloads 463

Authors: Lidija R. Jevrić, Sanja O. Podunavac-Kuzmanović, Strahinja Z. Kovačević

Abstract:

This study considers the selection of the most suitable in silico molecular descriptors that could be used for s-triazine pesticides characterization. Suitable descriptors among topological, geometrical and physicochemical are used for quantitative structure-retention relationships (QSRR) model establishment. Established models were obtained using linear regression (LR) and multiple linear regression (MLR) analysis. In this paper, MLR models were established avoiding multicollinearity among the selected molecular descriptors. Statistical quality of established models was evaluated by standard and cross-validation statistical parameters. For detection of similarity or dissimilarity among investigated s-triazine pesticides and their classification, principal component analysis (PCA) and hierarchical cluster analysis (HCA) were used and gave similar grouping. This study is financially supported by COST action TD1305.

Keywords: chemometrics, classification analysis, molecular descriptors, pesticides, regression analysis

Procedia PDF Downloads 386

Authors: Ravi Nandu, Anmol Taploo

Abstract:

As per current scenario, commercial aircrafts have been very well functioning with higher efficiency, but there is something that affects it. Every aircraft runs with the combustion produced by mixture of fuel and air. For example: A flight to travel from Mumbai to Kolkata it takes 2h: 30 min and from Kolkata to Mumbai it takes 2h: 45 min. It happens due to head and tail wind. Due to head wind air craft travels faster than its usual velocity and it takes 2h: 30 min to reach to Kolkata, while it takes 2h;45min vis versa. This lag in time is caused due to head wind that increases the drag and reduces the relative velocity of the plane. So in order to reduce this wastage of fuel there is an optimal flight altitude at which the head and tail wind action is reduced compared to the present scenario.

Keywords: drag, head wind, tail wind, aircraft

Procedia PDF Downloads 463

Authors: Selvakumar Ulaganathan, Ivo Couckuyt, Francesco Ferranti, Tom Dhaene, Eric Laermans

Abstract:

Variable-fidelity surrogate modelling offers an efficient way to approximate function data available in multiple degrees of accuracy each with varying computational cost. In this paper, a Kriging-based variable-fidelity surrogate modelling approach is introduced to approximate such deterministic data. Initially, individual Kriging surrogate models, which are enhanced with gradient data of different degrees of accuracy, are constructed. Then these Gradient enhanced Kriging surrogate models are strategically coupled using a recursive CoKriging formulation to provide an accurate surrogate model for the highest fidelity data. While, intuitively, gradient data is useful to enhance the accuracy of surrogate models, the primary motivation behind this work is to investigate if it is also worthwhile incorporating gradient data of varying degrees of accuracy.

Keywords: Kriging, CoKriging, Surrogate modelling, Variable- fidelity modelling, Gradients

Procedia PDF Downloads 549

Authors: Donglan Zha, Jiansong Si

Abstract:

Because of its flexibility, CES attracts much interest in economic growth and programming models, and the macroeconomics or micro-macro models. This paper focuses on the development, estimating methods of CES production function considering energy as an input. We leave for future research work of relaxing the assumption of constant returns to scale, the introduction of potential input factors, and the generalization method of the optimal nested form of multi-factor production functions.

Keywords: bias of technical change, CES production function, elasticity of substitution, energy input

Procedia PDF Downloads 276

Authors: Puttharapong Sakulwaropas, Uraiwan Jaroengeratikun

Abstract:

Casualty insurance business, the optimal premium pricing and adequate cost for an insurance company are important in risk management. Normally, the insurance pure premium can be determined by multiplying the claim frequency with the claim cost. The aim of this research was to study in the application of generalized linear models to select the risk factor for model of claim frequency and claim cost for estimating a pure premium. In this study, the data set was the claim of comprehensive motor insurance, which was provided by one of the insurance company in Thailand. The results of this study found that the risk factors significantly related to pure premium at the 0.05 level consisted of no claim bonus (NCB) and used of the car (Car code).

Keywords: generalized linear models, risk factor, pure premium, regression model

Procedia PDF Downloads 460

Authors: N. Nandan A. M. Nagaraj, L. Sanjeev Kumar

Abstract:

In the rapidly growing population, the requirement of electrical power is increasing day by day. In order to meet the needs, we need to generate the power using alternate method. In this paper, a presentable approach is developed by analysis and can be implemented by utilizing heat energy, which is generated in numerous ways in some of the rural areas in India. The thermoelectric generator unit will be developed by combing with control circuits and converts, which is used to light the LED lamps. The temperature difference which is available in the kitchens, especially the exhaust pipes/chimneys of wooden fire stoves, where more heat is dissipated into the atmosphere, can be utilized for electrical power generation. Hence, the temperature rise of surroundings atmosphere can be reduced.

Keywords: thermo electric generator, LED, converts, temperature

Procedia PDF Downloads 140

Authors: Edlira Kalemi, Sule Yildirim-Yayilgan

Abstract:

Online Social Networks (OSNs) are nowadays being used widely and intensively for crime investigation and prevention activities. As they provide a lot of information they are used by the law enforcement and intelligence. An extensive review on existing solutions and models for collecting intelligence from this source of information and making use of it for solving crimes has been presented in this article. The main focus is on smart solutions and models where ontologies have been used as the main approach for representing criminal domain knowledge. A framework for a prototype ontology named SC-Ont will be described. This defines terms of the criminal domain ontology and the relations between them. The terms and the relations are extracted during both this review and the discussions carried out with domain experts. The development of SC-Ont is still ongoing work, where in this paper, we report mainly on the motivation for using smart ontology models and the possible benefits of using them for solving crimes.

Keywords: criminal digital evidence, social media, ontologies, reasoning

Procedia PDF Downloads 383

Authors: Asma Rebaya, Kaouther Gasmi, Imen Amari, Salem Hasnaoui

Abstract:

Over these last years, applications such as telecommunications, signal processing, digital communication with advanced features (Multi-antenna, equalization..) witness a rapid evaluation accompanied with an increase of user exigencies in terms of latency, the power of computation… To satisfy these requirements, the use of hardware/software systems is a common solution; where hardware is composed of multi-cores and software is represented by models of computation, synchronous data flow (SDF) graph for instance. Otherwise, the most of the embedded system designers utilize Simulink for modeling. The issue is how to simplify the c code generation, for a multi-cores platform, of an application modeled by Simulink. To overcome this problem, we propose a workflow allowing an automatic transformation from the Simulink model to the SDF graph and providing an efficient schedule permitting to optimize the number of cores and to minimize latency. This workflow goes from a Simulink application and a hardware architecture described by IP.XACT language. Based on the synchronous and hierarchical behavior of both models, the Simulink block diagram is automatically transformed into an SDF graph. Once this process is successfully achieved, the scheduler calculates the optimal cores’ number needful by minimizing the maximum density of the whole application. Then, a core is chosen to execute a specific graph task in a specific order and, subsequently, a compatible C code is generated. In order to perform this proposal, we extend Preesm, a rapid prototyping tool, to take the Simulink model as entry input and to support the optimal schedule. Afterward, we compared our results to this tool results, using a simple illustrative application. The comparison shows that our results strictly dominate the Preesm results in terms of number of cores and latency. In fact, if Preesm needs m processors and latency L, our workflow need processors and latency L'< L.

Keywords: hardware/software system, latency, modeling, multi-cores platform, scheduler, SDF graph, Simulink model, workflow

Procedia PDF Downloads 260

Authors: Hassan Jebreen

Abstract:

These models of a conservative pollutant in groundwater do not include representation of processes in soils and in the unsaturated zone, or biogeochemical processes in groundwater, These demonstration models can be used as the basis for more detailed simulations of the impacts of pollution sources at a local scale, but such studies should address processes related to specific pollutant species, and should consider local hydrogeology in more detail, particularly in relation to possible impacts on shallow systems which are likely to respond more quickly to changes in pollutant inputs. The results have demonstrated the interaction between groundwater flow fields and pollution sources in abstraction areas, and help to emphasise that wadi development is one of the key elements of water resources planning. The quality of groundwater in the Hebron area indicates a gradual increase in chloride and nitrate with time. Since the aquifers in Hebron districts are highly vulnerable due to their karstic nature, continued disposal of untreated domestic and industrial wastewater into the wadi will lead to unacceptably poor water quality in drinking water, which may ultimately require expensive treatment if significant health problems are to be avoided. Improvements are required in wastewater treatment at the municipal and domestic levels, the latter requiring increased public awareness of the issues, as well as improved understanding of the hydrogeological behaviour of the aquifers.

Keywords: groundwater, models, pollutants, wadis, hebron

Procedia PDF Downloads 429

Authors: Said Benkaciali, Mourad Haddadi, Abdallah Khellaf, Kacem Gairaa, Mawloud Guermoui

Abstract:

In this study, many experiments were carried out to assess the influence of the input parameters on the performance of multilayer perceptron which is one the configuration of the artificial neural networks. To estimate the daily global solar radiation on the horizontal surface, we have developed some models by using seven combinations of twelve meteorological and geographical input parameters collected from a radiometric station installed at Ghardaïa city (southern of Algeria). For selecting of best combination which provides a good accuracy, six statistical formulas (or statistical indicators) have been evaluated, such as the root mean square errors, mean absolute errors, correlation coefficient, and determination coefficient. We noted that multilayer perceptron techniques have the best performance, except when the sunshine duration parameter is not included in the input variables. The maximum of determination coefficient and correlation coefficient are equal to 98.20 and 99.11%. On the other hand, some empirical models were developed to compare their performances with those of multilayer perceptron neural networks. Results obtained show that the neural networks techniques give the best performance compared to the empirical models.

Keywords: empirical models, multilayer perceptron neural network, solar radiation, statistical formulas

Procedia PDF Downloads 339

Authors: Te-Jen Chang, I-Hui Pan, Ping-Sheng Huang, Shan-Jen Cheng

Abstract:

In this paper, a fast multiplication computing method utilizing the complement representation method and canonical recoding technique is proposed. By performing complements and canonical recoding technique, the number of partial products can be reduced. Based on these techniques, we propose an algorithm that provides an efficient multiplication method. On average, our proposed algorithm is to reduce the number of k-bit additions from (0.25k+logk/k+2.5) to (k/6 +logk/k+2.5), where k is the bit-length of the multiplicand A and multiplier B. We can therefore efficiently speed up the overall performance of the multiplication. Moreover, if we use the new proposes to compute common-multiplicand multiplication, the computational complexity can be reduced from (0.5 k+2 logk/k+5) to (k/3+2 logk/k+5) k-bit additions.

Keywords: algorithm design, complexity analysis, canonical recoding, public key cryptography, common-multiplicand multiplication

Procedia PDF Downloads 425

Authors: F. Amireche, I. Mokbel, J. Jose, B. F. Belaribi

Abstract:

The experimental vapour-liquid equilibria (VLE) at isothermal conditions and excess molar Gibbs energies GE are carried out for the three binary mixtures: 1, 2- dichloroethane + ethylene glycol dimethyl ether, + diethylene glycol dimethyl ether or + diethylene glycol diethyl ether, at ten temperatures ranging from 273 to 353.15 K. A good static device was employed for these measurements. The VLE data were reduced using the Redlich-Kister equation by taking into consideration the vapour pressure non-ideality in terms of the second molar virial coefficient. The experimental data were compared to the results predicted with the DISQUAC and Dortmund UNIFAC group contribution models for the total pressures P, the excess molar Gibbs energies GE and the excess molar enthalpies HE.

Keywords: Disquac model, Dortmund UNIFAC model, 1, 2- dichloroethane, excess molar Gibbs energies GE, polyethers, VLE

Procedia PDF Downloads 264

Authors: Aurelija Burinskiene

Abstract:

This paper presents the study for an e-commerce warehouse. The study is aiming to improve order fulfillment activity by identifying the strategy presenting the best performance. A simulation model was proposed to reach the target of this research. This model enables various scenario tests in an e-commerce warehouse, allowing them to find out for the best order fulfillment strategy. By using simulation, model authors investigated customers’ orders representing on-line purchases for one month. Experiments were designed to evaluate various order picking methods applicable to the fulfillment of customers’ orders. The research uses cost components analysis and helps to identify the best possible order picking method improving the overall performance of e-commerce warehouse and fulfillment service to the customers. The results presented show that the application of order batching strategy is the most applicable because it brings distance savings of around 6.7 percentage. This result could be improved by taking an assortment clustering action until 8.34 percentage. So, the recommendations were given to apply the method for future e-commerce warehouse operations.

Keywords: e-commerce, order, fulfilment, strategy, simulation

Procedia PDF Downloads 147

Authors: Leonard Maaya, Michel Meulders, Martina Vandebroek

Abstract:

Discrete Choice Experiments (DCE) are used to investigate how product attributes affect decision-makers’ choices. In DCEs, choice situations consisting of several alternatives are presented from which choice-makers select the preferred alternative. Standard multinomial logit models based on random utility theory can be used to estimate the utilities for the attributes. The overarching principle in these models is that respondents understand and use all the attributes when making choices. However, studies suggest that respondents sometimes ignore some attributes (commonly referred to as Attribute Non-Attendance/ANA). The choice modeling literature presents ANA as a static process, i.e., respondents’ ANA behavior does not change throughout the experiment. However, respondents may ignore attributes due to changing factors like availability of information on attributes, learning/fatigue in experiments, etc. We develop a dynamic mixture latent Markov model to model changes in ANA when information on attributes is provided. The model is illustrated on e-consumers’ webshop choices. The results indicate that the dynamic ANA model describes the behavioral changes better than modeling the impact of information using changes in parameters. Further, we find that providing information on attributes leads to an increase in the attendance probabilities for the investigated attributes.

Keywords: choice models, discrete choice experiments, dynamic models, e-commerce, statistical modeling

Procedia PDF Downloads 133

Authors: N. Rajeswara Rao, T. Venkatappa Rao, K. Sowri Babu, N. Srinivas Rao, P. S. V. Shanmukhi

Abstract:

Carboxymethyl Starch (CMS) is a biopolymer derived from starch by the substitution method. CMS is proclaimed to have improved physicochemical properties than native starch. The present work deals with the effect of gamma radiation on the physicochemical properties of CMS. The samples were exposed to gamma irradiation of doses 30, 60 and 90 kGy. The resultant properties were studied with electron spin resonance (ESR), fourier transform infrared spectrometer (FTIR), differential scanning calorimeter (DSC), X-ray diffractometer (XRD) and scanning electron microscopy. Irradiation of CMS by gamma rays initiates cleavage of glucosidic bonds producing different types of radicals. Some of these radicals convert to peroxy radicals by abstracting oxygen. The ESR spectrum of CMS is anisotropic and is thought to be due to the superposition of various component spectra. In order to analyze the ESR spectrum, computer simulations were also employed. ESR spectra are also recorded under different conditions like post-irradiation times, variable temperatures and saturation behavior in order to evaluate the stability of free radicals produced on irradiation. Thermal studies from DSC depict that for CMS the gelatization process was absconded at higher doses. Relative crystallinity was reduced significantly after irradiation from XRD Studies. FTIR studies also confirm the same aspect. From ESR studies, it was concluded that irradiated CMS could be a potential reference material in ESR dosimetry.

Keywords: gamma rays, free radicals, ESR simulations, gelatization

Procedia PDF Downloads 99

Authors: M. A. Khanday, Aasma Rafiq, Khalid Nazir

Abstract:

This paper is an attempt to establish the mathematical models to understand the distribution of drug administration in the human body through oral and intravenous routes. Three models were formulated based on diffusion process using Fick’s principle and the law of mass action. The rate constants governing the law of mass action were used on the basis of the drug efficacy at different interfaces. The Laplace transform and eigenvalue methods were used to obtain the solution of the ordinary differential equations concerning the rate of change of concentration in different compartments viz. blood and tissue medium. The drug concentration in the different compartments has been computed using numerical parameters. The results illustrate the variation of drug concentration with respect to time using MATLAB software. It has been observed from the results that the drug concentration decreases in the first compartment and gradually increases in other subsequent compartments.

Keywords: Laplace transform, diffusion, eigenvalue method, mathematical model

Procedia PDF Downloads 328

Authors: Md Murshed Bhuyan, Hirotaka Okabe, Yoshiki Hidaka, Kazuhiro Hara

Abstract:

Pectin-Acrylamide- (Vinyl Phosphonic Acid) Hydrogels were prepared from their blend by using gamma radiation of various doses. It was found that the gel fraction of hydrogel increases with increasing the radiation dose reaches a maximum and then started decreasing with increasing the dose. The optimum radiation dose and the composition of raw materials were determined on the basis of equilibrium swelling which resulted in 20 kGy absorbed dose and 1:2:4 (Pectin:AAm:VPA) composition. Differential scanning calorimetry reveals the gel strength for using them as the adsorbent. The FTIR-spectrum confirmed the grafting/ crosslinking of the monomer on the backbone of pectin chain. The hydrogels were applied in adsorption of Al, Ga, and In from multielement solution where the adsorption capacity order for those three elements was found as – In>Ga>Al. SEM images of hydrogels and metal adsorbed hydrogels indicate the gel network and adherence of the metal ions in the interpenetrating network of the hydrogel which were supported by EDS spectra. The adsorption isotherm models were studied and found that the Langmuir adsorption isotherm model was well fitted with the data. Adsorption data were also fitted to different adsorption kinetic and diffusion models. Desorption of metal adsorbed hydrogels was performed in 5% nitric acid where desorption efficiency was found around 90%.

Keywords: hydrogel, gamma radiation, vinyl phosphonic acid, metal adsorption

Procedia PDF Downloads 149

Authors: Yasamin Rahimi, Ali Kamandi, Abbas Hoseini, Hesam Haddad

Abstract:

Online/offline chat is a convenient approach in the electronic markets of second-hand products in which potential customers would like to have more information about the products to fill the information gap between buyers and sellers. Online peer in peer market is trying to create artificial intelligence-based systems that help customers ask more informative questions in an easier way. In this article, we introduce a method for the question/answer system that we have developed for the top-ranked electronic market in Iran called Divar. When it comes to secondhand products, incomplete product information in a purchase will result in loss to the buyer. One way to balance buyer and seller information of a product is to help the buyer ask more informative questions when purchasing. Also, the short time to start and achieve the desired result of the conversation was one of our main goals, which was achieved according to A/B tests results. In this paper, we propose and evaluate a method for suggesting questions and answers in the messaging platform of the e-commerce website Divar. Creating such systems is to help users gather knowledge about the product easier and faster, All from the Divar database. We collected a dataset of around 2 million messages in Persian colloquial language, and for each category of product, we gathered 500K messages, of which only 2K were Tagged, and semi-supervised methods were used. In order to publish the proposed model to production, it is required to be fast enough to process 10 million messages daily on CPU processors. In order to reach that speed, in many subtasks, faster and simplistic models are preferred over deep neural models. The proposed method, which requires only a small amount of labeled data, is currently used in Divar production on CPU processors, and 15% of buyers and seller’s messages in conversations is directly chosen from our model output, and more than 27% of buyers have used this model suggestions in at least one daily conversation.

Keywords: smart reply, spell checker, information retrieval, intent detection, question answering

Procedia PDF Downloads 180

Authors: Mario Isaza-Ruget, Claudia C. Colmenares-Mejia, Nancy Yomayusa, Camilo A. González, Andres Cely, Jossie Murcia

Abstract:

Quantification of risks associated with complications development from chronic kidney disease (CKD) through accurate survival models can help with patient management. A retrospective cohort that included patients diagnosed with CKD from a primary care program and followed up between 2013 and 2018 was carried out. Time-dependent and static covariates associated with demographic, clinical, and laboratory factors were included. Deep Learning (DL) survival analyzes were developed for three CKD outcomes: CKD stage progression, >25% decrease in Estimated Glomerular Filtration Rate (eGFR), and Renal Replacement Therapy (RRT). Models were evaluated and compared with Random Survival Forest (RSF) based on concordance index (C-index) metric. 2.143 patients were included. Two models were developed for each outcome, Deep Neural Network (DNN) model reported C-index=0.9867 for CKD stage progression; C-index=0.9905 for reduction in eGFR; C-index=0.9867 for RRT. Regarding the RSF model, C-index=0.6650 was reached for CKD stage progression; decreased eGFR C-index=0.6759; RRT C-index=0.8926. DNN models applied in survival analysis context with considerations of longitudinal covariates at the start of follow-up can predict renal stage progression, a significant decrease in eGFR and RRT. The success of these survival models lies in the appropriate definition of survival times and the analysis of covariates, especially those that vary over time.

Keywords: artificial intelligence, chronic kidney disease, deep neural networks, survival analysis

Procedia PDF Downloads 129

Authors: Arash Parehkar, S. Jamshid Mousavi, Shoubo Bayazidi, Vahid Karami, Laleh Shahidi, Arash Azaranfar, Ali Moridi, M. Shabakhti, Tayebeh Ariyan, Mitra Tofigh, Kaveh Masoumi, Alireza Motahari

Abstract:

Water resource systems modelling have constantly been a challenge through history for human being. As the innovative methodological development is evolving alongside computer sciences on one hand, researches are likely to confront more complex and larger water resources systems due to new challenges regarding increased water demands, climate change and human interventions, socio-economic concerns, and environment protection and sustainability. In this research, an automatic calibration scheme has been applied on the Gilan’s large-scale water resource model using mathematical programming. The water resource model’s calibration is developed in order to attune unknown water return flows from demand sites in the complex Sefidroud irrigation network and other related areas. The calibration procedure is validated by comparing several gauged river outflows from the system in the past with model results. The calibration results are pleasantly reasonable presenting a rational insight of the system. Subsequently, the unknown optimized parameters were used in a basin-scale linear optimization model with the ability to evaluate the system’s performance against a reduced inflow scenario in future. Results showed an acceptable match between predicted and observed outflows from the system at selected hydrometric stations. Moreover, an efficient operating policy was determined for Sefidroud dam leading to a minimum water shortage in the reduced inflow scenario.

Keywords: auto-calibration, Gilan, large-scale water resources, simulation

Procedia PDF Downloads 329

Authors: Heerak Banerjee, Sourov Roy

Abstract:

Supersymmetry, one of the most celebrated fields of study for explaining experimental observations where the standard model (SM) falls short, is reeling from the lack of experimental vindication. At the same time, the idea of additional gauge symmetry, in particular, the gauged Lμ-Lτ symmetric models have also generated significant interest. They have been extensively proposed in order to explain the tantalizing discrepancy in the predicted and measured value of the muon anomalous magnetic moment alongside several other issues plaguing the SM. While very little parameter space within these models remain unconstrained, this work finds that the γ + Missing Energy (ME) signal at the Belle-II detector will be a smoking gun for supersymmetry (SUSY) in the presence of a gauged U(1)Lμ-Lτ symmetry. A remarkable consequence of breaking the enhanced symmetry appearing in the limit of degenerate (s)leptons is the nondecoupling of the radiative contribution of heavy charged sleptons to the γ-Z΄ kinetic mixing. The signal process, e⁺e⁻ →γZ΄→γ+ME, is an outcome of this ubiquitous feature. Taking the severe constraints on gauged Lμ-Lτ models by several low energy observables into account, it is shown that any significant excess in all but the highest photon energy bin would be an undeniable signature of such heavy scalar fields in SUSY coupling to the additional gauge boson Z΄. The number of signal events depends crucially on the logarithm of the ratio of stau to smuon mass in the presence of SUSY. In addition, the number is also inversely proportional to the e⁺e⁻ collision energy, making a low-energy, high-luminosity collider like Belle-II an ideal testing ground for this channel. This process can probe large swathes of the hitherto free slepton mass ratio vs. additional gauge coupling (gₓ) parameter space. More importantly, it can explore the narrow slice of Z΄ mass (MZ΄) vs. gₓ parameter space still allowed in gauged U(1)Lμ-Lτ models for superheavy sparticles. The spectacular finding that the signal significance is independent of individual slepton masses is an exciting prospect indeed. Further, the prospect that signatures of even superheavy SUSY particles that may have escaped detection at the LHC may show up at the Belle-II detector is an invigorating revelation.

Keywords: additional gauge symmetry, electron-positron collider, kinetic mixing, nondecoupling radiative effect, supersymmetry

Procedia PDF Downloads 124

Authors: Shan-Hong Ying, Ying-Fang Wang

Abstract:

A surrogate approach was deployed for assessing exposures of multiple chemicals at the selected working area of coating processes and applied to assess the exposure concentration of similar exposed groups using the same chemicals but different formula ratios. For the selected area, 6 to 12 portable photoionization detector (PID) were placed uniformly in its workplace to measure its total VOCs concentrations (CT-VOCs) for 6 randomly selected workshifts. Simultaneously, one sampling strain was placed beside one of these portable PIDs, and the collected air sample was analyzed for individual concentration (CVOCi) of 5 VOCs (xylene, butanone, toluene, butyl acetate, and dimethylformamide). Predictive models were established by relating the CT-VOCs to CVOCi of each individual compound via simple regression analysis. The established predictive models were employed to predict each CVOCi based on the measured CT-VOC for each the similar working area using the same portable PID. Results show that predictive models obtained from simple linear regression analyses were found with an R2 = 0.83~0.99 indicating that CT-VOCs were adequate for predicting CVOCi. In order to verify the validity of the exposure prediction model, the sampling analysis of the above chemical substances was further carried out and the correlation between the measured value (Cm) and the predicted value (Cp) was analyzed. It was found that there is a good correction between the predicted value and measured value of each measured chemical substance (R2=0.83~0.98). Therefore, the surrogate approach could be assessed the exposure concentration of similar exposed groups using the same chemicals but different formula ratios. However, it is recommended to establish the prediction model between the chemical substances belonging to each coater and the direct-reading PID, which is more representative of reality exposure situation and more accurately to estimate the long-term exposure concentration of operators.

Keywords: exposure assessment, exposure prediction model, surrogate approach, TVOC

Procedia PDF Downloads 142

Authors: Ka C. Lam, Oluwafunmibi S. Idowu

Abstract:

Uncertainty in cost is a major factor affecting performance of construction projects. To our knowledge, several conceptual cost models have been developed with varying degrees of accuracy. Incorporating conceptual quantities into conceptual cost models could improve the accuracy of early predesign cost estimates. Hence, the development of quantity models for estimating conceptual quantities of framed reinforced concrete structures using supervised machine learning is the aim of the current research. Using measured quantities of structural elements and design variables such as live loads and soil bearing pressures, response and predictor variables were defined and used for constructing conceptual quantities models. Twenty-four models were developed for comparison using a combination of non-parametric support vector regression, linear regression, and bootstrap resampling techniques. R programming language was used for data analysis and model implementation. Gross soil bearing pressure and gross floor loading were discovered to have a major influence on the quantities of concrete and reinforcement used for foundations. Building footprint and gross floor loading had a similar influence on beams and slabs. Future research could explore the modelling of other conceptual quantities for walls, finishes, and services using machine learning techniques. Estimation of conceptual quantities would assist construction planners in early resource planning and enable detailed performance evaluation of early cost predictions.

Keywords: bootstrapping, conceptual quantities, modelling, reinforced concrete, support vector regression

Procedia PDF Downloads 203

Authors: Flackson Tshuma, James Bennett, Pieter Andreas Swanepoel, Johan Labuschagne, Stephan van der Westhuizen, Francis Rayns

Abstract:

Amongst other things, tillage and synthetic agrochemicals can be effective methods of seedbed preparation and pest control. Nonetheless, frequent and intensive tillage and excessive application of synthetic agrochemicals, such as herbicides and insecticides, can reduce soil microbial enzyme activity. A decline in soil microbial enzyme activity can negatively affect nutrient cycling and crop productivity. In this study, the effects of four tillage treatments; continuous mouldboard plough; shallow tine-tillage to a depth of about 75 mm; no-tillage; and tillage rotation (involving shallow tine-tillage once every four years in rotation with three years of no-tillage), and two rates of synthetic agrochemicals (standard: with regular application of synthetic agrochemicals; and reduced: fewer synthetic agrochemicals in combination with bio-chemicals/ or bio-stimulants) on soil microbial enzyme activity were investigated between 2018 and 2020 in a typical Mediterranean climate zone in South Africa. Four different bio-stimulants applied contained: Trichoderma asperellum, fulvic acid, silicic acid, and Nereocystis luetkeana extracts, respectively. The study was laid out as a complete randomised block design with four replicated blocks. Each block had 14 plots, and each plot measured 50 m x 6 m. The study aimed to assess the combined impact of tillage practices and reduced rates of synthetic agrochemical application on soil microbial enzyme activity in a dryland cropping system. It was hypothesised that the application of bio-stimulants in combination with minimum soil disturbance will lead to a greater increase in microbial enzyme activity than the effect of applying either in isolation. Six soil cores were randomly and aseptically collected from each plot for microbial enzyme activity analysis from the 0-150 mm layer of a field trial under a dryland crop rotation system in the Swartland region. The activities of four microbial enzymes, β-glucosidase, acid phosphatase, alkaline phosphatase and urease, were assessed. The enzymes are essential for the cycling of glucose, phosphorus, and nitrogen, respectively. Microbial enzyme activity generally increased with a reduction of both tillage intensity and synthetic agrochemical application. The use of the mouldboard plough led to the least (P<0.05) microbial enzyme activity relative to the reduced tillage treatments, whereas the system with bio-stimulants (reduced synthetic agrochemicals) led to the highest (P<0.05) microbial enzyme activity relative to the standard systems. The application of bio-stimulants in combination with reduced tillage, particularly no-tillage, could be beneficial for enzyme activity in a dryland farming system.

Keywords: bio-stimulants, soil microbial enzymes, synthetic agrochemicals, tillage

Procedia PDF Downloads 74

Authors: Sanyka Banerjee, Saikat Nandi, P. K. Dan

Abstract:

Synchronizers as an integral part of gearbox is a key element in the transmission system in automotive. The performance of synchronizer affects transmission efficiency and driving comfort. Synchronizing mechanism as a major component of transmission system must be capable of preventing vibration and noise in the gears. Gear shifting efficiency improvement with an aim to achieve smooth, quick and energy efficient power transmission remains a challenge for the automotive industry. Performance of the synchronizer is dependent on the features and characteristics of its sub-components and therefore analysis of the contribution of such characteristics is necessary. An important exercise involved is to identify all such characteristics or factors which are associated with the modeling and analysis and for this purpose the literature was reviewed, rather extensively, to study the mathematical models, formulated considering such. It has been observed that certain factors are rather common across models; however, there are few factors which have specifically been selected for individual models, as reported. In order to obtain a more realistic model, an attempt here has been made to identify and assimilate practically all possible factors which may be considered in formulating the model more comprehensively. A simulation study, formulated as a block model, for such analysis has been carried out in a reliable environment like MATLAB. Lower synchronization time is desirable and hence, it has been considered here as the output factors in the simulation modeling for evaluating transmission efficiency. An improved synchronizer model requires optimized values of sub-component design parameters. A parametric optimization utilizing Taguchi’s design of experiment based response data and their analysis has been carried out for this purpose. The effectiveness of the optimized parameters for the improved synchronizer performance has been validated by the simulation study of the synchronizer block model with improved parameter values as input parameters for better transmission efficiency and driver comfort.

Keywords: design of experiments, modeling, parametric optimization, simulation, synchronizer

Procedia PDF Downloads 306