Search results for: gradient method

Authors: Zeynep Yazgan, Gökhan Aygün, Reyhan Çalışkan

Abstract:

Objective: Invasive fungal infections are a serious threat in terms of morbidity and mortality, especially in immunocompromised patients. The most frequently isolated agents are Aspergillus genus fungi, and sensitivity to azoles, which are the first choice in treatment, decreases. In our study, we aimed to investigate the use of the agar plate screening method as a fast, easy, and practical method in determining azole resistance in Aspergillus spp. species. Methods: Our study was conducted with 125 Aspergillus spp. isolates produced from various clinical samples. Aspergillus spp. isolates were identified by conventional methods and azole resistance was determined by gradient test and agar plate screening method. Broth microdilution method was applied to resistant isolates, and CypA-L98H and CypA-M220 mutations in the cyp51A gene were investigated. Results: In our study, 55 A. fumigatus complex (44%), 42 A. flavus (33.6%), 6 A. terreus (5%), 4 A. niger (3%) and 18 Aspergillus spp. (14%) were identified. With the gradient test method, resistance to VOR and POS was detected in 1 (1.8%) of A.fumigatus isolates, and resistance to ITR was detected in 3 (5.45%). With the agar plate method, 1 of the A.fumigatus isolates (1.8%) had VOR, ITR, POS, 1 of the A.terreus isolates (16.7%) had VOR, 1 of the A.niger isolates (25%) had ITR. Resistance to VOR and POS was detected in 2 Aspergillus spp. isolates (11%), and resistance to ITR was detected in 1 (5.6%). Sensitivity and specificity were determined as 100% for VOR and POS in A. fumigatus species, 33.3% and 100% for ITR, respectively, 100% for ITR in A. flavus species, and 100% for ITR and POS in A. terreus species. By broth microdilution method in 7 isolates in which resistance was detected by gradient test and/or agar plate screening method; 1 A.fumigatus resistant to ITR, VOR, POS, 2 A.fumigatus resistant to ITR, 2 Aspergillus spp. ITR, VOR, POS MICs were determined as 2µg/ml and 8µg/ml, 8µg/ml and >32µg/ml, 0.5µg/ml and 4µg/ml, respectively. CypA-L98H mutations were detected in 5 of these isolates, CypA-M220 mutations were detected in 6, and no mutation was detected in 1. CypA-L98H and CypA-M220 mutations were detected in 1 isolate for which resistance was not detected. Conclusion: The need for rapid antifungal susceptibility screening tests is increasing in the treatment of aspergillosis. Although the sensitivity of the agar plate method was determined to be 33.3% for A.fumigatus ITR in our study, its sensitivity and specificity were determined to be 100% for ITR, VOR, and POS in other species. The low sensitivity value detected for A.fumigatus showed that agar plate drug concentrations should be updated in accordance with the latest regulations of EUCAST guidelines. The CypA-L98H and CypA-M220 mutations detected in our study suggested that the distribution of azole resistance-related mutations in different regions in our country should be investigated. In conclusion, it is thought that the agar plate method, which can be easily applied to detect azole resistance, is a fast and practical method in routine use and can contribute to both the determination of effective treatment strategies and the generation of epidemiological data.

Keywords: Aspergillus, agar plate, azole resistance, cyp51A, cypA-L98H, cypA-M220

Procedia PDF Downloads 38

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 525

Authors: A. Archibong-Eso, J. Shi, Y Baba, S. Alagbe, W. Yan, H. Yeung

Abstract:

This study presents over 100 experiments conducted in a 25.4 mm internal diameter (ID) horizontal pipeline. Oil viscosity ranging from 3.5 Pa.s–5.0 Pa.s are used with superficial velocities of oil and water ranging from 0.06 to 0.55 m/s and 0.01 m/s to 1.0 m/s, respectively. Pressure gradient measurements and flow pattern observations are discussed. Numerical simulation of some flow conditions is performed using the commercial CFD code ANSYS Fluent® and the simulation results are compared with experimental results. Results indicate that CFD numerical simulation performed moderately well in predicting the flow configurations observed in this study while discrepancies were observed in the pressure gradient predictions.

Keywords: flow patterns, plug, pressure gradient, rivulet

Procedia PDF Downloads 394

Authors: R. O. Osaseri, A. R. Usiobaifo

Abstract:

The power transformer which is responsible for the voltage transformation is of great relevance in the power system and oil-immerse transformer is widely used all over the world. A prompt and proper maintenance of the transformer is of utmost importance. The dissolved gasses content in power transformer, oil is of enormous importance in detecting incipient fault of the transformer. There is a need for accurate prediction of the incipient fault in transformer oil in order to facilitate the prompt maintenance and reducing the cost and error minimization. Study on fault prediction and diagnostic has been the center of many researchers and many previous works have been reported on the use of artificial intelligence to predict incipient failure of transformer faults. In this study machine learning technique was employed by using gradient decent algorithms and Support Vector Machine (SVM) in predicting incipient fault diagnosis of transformer. The method focuses on creating a system that improves its performance on previous result and historical data. The system design approach is basically in two phases; training and testing phase. The gradient decent algorithm is trained with a training dataset while the learned algorithm is applied to a set of new data. This two dataset is used to prove the accuracy of the proposed model. In this study a transformer fault diagnostic model based on Support Vector Machine (SVM) and gradient decent algorithms has been presented with a satisfactory diagnostic capability with high percentage in predicting incipient failure of transformer faults than existing diagnostic methods.

Keywords: diagnostic model, gradient decent, machine learning, support vector machine (SVM), transformer fault

Procedia PDF Downloads 287

Authors: Zhenqiu Liu

Abstract:

Visualizing the heterogeneity within cell-populations for single-cell RNA-seq data is crucial for studying the functional diversity of a cell. However, because of the high level of noises, outlier, and dropouts, it is very challenging to measure the cell-to-cell similarity (distance), visualize and cluster the data in a low-dimension. Minimum volume embedding (MVE) projects the data into a lower-dimensional space and is a promising tool for data visualization. However, it is computationally inefficient to solve a semi-definite programming (SDP) when the sample size is large. Therefore, it is not applicable to single-cell RNA-seq data with thousands of samples. In this paper, we develop an efficient algorithm with an accelerated proximal gradient method and visualize the single-cell RNA-seq data efficiently. We demonstrate that the proposed approach separates known subpopulations more accurately in single-cell data sets than other existing dimension reduction methods.

Keywords: single-cell RNA-seq, minimum volume embedding, visualization, accelerated proximal gradient method

Procedia PDF Downloads 197

Authors: Morteza Mirhosseini, Amir B. Khoshnevis

Abstract:

The boundary layer separation and new active flow control of a NACA 0025 airfoil were studied experimentally. This new flow control is sometimes known as a co-flow jet (cfj) airfoil. This paper presents the fluctuating velocity in a wall jet over the co-flow jet airfoil subjected to an adverse pressure gradient and a curved surface. In these results, the fluctuating velocity at the inner part increasing by increased the angle of attack up to 12^o and this has due to the jet energized, while the angle of attack 20^o has different. The airfoil cord based Reynolds number has 10⁵.

Keywords: adverse pressure gradient, fluctuating velocity, wall jet, co-flow jet airfoil

Procedia PDF Downloads 459

Authors: Lican Wang, Rongqian Chen, Yancheng You, Ruofan Qiu

Abstract:

In the applications of jet engine, open-jet wind tunnel and airframe, there wildly exists a shear layer formed by the velocity and temperature gradients between jet flow and surrounded medium. The presence of shear layer will refract and reflect the sound path that consequently influences the measurement results in far-field. To investigate and evaluate the shear layer effect, a gradient term suppression and filtering method is adopted to simulate sound propagation through a steady sheared flow in three dimensions. Two typical configurations are considered: one is an incompressible and cold jet flow in wind tunnel and the other is a compressible and hot jet flow in turbofan engine. A numerically linear microphone array is used to localize the position of given sound source. The localization error is presented and linearly fitted.

Keywords: aeroacoustic, linearized Euler equation, acoustic propagation, source localization

Procedia PDF Downloads 154

Authors: Namita Mittal, Himakshi Shekhawat, Ankit Vidyarthi

Abstract:

In medical research study, doctors and radiologists face lot of complexities in analysing the brain tumors in Magnetic Resonance (MR) images. Brain tumor detection is difficult due to amorphous tumor shape and overlapping of similar tissues in nearby region. So, radiologists require one such clinically viable solution which helps in automatic segmentation of tumor inside brain MR image. Initially, segmentation methods were used to detect tumor, by dividing the image into segments but causes loss of information. In this paper, a hybrid method is proposed which detect Region of Interest (ROI) on the basis of difference in intensity values and texture values of tumor region using nearby tissues with Gradient Vector Flow (GVF) technique in the identification of ROI. Proposed approach uses both intensity and texture values for identification of abnormal section of the brain MR images. Experimental results show that proposed method outperforms GVF method without any loss of information.

Keywords: brain tumor, GVF, intensity, MR images, segmentation, texture

Procedia PDF Downloads 402

Authors: Pu Ying-Chih, Yang Yin-Ju, Hang Jian-Yi, Jang Guang-Way

Abstract:

Organic-Inorganic hybrid materials consisting of graded distributions of inorganic nano particles in organic polymer matrices were successfully prepared by the sol-gel process. Optical and surface properties of the resulting nano composites can be manipulated by changing their compositions and nano particle distribution gradients. Applications of gradient nano composite materials include sealants for LED packaging and screen lenses for smartphones. Optical transparency, prism coupler, TEM, SEM, Energy Dispersive X-ray Spectrometer (EDX), Izod impact strength, conductivity, pencil hardness, and thermogravimetric characterizations of the nano composites were performed and the results will be presented.

Keywords: Gradient, Hybrid, Nanocomposite, Organic-Inorganic

Procedia PDF Downloads 477

Authors: U. Amin, A. Talib, S. A. Qureshi, M. J. Hossain, G. Ahmad

Abstract:

The choice evaluation between oil-immersed and dry-type transformers is often controlled by cost, location, and application. This paper compares the electrical performance of liquid- filled and dry-type transformers, which will assist the customer to choose the right and efficient ones for particular applications. An accurate assessment of the time-average flux density, electric field intensity and voltage distribution in an oil-insulated and a dry-type transformer have been computed and investigated. The detailed transformer modeling and analysis has been carried out to determine electrical parameter distributions. The models of oil-immersed and dry-type transformers are developed and solved by using the finite element method (FEM) to compare the electrical parameters. The effects of non-uniform and non-coherent voltage gradient, flux density and electric field distribution on the power losses and insulation properties of transformers are studied in detail. The results show that, for the same voltage and kilo-volt-ampere (kVA) rating, oil-immersed transformers have better insulation properties and less hysteresis losses than the dry-type.

Keywords: finite element method, flux density, transformer, voltage gradient

Procedia PDF Downloads 247

Authors: Likewin Thomas, M. V. Manoj Kumar, B. Annappa

Abstract:

The aim of this study was to develop an Artificial Neural Network0 s recommendation model for an online process using the complexity of load, performance, and average servicing time of the resources. Here, the proposed model investigates the resource performance using stochastic gradient decent method for learning ranking function. A probabilistic cost function is implemented to identify the optimal θ values (load) on each resource. Based on this result the recommendation of resource suitable for performing the currently executing task is made. The test result of CoSeLoG project is presented with an accuracy of 72.856%.

Keywords: ADALINE, neural network, gradient decent, process mining, resource behaviour, polynomial regression model

Procedia PDF Downloads 354

Authors: Bingxi Huang, Yiqing Li, Marek Morzynski, Bernd R. Noack

Abstract:

We propose a Kriging-based global optimization method for active flow control with multiple actuation parameters. This method is designed to converge quickly and avoid getting trapped into local minima. We follow the model-free explorative gradient method (EGM) to alternate between explorative and exploitive steps. This facilitates a convergence similar to a gradient-based method and the parallel exploration of potentially better minima. In contrast to EGM, both kinds of steps are performed with Kriging surrogate model from the available data. The explorative step maximizes the expected improvement, i.e., favors regions of large uncertainty. The exploitive step identifies the best location of the cost function from the Kriging surrogate model for a subsequent weight-biased linear-gradient descent search method. To verify the effectiveness and robustness of the improved Kriging-based optimization method, we have examined several comparative test problems of varying dimensions with limited evaluation budgets. The results show that the proposed algorithm significantly outperforms some model-free optimization algorithms like genetic algorithm and differential evolution algorithm with a quicker convergence for a given budget. We have also performed direct numerical simulations of the fluidic pinball (N. Deng et al. 2020 J. Fluid Mech.) on three circular cylinders in equilateral-triangular arrangement immersed in an incoming flow at Re=100. The optimal cylinder rotations lead to 44.0% net drag power saving with 85.8% drag reduction and 41.8% actuation power. The optimal results for active flow control based on this configuration have achieved boat-tailing mechanism by employing Coanda forcing and wake stabilization by delaying separation and minimizing the wake region.

Keywords: direct numerical simulations, flow control, kriging, stochastic optimization, wake stabilization

Procedia PDF Downloads 79

Authors: Sol Girouard, Zona Kostic

Abstract:

A home is often the largest and most expensive purchase a person makes. Whether the decision leads to a successful outcome will be determined by a combination of critical factors. In this paper, we propose a method that efficiently handles all the factors in residential real estate and performs predictions given a feature space with high dimensionality while controlling for overfitting. The proposed method was built on gradient descent and boosting algorithms and uses a mixed optimizing technique to improve the prediction power. Usually, a single model cannot handle all the cases thus our approach builds multiple models based on different subsets of the predictors. The algorithm was tested on 3 million homes across the U.S., and the experimental results demonstrate the efficiency of this approach by outperforming techniques currently used in forecasting prices. With everyday changes on the real estate market, our proposed algorithm capitalizes from new events allowing more efficient predictions.

Keywords: real estate prediction, gradient descent, boosting, ensemble methods, active learning, training

Procedia PDF Downloads 245

Authors: Traore Papa Touty

Abstract:

This article presents a numerical method for determining the thermal conductivity of local materials, kapok plaster and tow plaster. It consists of heating the front face of a wall made from these two materials and at the same time insulating its rear face. We simultaneously study the curves of the evolution of the heat flux density as a function of time on the rear face and the evolution of the temperature gradient as a function of time between the heated face and the insulated face. Thermal conductivity is obtained when reaching a steady state when the evolution of the heat flux density and the temperature gradient no longer depend on time. The results showed that the theoretical value of thermal conductivity is obtained when the material has reached its equilibrium state. And the values obtained for different values of the convective exchange coefficients are appreciably equal to the experimental value.

Keywords: thermal conductivity, numerical method, heat exchange coefficient, transitional regime

Procedia PDF Downloads 184

Authors: Huda Algharib, Amal Algharib, Hanan Algharib, Ali Mohammad Alqudah

Abstract:

Image registration is an important topic for many imaging systems and computer vision applications. The standard image registration techniques such as Mutual information/ Normalized mutual information -based methods have a limited performance because they do not consider the spatial information or the relationships between the neighbouring pixels or voxels. In addition, the amount of image noise may significantly affect the registration accuracy. Therefore, this paper proposes an efficient method that explicitly considers the relationships between the adjacent pixels, where the gradient information of the reference and scene images is extracted first, and then the cosine similarity of the extracted gradient information is computed and used to improve the accuracy of the standard normalized mutual information measure. Our experimental results on different data types (i.e. CT, MRI and thermal images) show that the proposed method outperforms a number of image registration techniques in terms of the accuracy.

Keywords: image registration, mutual information, image gradients, image transformations

Procedia PDF Downloads 216

Authors: Jisun Lee, Jay Hyoun Kwon

Abstract:

As an alternative way to compensate the INS (inertial navigation system) error in non-GNSS (Global Navigation Satellite System) environment, geophysical database referenced navigation is being studied. In this study, both gravity gradient and terrain data were combined to complement the weakness of sole geophysical data as well as to improve the stability of the positioning. The main process to compensate the INS error using geophysical database was constructed on the basis of the EKF (Extended Kalman Filter). In detail, two type of combination method, centralized and decentralized filter, were applied to check the pros and cons of its algorithm and to find more robust results. The performance of each navigation algorithm was evaluated based on the simulation by supposing that the aircraft flies with precise geophysical DB and sensors above nine different trajectories. Especially, the results were compared to the ones from sole geophysical database referenced navigation to check the improvement due to a combination of the heterogeneous geophysical database. It was found that the overall navigation performance was improved, but not all trajectories generated better navigation result by the combination of gravity gradient with terrain data. Also, it was found that the centralized filter generally showed more stable results. It is because that the way to allocate the weight for the decentralized filter could not be optimized due to the local inconsistency of geophysical data. In the future, switching of geophysical data or combining different navigation algorithm are necessary to obtain more robust navigation results.

Keywords: Extended Kalman Filter, geophysical database referenced navigation, gravity gradient, terrain

Procedia PDF Downloads 316

Authors: Jeong-Min Lee, Hyunwoo Song, Yonseok Kim, Junghan Yun, Jungin Byun, Jae-Mean Koo, Chang-Sung Seok

Abstract:

The Thermal Barrier Coating (TBC) prevents heat directly transferring from the high-temperature flame to the substrate. Top coat and bond coat compose the TBC and top coat consists of a ceramic and bond coat increases adhesion between the top coat and the substrate. The TBC technology drops the substrate surface temperature by about 150~200°C. In addition, the TBC system has a cooling system to lower the blade temperature by the air flow inside the blade. Then, as a result, the thermal gradient occurs inside the blade by cooling. Also, the internal stress occurs due to the difference in thermal expansion. In this paper, the finite element analyses (FEA) were performed and stress changes were derived according to the thermal gradient of the TBC system. The stress was increased due to the cooling, but difference of the stress between the top coat and bond coat was decreased. So, delamination in the interface between top coat and bond coat.

Keywords: gas turbine blade, Thermal Barrier Coating (TBC), thermal gradient, Finite Element Analysis (FEA)

Procedia PDF Downloads 580

Authors: Muhammad Danladi, Muhammad Bura Garba, Muhammad Yahaya, Dahiru Muhammad

Abstract:

Biomass is one of the primary sources of energy supply, which contributes to about 78% of Nigeria. In this work, a comparative analysis of the heating values of sawdust and rice husks in the thermal generation of electricity was carried out. In the study, different masses of biomass were used and the corresponding electromotive force in millivolts was obtained. A graph of e.m.f was plotted against the mass of each biomass and a gradient was obtained. Bar graphs were plotted to represent the values of e.m.f and masses of the biomass. Also, a graph of e.m.f against eating values of sawdust and rice husks was plotted, and in each case, as the e.m.f increases also, the heating values increases. The result shows that saw dust with 0.033Mv/g gradient and 3.5 points of intercept had the highest gradient, followed by rice husks with 0.026Mv/g gradient and 2.6 points of intercept. It is, therefore, concluded that sawdust is the most efficient of the two types of biomass in the thermal generation of electricity.

Keywords: biomass, electricity, thermal, generation

Procedia PDF Downloads 63

Authors: Sung-Sik Park, Han Chang, Kyung-Hun Chae, Sae-Byeok Lee

Abstract:

In this study, a three-dimensional (3D) Particle method without using grid was developed for analyzing large deformation of soils instead of using ordinary finite element method (FEM) or finite difference method (FDM). In the 3D Particle method, the governing equations were discretized by various particle interaction models corresponding to differential operators such as gradient, divergence, and Laplacian. The Mohr-Coulomb failure criterion was incorporated into the 3D Particle method to determine soil failure. The yielding and hardening behavior of soil before failure was also considered by varying viscosity of soil. First of all, an unconfined compression test was carried out and the large deformation following soil yielding or failure was simulated by the developed 3D Particle method. The results were also compared with those of a commercial FEM software PLAXIS 3D. The developed 3D Particle method was able to simulate the 3D large deformation of soils due to soil yielding and calculate the variation of normal and shear stresses following clay deformation.

Keywords: particle method, large deformation, soil column, confined compressive stress

Procedia PDF Downloads 547

Authors: Armin Rahimi

Abstract:

In this study, we reveal co-seismic signals of two combined earthquakes, the 2006 Mw8.3 thrust and 2007 Mw8.1 normal fault earthquakes of the central Kuril Islands from GRACE observations. We compute monthly full gravitational gradient tensor in the local north-east-down frame for Kuril Islands earthquakes without spatial averaging and de-striping filters. Some of the gravitational gradient components (e.g. ΔVxx, ΔVxz) enhance high frequency components of the earth gravity field and reveal more details in spatial and temporal domain. Therefore that preseismic activity can be better illustrated. We show that the positive-negative-positive co-seismic ΔVxx due to the Kuril Islands earthquakes ranges from − 0.13 to + 0.11 milli Eötvös, and ΔVxz shows a positive-negative-positive pattern ranges from − 0.16 to + 0.13 milli Eötvös, agree well with seismic model predictions.

Keywords: GRACE observation, gravitational gradient changes, Kuril island earthquakes, PSGRN/PSCMP

Procedia PDF Downloads 240

Authors: A. K. Borah, A. K. Singh

Abstract:

In this paper we deal building blocks of the computer simulation of the multiphase flows. Whole simulation procedure can be viewed as two super procedures; The implementation of VOF method and the solution of Navier Stoke’s Equation. Moreover, a sequential code for a Navier Stoke’s solver has been studied.

Keywords: Bi-conjugate gradient stabilized (Bi-CGSTAB), ILUT function, krylov subspace, multifluid flows preconditioner, simple algorithm

Procedia PDF Downloads 494

Authors: Zhifeng Kong

Abstract:

Over-parameterized neural networks have attracted a great deal of attention in recent deep learning theory research, as they challenge the classic perspective of over-fitting when the model has excessive parameters and have gained empirical success in various settings. While a number of theoretical works have been presented to demystify properties of such models, the convergence properties of such models are still far from being thoroughly understood. In this work, we study the convergence properties of training two-hidden-layer partially over-parameterized fully connected networks with the Rectified Linear Unit activation via gradient descent. To our knowledge, this is the first theoretical work to understand convergence properties of deep over-parameterized networks without the equally-wide-hidden-layer assumption and other unrealistic assumptions. We provide a probabilistic lower bound of the widths of hidden layers and proved linear convergence rate of gradient descent. We also conducted experiments on synthetic and real-world datasets to validate our theory.

Keywords: over-parameterization, rectified linear units ReLU, convergence, gradient descent, neural networks

Procedia PDF Downloads 107

Authors: O. W. Lawal, L. O. Ahmed, Y. K. Ali

Abstract:

The unsteady MHD Poiseuille flow of a third grade fluid between two parallel horizontal nonconducting porous plates is studied with heat transfer. The two plates are fixed but maintained at different constant temperature with the Joule and viscous dissipation taken into consideration. The fluid motion is produced by a sudden uniform exponential decaying pressure gradient and external uniform magnetic field that is perpendicular to the plates. The momentum and energy equations governing the flow are solved numerically using Maple program. The effects of magnetic field and third grade fluid parameters on velocity and temperature profile are examined through several graphs.

Keywords: exponential decaying pressure gradient, MHD flow, Poiseuille flow, third grade fluid

Procedia PDF Downloads 444

Authors: Bastien Batardière, Joon Kwon

Abstract:

For finite-sum optimization, variance-reduced gradient methods (VR) compute at each iteration the gradient of a single function (or of a mini-batch), and yet achieve faster convergence than SGD thanks to a carefully crafted lower-variance stochastic gradient estimator that reuses past gradients. Another important line of research of the past decade in continuous optimization is the adaptive algorithms such as AdaGrad, that dynamically adjust the (possibly coordinate-wise) learning rate to past gradients and thereby adapt to the geometry of the objective function. Variants such as RMSprop and Adam demonstrate outstanding practical performance that have contributed to the success of deep learning. In this work, we present AdaLVR, which combines the AdaGrad algorithm with loopless variance-reduced gradient estimators such as SAGA or L-SVRG that benefits from a straightforward construction and a streamlined analysis. We assess that AdaLVR inherits both good convergence properties from VR methods and the adaptive nature of AdaGrad: in the case of L-smooth convex functions we establish a gradient complexity of O(n + (L + √ nL)/ε) without prior knowledge of L. Numerical experiments demonstrate the superiority of AdaLVR over state-of-the-art methods. Moreover, we empirically show that the RMSprop and Adam algorithm combined with variance-reduced gradients estimators achieve even faster convergence.

Keywords: convex optimization, variance reduction, adaptive algorithms, loopless

Procedia PDF Downloads 26

Authors: Martin K. Steiger, Lukas Heisler, Hans-Georg Brachtendorf

Abstract:

Deep neural networks and their variants form the backbone of many AI applications. Based on the so-called residual networks, a continuous formulation of such models as ordinary differential equations (ODEs) has proven advantageous since different techniques may be applied that significantly increase the learning speed and enable controlled trade-offs with the resulting error at the same time. For the evaluation of such models, high-performance numerical differential equation solvers are used, which also provide the gradients required for training. However, whether classical gradient-based methods are even applicable or which one yields the best results has not been discussed yet. This paper aims to redeem this situation by providing empirical results for different applications.

Keywords: deep neural networks, gradient-based learning, image processing, ordinary differential equation networks

Procedia PDF Downloads 130

Authors: Alexander V. Spirov, David M. Holloway, Ekaterina M. Myasnikova

Abstract:

Segmentation of the early Drosophila embryo results from the dynamic establishment of spatial gene expression patterns. Patterning occurs on an embryo geometry which is a 'deformed' prolate ellipsoid, with anteroposterior and dorsal-ventral major and minor axes, respectively. Patterning is largely independent along each axis, but some interaction can be seen in the 'bending' of the segmental expression stripes. This interaction is not well understood. In this report, we investigate how 3D geometrical features of the early embryo affect the segmental expression patterning. Specifically, we study the effect of geometry on formation of the Bicoid primary morphogenetic gradient. Our computational results demonstrate that embryos with a much longer ventral than dorsal surface ('bellied') can produce curved Bicoid concentration contours which could activate curved stripes in the downstream pair-rule segmentation genes. In addition, we show that having an extended source for Bicoid in the anterior of the embryo may be necessary for producing the observed exponential form of the Bicoid gradient along the anteroposterior axis.

Keywords: Drosophila embryo, bicoid morphogenetic gradient, exponential expression profile, expression surface form, segmentation genes, 3D modelling

Procedia PDF Downloads 239

Authors: Kanchan Vishwakarma, Shivesh Sharma, Nitin Kumar

Abstract:

Soil is incoherent matter on Earth’s surface having organic and mineral content. The spatial variation of 4 soil enzyme activities and microbial biomass were assessed for two seasons’ viz. monsoon and winter along the latitudinal gradient in North-central India as the area of this study is fettered with respect to national status. The study was facilitated to encompass the effect of climate change, enzyme activity and biomass on nutrient cycling. Top soils were sampled from 4 sites in North-India. There were significant correlations found between organic C, N & P wrt to latitude gradient in two seasons. This distribution of enzyme activities and microbial biomass was consequence of alterations in temperature and moisture of soil because of which soil properties change along the latitude transect.

Keywords: latitude gradient, microbial biomass, moisture, soil, organic carbon, temperature

Procedia PDF Downloads 364

Authors: Gayathri Nagarajan, L. D. Dhinesh Babu

Abstract:

Health care is one of the prominent industries that generate voluminous data thereby finding the need of machine learning techniques with big data solutions for efficient processing and prediction. Missing data, incomplete data, real time streaming data, sensitive data, privacy, heterogeneity are few of the common challenges to be addressed for efficient processing and mining of health care data. In comparison with other applications, accuracy and fast processing are of higher importance for health care applications as they are related to the human life directly. Though there are many machine learning techniques and big data solutions used for efficient processing and prediction in health care data, different techniques and different frameworks are proved to be effective for different applications largely depending on the characteristics of the datasets. In this paper, we present a framework that uses ensemble machine learning technique gradient boosted trees for data classification in health care big data. The framework is built on Spark platform which is fast in comparison with other traditional frameworks. Unlike other works that focus on a single technique, our work presents a comparison of six different machine learning techniques along with gradient boosted trees on datasets of different characteristics. Five benchmark health care datasets are considered for experimentation, and the results of different machine learning techniques are discussed in comparison with gradient boosted trees. The metric chosen for comparison is misclassification error rate and the run time of the algorithms. The goal of this paper is to i) Compare the performance of gradient boosted trees with other machine learning techniques in Spark platform specifically for health care big data and ii) Discuss the results from the experiments conducted on datasets of different characteristics thereby drawing inference and conclusion. The experimental results show that the accuracy is largely dependent on the characteristics of the datasets for other machine learning techniques whereas gradient boosting trees yields reasonably stable results in terms of accuracy without largely depending on the dataset characteristics.

Keywords: big data analytics, ensemble machine learning, gradient boosted trees, Spark platform

Procedia PDF Downloads 216

Authors: Manjula Weerasekera, Chris Sissons, Lisa Wong, Sally Anderson, Ann Holmes, Richard Cannon

Abstract:

The aim of this study was to characterize bacterial population and yeasts in saliva by Polymerase chain reaction followed by denaturing gradient gel electrophoresis (PCR-DGGE) and measure yeast levels by culture. PCR-DGGE was performed to identify oral bacteria and yeasts in 24 saliva samples. DNA was extracted and used to generate DNA amplicons of the V2–V3 hypervariable region of the bacterial 16S rDNA gene using PCR. Further universal primers targeting the large subunit rDNA gene (25S-28S) of fungi were used to amplify yeasts present in human saliva. Resulting PCR products were subjected to denaturing gradient gel electrophoresis using Universal mutation detection system. DGGE bands were extracted and sequenced using Sanger method. A potential relationship was evaluated between groups of bacteria identified by cluster analysis of DGGE fingerprints with the yeast levels and with their diversity. Significant interpersonal variation of salivary microbiome was observed. Cluster and principal component analysis of the bacterial DGGE patterns yielded three significant major clusters, and outliers. Seventeen of the 24 (71%) saliva samples were yeast positive going up to 10³ cfu/mL. Predominately, C. albicans, and six other species of yeast were detected. The presence, amount and species of yeast showed no clear relationship to the bacterial clusters. Microbial community in saliva showed a significant variation between individuals. A lack of association between yeasts and the bacterial fingerprints in saliva suggests the significant ecological person-specific independence in highly complex oral biofilm systems under normal oral conditions.

Keywords: bacteria, denaturing gradient gel electrophoresis, oral biofilm, yeasts

Procedia PDF Downloads 196

Authors: Jurriaan Gillissen

Abstract:

This work is concerned with stabilizing the numerical integration of the Navier-Stokes equation (NSE), backwards in time. Applications involve the detection of sources of, e.g., sound, heat, and pollutants. Stable reverse numerical integration of parabolic differential equations is also relevant for image de-blurring. While the literature addresses the reverse integration problem of the advection-diffusion equation, the problem of numerical reverse integration of the NSE has, to our knowledge, not yet been addressed. Owing to the presence of viscosity, the NSE is irreversible, i.e., when going backwards in time, the fluid behaves, as if it had a negative viscosity. As an effect, perturbations from the perfect solution, due to round off errors or discretization errors, grow exponentially in time, and reverse integration of the NSE is inherently unstable, regardless of using an implicit time integration scheme. Consequently, some sort of filtering is required, in order to achieve a stable, numerical, reversed integration. The challenge is to find a filter with a minimal adverse affect on the accuracy of the reversed integration. In the present work, we explore an adjoint gradient method (AGM) to achieve this goal, and we apply this technique to two-dimensional (2D), decaying turbulence. The AGM solves for the initial velocity field u0 at t = 0, that, when integrated forward in time, produces a final velocity field u1 at t = 1, that is as close as is feasibly possible to some specified target field v1. The initial field u0 defines a minimum of a cost-functional J, that measures the distance between u1 and v1. In the minimization procedure, the u0 is updated iteratively along the gradient of J w.r.t. u0, where the gradient is obtained by transporting J backwards in time from t = 1 to t = 0, using the adjoint NSE. The AGM thus effectively replaces the backward integration by multiple forward and backward adjoint integrations. Since the viscosity is negative in the adjoint NSE, each step of the AGM is numerically stable. Nevertheless, when applied to turbulence, the AGM develops instabilities, which limit the backward integration to small times. This is due to the exponential divergence of phase space trajectories in turbulent flow, which produces a multitude of local minima in J, when the integration time is large. As an effect, the AGM may select unphysical, noisy initial conditions. In order to improve this situation, we propose two remedies. First, we replace the integration by a sequence of smaller integrations, i.e., we divide the integration time into segments, where in each segment the target field v1 is taken as the initial field u0 from the previous segment. Second, we add an additional term (regularizer) to J, which is proportional to a high-order Laplacian of u0, and which dampens the gradients of u0. We show that suitable values for the segment size and for the regularizer, allow a stable reverse integration of 2D decaying turbulence, with accurate results for more then O(10) turbulent, integral time scales.

Keywords: time reversed integration, parabolic differential equations, adjoint gradient method, two dimensional turbulence

Procedia PDF Downloads 199