Search results for: neural interface

Authors: Mostafa Zandi, Ramin Mansouri

Abstract:

Spillways are one of the most important hydraulic structures of dams that provide the stability of the dam and downstream areas at the time of flood. Morning-Glory spillway is one of the common spillways for discharging the overflow water behind dams, these kinds of spillways are constructed in dams with small reservoirs. In this research, the hydraulic flow characteristics of a morning-glory spillways are investigated with CFD model. Two dimensional unsteady RANS equations were solved numerically using Finite Volume Method. The PISO scheme was applied for the velocity-pressure coupling. The mostly used two-equation turbulence models, k- and k-, were chosen to model Reynolds shear stress term. The power law scheme was used for discretization of momentum, k , and  equations. The VOF method (geometrically reconstruction algorithm) was adopted for interface simulation. The results show that the fine computational grid, the input speed condition for the flow input boundary, and the output pressure for the boundaries that are in contact with the air provide the best possible results. Also, the standard wall function is chosen for the effect of the wall function, and the turbulent model k -ε (Standard) has the most consistent results with experimental results. When the jet is getting closer to end of basin, the computational results increase with the numerical results of their differences. The lower profile of the water jet has less sensitivity to the hydraulic jet profile than the hydraulic jet profile. In the pressure test, it was also found that the results show that the numerical values of the pressure in the lower landing number differ greatly in experimental results. The characteristics of the complex flows over a Morning-Glory spillway were studied numerically using a RANS solver. Grid study showed that numerical results of a 57512-node grid had the best agreement with the experimental values. The desired downstream channel length was preferred to be 1.5 meter, and the standard k-ε turbulence model produced the best results in Morning-Glory spillway. The numerical free-surface profiles followed the theoretical equations very well.

Keywords: morning-glory spillway, CFD model, hydraulic characteristics, wall function

Procedia PDF Downloads 77

Authors: Chao Wang

Abstract:

Under rapid innovation of information technology, the media plays a very important role in the dissemination of information, and it has a totally different analogy generations face. However, the involvement of narrative images provides more possibilities of narrative text. "Images" through the process of aperture, a camera shutter and developable photosensitive processes are manufactured, recorded and stamped on paper, displayed on a computer screen-concretely saved. They exist in different forms of files, data, or evidence as the ultimate looks of events. By the interface of media and network platforms and special visual field of the viewer, class body space exists and extends out as thin as sebum layer, extremely soft and delicate with real full tension. The physical space of sebum layer of confuses the fact that physical objects exist, needs to be established under a perceived consensus. As at the scene, the existing concepts and boundaries of physical perceptions are blurred. Sebum layer physical simulation shapes the “Topical-Similarity" immersing, leading the contemporary social practice communities, groups, network users with a kind of illusion without the presence, i.e. a non-real illusion. From the investigation and discussion of literatures, digital movies editing manufacture and produce the variability characteristics of time (for example, slices, rupture, set, and reset) are analyzed. Interactive eBook has an unique interaction in "Waiting-Greeting" and "Expectation-Response" that makes the operation of image narrative structure more interpretations functionally. The works of digital editing and interactive technology are combined and further analyze concept and results. After digitization of Interventional Imaging and interactive technology, real events exist linked and the media handing cannot be cut relationship through movies, interactive art, practical case discussion and analysis. Audience needs more rational thinking about images carried by the authenticity of the text.

Keywords: sebum layer, topical and similarity, interactive technology, image narrative

Procedia PDF Downloads 389

Authors: Tamas Molnar

Abstract:

Electronic government systems are currently in the same development stage as e-commerce applications were about in the late 1990s. Wide adoption by the majority of population is near, as such services are not only more and more desired by the users, but also strongly advocated and pushed by the state, as a means to increase effectiveness and cut expenses at the same time. Diffusion is however hampered by the low motivation caused by usability issues which will cause more and more frustration as the general population ages. Usability centred design is essential when creating such services. Elderly users, who have statistically the least experience, have the most problems, and therefore reject unusable systems first. The goal of our research was to find a way to map the needs of the elderly and create guidelines for the design of electronic government systems which are usable for the whole population. The first phase of our research, started mid-2009, was centred on the idea to gather information about the needs of the target group, in both Germany and Hungary with over 70 participants. This was done with the help of scenarios, interviews and questionnaires. The supplied data enabled to choose an eGovernment system for tests on the target group. Tests conducted in Germany and Hungary were based on the design and functions of the German electronic ID card, in the native languages. Scenarios mirroring common, every day transactions requiring an identification procedure were used. The obtained results allowed us to develop a generalised solution, the IGUAN guideline. This guideline makes a standardised approach to the usability improvement process possible. It contains the special requirements of elderly users, and a catalogue of criteria, which helps to develop an application in line with the set requirements. The third phase of our research was used a proof of concept for the IGUAN. The guideline was evaluated and tested with an iterative prototyping. The successful completion of this phase indicates that the IGUAN can be used to measurably increase the acceptance of e-government systems by elderly users. We could therefore demonstrate that improvements in the interface make e-government application possible which are perceived useful and easy to use by elderly users. These improvements will measurably increase the user motivation and experience. This can however only be achieved with a structured design process, and requires a framework which takes the requirements of the elderly users into account.

Keywords: e-Government, usability, acceptance, guidelines

Procedia PDF Downloads 544

Authors: Sonia Sassi, Mostapha Tarfaoui, Hamza Ben Yahia

Abstract:

In this investigation, an experimental technique in which the dynamic response, damage kinetic and heat dissipation are measured simultaneously during high strain rates on adhesively bonded joints materials. The material used in this study is widely used in the design of structures for military applications. It was composed of a 45° Bi-axial fiber-glass mat of 0.286 mm thickness in a Polyester resin matrix. In adhesive bonding, a NORPOL Polyvinylester of 1 mm thickness was used to assemble the composite substrate. The experimental setup consists of a compression Split Hopkinson Pressure Bar (SHPB), a high-speed infrared camera and a high-speed Fastcam rapid camera. For the dynamic compression tests, 13 mm x 13 mm x 9 mm samples for out-of-plane tests were considered from 372 to 1030 s-1. Specimen surface is controlled and monitored in situ and in real time using the high-speed camera which acquires the damage progressive in specimens and with the infrared camera which provides thermal images in time sequence. Preliminary compressive stress-strain vs. strain rates data obtained show that the dynamic material strength increases with increasing strain rates. Damage investigations have revealed that the failure mainly occurred in the adhesive/adherent interface because of the brittle nature of the polymeric adhesive. Results have shown the dependency of the dynamic parameters on strain rates. Significant temperature rise was observed in dynamic compression tests. Experimental results show that the temperature change depending on the strain rate and the damage mode and their maximum exceed 100 °C. The dependence of these results on strain rate indicates that there exists a strong correlation between damage rate sensitivity and heat dissipation, which might be useful when developing damage models under dynamic loading tacking into account the effect of the energy balance of adhesively bonded joints.

Keywords: adhesive bonded joints, Hopkinson bars, out-of-plane tests, dynamic compression properties, damage mechanisms, heat dissipation

Procedia PDF Downloads 212

Authors: Soumen Mitra, Aparna Saha

Abstract:

The perceived link between the city and the countryside is evolving rapidly and is getting shifted away from the assumptions of mainstream paradigms to new conceptual networks where rural-urban links are being redefined. In this conceptual field, the fringe interface is still considered as a transitional zone between city and countryside, and is defined as a diffused area rather than a discrete territory. In developing countries fringe areas are said to have both rural and urban characteristics but are devoid of basic municipal facilities. Again, when the urban core areas envelopes the fringe areas along with it the character of fringe changes but services are not well facilitated which in turn results to uneven growth, rapid and haphazard development. One of the major services present in fringe areas is inter-linkages in terms of transit corridors. Planning for the appropriate and sustainable future of fringe areas requires a sheer focus on these corridors pertaining to transit facility, for better accessibility and mobility. Inducing a transit facility plan enhances the various facilities and also increases their proximity for user groups. The study focuses on the western fringe region of Kolkata metropolis which is a major source of industrial hub and housing sector, thus converting the agricultural lands into non-agricultural use. The study emphasizes on providing transit facilities both physical (stops, sheds, terminals, etc.) and operational (ticketing system, route prioritization, integration of transit modes, etc.), to facilitate the region as well as accelerate the growth pattern systematically. Hence, the scope of this work is on the basis of prevailing conditions in fringe areas and attempts for an effective transit facility plan. The strategies and recommendations are in terms of road widening, service coverage, feeder route prioritization, bus stops facilitation, pedestrian facilities, etc, which in turn enhances the region’s growth pattern. Thus, this context of transit facility planning acts as a catalytic agent to avoid the future unplanned growth and accelerates it towards an integrated development.

Keywords: feeder route, fringe, municipal planning, transit facility

Procedia PDF Downloads 179

Authors: Sadia Munir

Abstract:

The aim of this study is to investigate changes in placental development and vascular dysfunction which subsequently affect feto-maternal health in pregnancies complicated by gestational diabetes mellitus (GDM). Fetal and postnatal adverse health outcomes of GDM are shown to be associated with disturbances in placental structure and function. Children of women with GDM are more likely to be obese and diabetic in childhood and adulthood. GDM also increases the risk of adverse pregnancy outcomes, including preeclampsia, birth injuries, macrosomia and neonatal hypoglycemia, respiratory distress syndrome, neonatal cardiac dysfunction and stillbirth. Incidences of type 2 diabetes in the MENA region are growing at an alarming rate which is estimated to become more than double by 2030. Five of the top 10 countries for diabetes prevalence in 2010 were in the Gulf region. GDM also increases the risk of development of type 2 diabetes. Interestingly, more than half of the women with GDM develop diabetes later in their life. The human placenta is a temporary organ located at the interface between mother and fetal blood circulation. Placenta has a central role as both a producer as well as a target of several molecules that are involved in placental development and function. We have investigated performed a Pubmed search with key words placenta, GDM, placental villi, vascularization, cytokines, growth factors, inflammation, hypoxia, oxidative stress and pathophysiology. We have investigated differences in the development and vascularization of placenta, their underlying causes and impact on feto-maternal health through literature review. We have also identified gaps in the literature and research questions that need to be answered to completely understand the central role of placenta in the GDM. This study is important in understanding the pathophysiology of placenta due to changes in the vascularization of villi, surface area and diameter of villous capillaries in pregnancies complicated by GDM. It is necessary to understand these mechanisms in order to develop treatments to reverse their effects on placental malfunctioning, which in turn, will result in improved mother and child health.

Keywords: gestational diabetes mellitus, placenta, vasculature, villi

Procedia PDF Downloads 318

Authors: Alexander A. Karabutov, Natalia B. Podymova, Elena B. Cherepetskaya

Abstract:

The theoretical analysis is carried out to get the relation between the ultrasonic wave velocity and the value of residual stresses. The laser-ultrasonic method is developed to evaluate the residual stresses and subsurface defects in metals. The method is based on the laser thermooptical excitation of longitudinal ultrasonic wave sand their detection by a broadband piezoelectric detector. A laser pulse with the time duration of 8 ns of the full width at half of maximum and with the energy of 300 µJ is absorbed in a thin layer of the special generator that is inclined relative to the object under study. The non-uniform heating of the generator causes the formation of a broadband powerful pulse of longitudinal ultrasonic waves. It is shown that the temporal profile of this pulse is the convolution of the temporal envelope of the laser pulse and the profile of the in-depth distribution of the heat sources. The ultrasonic waves reach the surface of the object through the prism that serves as an acoustic duct. At the interface ‚laser-ultrasonic transducer-object‘ the conversion of the most part of the longitudinal wave energy takes place into the shear, subsurface longitudinal and Rayleigh waves. They spread within the subsurface layer of the studied object and are detected by the piezoelectric detector. The electrical signal that corresponds to the detected acoustic signal is acquired by an analog-to-digital converter and when is mathematically processed and visualized with a personal computer. The distance between the generator and the piezodetector as well as the spread times of acoustic waves in the acoustic ducts are the characteristic parameters of the laser-ultrasonic transducer and are determined using the calibration samples. There lative precision of the measurement of the velocity of longitudinal ultrasonic waves is 0.05% that corresponds to approximately ±3 m/s for the steels of conventional quality. This precision allows one to determine the mechanical stress in the steel samples with the minimal detection threshold of approximately 22.7 MPa. The results are presented for the measured dependencies of the velocity of longitudinal ultrasonic waves in the samples on the values of the applied compression stress in the range of 20-100 MPa.

Keywords: laser-ultrasonic method, longitudinal ultrasonic waves, metals, residual stresses

Procedia PDF Downloads 326

Authors: Lucilla Crosta, Anthony Edwards

Abstract:

Companies and the market place nowadays struggle to find employees with adequate skills in relation to anticipated growth of their businesses. At least half of workers will need to undertake some form of up-skilling process in the next five years in order to remain aligned with the requests of the market . In order to meet these challenges, there is a clear need to explore the potential uses of AI (artificial Intelligence) based tools in assessing transversal skills (critical thinking, communication and soft skills of different types in general) of workers and adult students while empowering them to develop those same skills in a reliable trustworthy way. Companies seek workers with key transversal skills that can make a difference between workers now and in the future. However, critical thinking seems to be the one of the most imprtant skill, bringing unexplored ideas and company growth in business contexts. What employers have been reporting since years now, is that this skill is lacking in the majority of workers and adult students, and this is particularly visible trough their writing. This paper investigates how critical thinking and communication skills are currently developed in Higher Education environments through use of AI tools at postgraduate levels. It analyses the use of a branch of AI namely Machine Learning and Big Data and of Neural Network Analysis. It also examines the potential effect the acquisition of these skills through AI tools and what kind of effects this has on employability This paper will draw information from researchers and studies both at national (Italy & UK) and international level in Higher Education. The issues associated with the development and use of one specific AI tool Edulai, will be examined in details. Finally comparisons will be also made between these tools and the more recent phenomenon of Chat GPT and forthcomings and drawbacks will be analysed.

Keywords: critical thinking, artificial intelligence, higher education, soft skills, chat GPT

Procedia PDF Downloads 111

Authors: Lanchi Xie, Zhihui Li, Zhigang Li, Guiqiang Wang, Lei Xu, Yuwen Yan

Abstract:

Image-based facial features can be classified into category recognition features and individual recognition features. Current automated face recognition systems extract a specific feature vector of different dimensions from a facial image according to their pre-trained neural network. However, to improve the efficiency of parameter calculation, an algorithm generally reduces the image details by pooling. The operation will overlook the details concerned much by forensic experts. In our experiment, we adopted a variety of face recognition algorithms based on deep learning, compared a large number of naturally collected face images with the known data of the same person's frontal ID photos. Downscaling and manual handling were performed on the testing images. The results supported that the facial recognition algorithms based on deep learning detected structural and morphological information and rarely focused on specific markers such as stains and moles. Overall performance, distribution of genuine scores and impostor scores, and likelihood ratios were tested to evaluate the accuracy of biometric systems and forensic experts. Experiments showed that the biometric systems were skilled in distinguishing category features, and forensic experts were better at discovering the individual features of human faces. In the proposed approach, a fusion was performed at the score level. At the specified false accept rate, the framework achieved a lower false reject rate. This paper contributes to improving the interpretability of the objective method of facial comparison and provides a novel method for human-machine collaboration in this field.

Keywords: likelihood ratio, automated facial recognition, facial comparison, biometrics

Procedia PDF Downloads 130

Authors: Annalakshmi G., Sakthivel Murugan S.

Abstract:

This paper presents a local directional encoded derivative binary pattern (LDEDBP) feature extraction method that can be applied for the classification of submarine coral reef images. The classification of coral reef images using texture features is difficult due to the dissimilarities in class samples. In coral reef image classification, texture features are extracted using the proposed method called local directional encoded derivative binary pattern (LDEDBP). The proposed approach extracts the complete structural arrangement of the local region using local binary batten (LBP) and also extracts the edge information using local directional pattern (LDP) from the edge response available in a particular region, thereby achieving extra discriminative feature value. Typically the LDP extracts the edge details in all eight directions. The process of integrating edge responses along with the local binary pattern achieves a more robust texture descriptor than the other descriptors used in texture feature extraction methods. Finally, the proposed technique is applied to an extreme learning machine (ELM) method with a meta-heuristic algorithm known as weighted distance grey wolf optimizer (GWO) to optimize the input weight and biases of single-hidden-layer feed-forward neural networks (SLFN). In the empirical results, ELM-WDGWO demonstrated their better performance in terms of accuracy on all coral datasets, namely RSMAS, EILAT, EILAT2, and MLC, compared with other state-of-the-art algorithms. The proposed method achieves the highest overall classification accuracy of 94% compared to the other state of art methods.

Keywords: feature extraction, local directional pattern, ELM classifier, GWO optimization

Procedia PDF Downloads 164

Authors: Amin Naboureh, Ainong Li, Jinhu Bian, Guangbin Lei, Hamid Ebrahimy

Abstract:

Large-scale land cover mapping has become a new challenge in land change and remote sensing field because of involving a big volume of data. Moreover, selecting the right classification method, especially when there are different types of landscapes in the study area is quite difficult. This paper is an attempt to compare the performance of different machine learning (ML) algorithms for generating a land cover map of the China-Central Asia–West Asia Corridor that is considered as one of the main parts of the Belt and Road Initiative project (BRI). The cloud-based Google Earth Engine (GEE) platform was used for generating a land cover map for the study area from Landsat-8 images (2017) by applying three frequently used ML algorithms including random forest (RF), support vector machine (SVM), and artificial neural network (ANN). The selected ML algorithms (RF, SVM, and ANN) were trained and tested using reference data obtained from MODIS yearly land cover product and very high-resolution satellite images. The finding of the study illustrated that among three frequently used ML algorithms, RF with 91% overall accuracy had the best result in producing a land cover map for the China-Central Asia–West Asia Corridor whereas ANN showed the worst result with 85% overall accuracy. The great performance of the GEE in applying different ML algorithms and handling huge volume of remotely sensed data in the present study showed that it could also help the researchers to generate reliable long-term land cover change maps. The finding of this research has great importance for decision-makers and BRI’s authorities in strategic land use planning.

Keywords: land cover, google earth engine, machine learning, remote sensing

Procedia PDF Downloads 113

Authors: Asowata Osamede, Christo Pienaar, Johan Bekker

Abstract:

Mismatch in electrical energy (power) or outage from commercial providers, in general, does not promote development to the public and private sector, these basically limit the development of industries. The necessity for a well-structured photovoltaic (PV) system is of importance for an efficient and cost-effective monitoring system. The major renewable energy potential on earth is provided from solar radiation and solar photovoltaics (PV) are considered a promising technological solution to support the global transformation to a low-carbon economy and reduction on the dependence on fossil fuels. Solar arrays which consist of various PV module should be operated at the maximum power point in order to reduce the overall cost of the system. So power regulation and conditioning circuits should be incorporated in the set-up of a PV system. Power regulation circuits used in PV systems include maximum power point trackers, DC-DC converters and solar chargers. Inappropriate choice of power conditioning device in a basic off-grid PV system can attribute to power loss, hence the need for a right choice of power conditioning device to be coupled with the system of the essence. This paper presents the design and implementation of a power conditioning devices in order to improve the overall yield from the availability of solar energy and the system’s total efficiency. The power conditioning devices taken into consideration in the project includes the Buck and Boost DC-DC converters as well as solar chargers with MPPT. A logging interface circuit (LIC) is designed and employed into the system. The LIC is designed on a printed circuit board. It basically has DC current signalling sensors, specifically the LTS 6-NP. The LIC is consequently required to program the voltages in the system (these include the PV voltage and the power conditioning device voltage). The voltage is structured in such a way that it can be accommodated by the data logger. Preliminary results which include availability of power as well as power loss in the system and efficiency will be presented and this would be used to draw the final conclusion.

Keywords: tilt and orientation angles, solar chargers, PV panels, storage devices, direct solar radiation

Procedia PDF Downloads 136

Authors: Ruphi Naz, Altaf Ahmad, Mohammad Anis

Abstract:

Sialic acid (5-Acetylneuraminic acid, Neu5Ac) is a common sugar found as a terminal residue on glycoconjugates in many animals. Humans brain and the central nervous system contain the highest concentration of sialic acid (as N-acetylneuraminic acid) where these acids play an important role in neural transmission and ganglioside structure in synaptogenesis. Due to its important biological function, sialic acid is attracting increasing attention. To understand metabolic networks, fluxes and regulation, it is essential to be able to determine the cellular and subcellular levels of metabolites. Genetically-encoded fluorescence resonance energy transfer (FRET) sensors represent a promising technology for measuring metabolite levels and corresponding rate changes in live cells. Taking this, we developed a genetically encoded FRET (fluorescence resonance energy transfer) based nanosensor to analyse the sialic acid level in living cells. Sialic acid periplasmic binding protein (sia P) from Haemophilus influenzae was taken and ligated between the FRET pair, the cyan fluorescent protein (eCFP) and Venus. The chimeric sensor protein was expressed in E. coli BL21 (DE3) and purified by affinity chromatography. Conformational changes in the binding protein clearly confirmed the changes in FRET efficiency. So any change in the concentration of sialic acid is associated with the change in FRET ratio. This sensor is very specific to sialic acid and found stable with the different range of pH. This nanosensor successfully reported the intracellular level of sialic acid in bacterial cell. The data suggest that the nanosensors may be a versatile tool for studying the in vivo dynamics of sialic acid level non-invasively in living cells

Keywords: nanosensor, FRET, Haemophilus influenzae, metabolic networks

Procedia PDF Downloads 133

Authors: Xi Wang, Yoshio Bando

Abstract:

Lithium-ion batteries (LIBs) can deliver high levels of energy storage density and offer long operating lifetimes, but their power density is too low for many important applications. Therefore, we developed some new strategies and fabricated novel electrodes for fast Li transport and its facile synthesis including N-doped graphene-SnO2 sandwich papers, bicontinuous nanoporous Cu/Li4Ti5O12 electrode, and binder-free N-doped graphene papers. In addition, by using advanced in-TEM, STEM techniques and the theoretical simulations, we systematically studied and understood their storage mechanisms at the atomic scale, which shed a new light on the reasons of the ultrafast lithium storage property and high capacity for these advanced anodes. For example, by using advanced in-situ TEM, we directly investigated these processes using an individual CuO nanowire anode and constructed a LIB prototype within a TEM. Being promising candidates for anodes in lithium-ion batteries (LIBs), transition metal oxide anodes utilizing the so-called conversion mechanism principle typically suffer from the severe capacity fading during the 1st cycle of lithiation–delithiation. Also we report on the atomistic insights of the GN energy storage as revealed by in situ TEM. The lithiation process on edges and basal planes is directly visualized, the pyrrolic N "hole" defect and the perturbed solid-electrolyte-interface (SEI) configurations are observed, and charge transfer states for three N-existing forms are also investigated. In situ HRTEM experiments together with theoretical calculations provide a solid evidence that enlarged edge {0001} spacings and surface "hole" defects result in improved surface capacitive effects and thus high rate capability and the high capacity is owing to short-distance orderings at the edges during discharging and numerous surface defects; the phenomena cannot be understood previously by standard electron or X-ray diffraction analyses.

Keywords: in-situ TEM, STEM, advanced anode, lithium-ion batteries, storage mechanism

Procedia PDF Downloads 352

Authors: Fang Liu, JunShuai Xue, JiaJia Yao, GuanLin Wu, ZuMaoLi, XueYan Yang, HePeng Zhang, ZhiPeng Sun

Abstract:

Terahertz spectra is in great demand since last two decades for many photonic and electronic applications. III-Nitride resonant tunneling diode is one of the promising candidates for portable and compact THz sources. Room temperature microwave oscillator based on GaN/AlN resonant tunneling diode was reported in this work. The devices, grown by plasma-assisted molecular-beam epitaxy on free-standing c-plane GaN substrates, exhibit highly repeatable and robust negative differential resistance (NDR) characteristics at room temperature. To improve the interface quality at the active region in RTD, indium surfactant assisted growth is adopted to enhance the surface mobility of metal atoms on growing film front. Thanks to the lowered valley current associated with the suppression of threading dislocation scattering on low dislocation GaN substrate, a positive peak current density of record-high 2.1 MA/cm2 in conjunction with a peak-to-valley current ratio (PVCR) of 1.2 are obtained, which is the best results reported in nitride-based RTDs up to now considering the peak current density and PVCR values simultaneously. When biased within the NDR region, microwave oscillations are measured with a fundamental frequency of 0.31 GHz, yielding an output power of 5.37 µW. Impedance mismatch results in the limited output power and oscillation frequency described above. The actual measured intrinsic capacitance is only 30fF. Using a small-signal equivalent circuit model, the maximum intrinsic frequency of oscillation for these diodes is estimated to be ~260GHz. This work demonstrates a microwave oscillator based on resonant tunneling effect, which can meet the demands of terahertz spectral devices, more importantly providing guidance for the fabrication of the complex nitride terahertz and quantum effect devices.

Keywords: GaN resonant tunneling diode, peak current density, microwave oscillation, intrinsic capacitance

Procedia PDF Downloads 140

Authors: Yehjune Heo

Abstract:

As biometric systems become widely deployed, the security of identification systems can be easily attacked by various spoof materials. This paper contributes to finding a reliable and practical anti-spoofing method using Convolutional Neural Networks (CNNs) based on the types of loss functions and optimizers. The types of CNNs used in this paper include AlexNet, VGGNet, and ResNet. By using various loss functions including Cross-Entropy, Center Loss, Cosine Proximity, and Hinge Loss, and various loss optimizers which include Adam, SGD, RMSProp, Adadelta, Adagrad, and Nadam, we obtained significant performance changes. We realize that choosing the correct loss function for each model is crucial since different loss functions lead to different errors on the same evaluation. By using a subset of the Livdet 2017 database, we validate our approach to compare the generalization power. It is important to note that we use a subset of LiveDet and the database is the same across all training and testing for each model. This way, we can compare the performance, in terms of generalization, for the unseen data across all different models. The best CNN (AlexNet) with the appropriate loss function and optimizers result in more than 3% of performance gain over the other CNN models with the default loss function and optimizer. In addition to the highest generalization performance, this paper also contains the models with high accuracy associated with parameters and mean average error rates to find the model that consumes the least memory and computation time for training and testing. Although AlexNet has less complexity over other CNN models, it is proven to be very efficient. For practical anti-spoofing systems, the deployed version should use a small amount of memory and should run very fast with high anti-spoofing performance. For our deployed version on smartphones, additional processing steps, such as quantization and pruning algorithms, have been applied in our final model.

Keywords: anti-spoofing, CNN, fingerprint recognition, loss function, optimizer

Procedia PDF Downloads 137

Authors: Isaac Kayode Ogunlade

Abstract:

Data acquisition (DAQ) is the process by which physical phenomena from the real world are transformed into an electrical signal(s) that are measured and converted into a digital format for processing, analysis, and storage by a computer. The DAQ is designed using PIC18F4550 microcontroller, communicating with Personal Computer (PC) through USB (Universal Serial Bus). The research deployed initial knowledge of data acquisition system and embedded system to develop a weather data acquisition device using LM35 sensor to measure weather parameters and the use of Artificial Intelligence(Artificial Neural Network - ANN)and statistical approach(Autoregressive Integrated Moving Average – ARIMA) to predict precipitation (rainfall). The device is placed by a standard device in the Department of Meteorology, Federal University of Technology, Akure (FUTA) to know the performance evaluation of the device. Both devices (standard and designed) were subjected to 180 days with the same atmospheric condition for data mining (temperature, relative humidity, and pressure). The acquired data is trained in MATLAB R2012b environment using ANN, and ARIMAto predict precipitation (rainfall). Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Correction Square (R2), and Mean Percentage Error (MPE) was deplored as standardize evaluation to know the performance of the models in the prediction of precipitation. The results from the working of the developed device show that the device has an efficiency of 96% and is also compatible with Personal Computer (PC) and laptops. The simulation result for acquired data shows that ANN models precipitation (rainfall) prediction for two months (May and June 2017) revealed a disparity error of 1.59%; while ARIMA is 2.63%, respectively. The device will be useful in research, practical laboratories, and industrial environments.

Keywords: data acquisition system, design device, weather development, predict precipitation and (FUTA) standard device

Procedia PDF Downloads 92

Authors: Carol Anne Hargreaves

Abstract:

A key issue in stock investment is how to select representative features for stock selection. The objective of this paper is to firstly determine whether an automated stock investment system, using machine learning techniques, may be used to identify a portfolio of growth stocks that are highly likely to provide returns better than the stock market index. The second objective is to identify the technical features that best characterize whether a stock’s price is likely to go up and to identify the most important factors and their contribution to predicting the likelihood of the stock price going up. Unsupervised machine learning techniques, such as cluster analysis, were applied to the stock data to identify a cluster of stocks that was likely to go up in price – portfolio 1. Next, the principal component analysis technique was used to select stocks that were rated high on component one and component two – portfolio 2. Thirdly, a supervised machine learning technique, the logistic regression method, was used to select stocks with a high probability of their price going up – portfolio 3. The predictive models were validated with metrics such as, sensitivity (recall), specificity and overall accuracy for all models. All accuracy measures were above 70%. All portfolios outperformed the market by more than eight times. The top three stocks were selected for each of the three stock portfolios and traded in the market for one month. After one month the return for each stock portfolio was computed and compared with the stock market index returns. The returns for all three stock portfolios was 23.87% for the principal component analysis stock portfolio, 11.65% for the logistic regression portfolio and 8.88% for the K-means cluster portfolio while the stock market performance was 0.38%. This study confirms that an automated stock investment system using machine learning techniques can identify top performing stock portfolios that outperform the stock market.

Keywords: machine learning, stock market trading, logistic regression, cluster analysis, factor analysis, decision trees, neural networks, automated stock investment system

Procedia PDF Downloads 158

Authors: Alice Brites, Gerd Sparovek, Jean Paul Metzger, Ricardo Rodrigues

Abstract:

The segregation between science and policy in decision making process hinders nature conservation efforts worldwide. Scientists have been criticized for not producing information that leads to effective solutions for environmental problems. In an attempt to bridge this gap between science and practice, we conducted a project aimed at supporting the implementation of the Brazilian Native Vegetation Protection Law (NVPL) implementation in São Paulo State (SP), Brazil. To do so, we conducted multiple open meetings with the stakeholders involved in this discussion. Throughout this process, we raised stakeholders' demands for scientific information and brought feedbacks about our findings. However, our main scientific advice was not taken into account during the NVPL implementation in SP. The NVPL has a mechanism that exempts landholders who converted native vegetation without offending the legislation in place at the time of the conversion from restoration requirements. We found out that there were no accurate spatialized data for native vegetation cover before the 1960s. Thus, the initial benchmark for the mechanism application should be the 1965 Brazilian Forest Act. Even so, SP kept the 1934 Brazilian Forest Act as the initial legal benchmark for the law application. This decision implies the use of a probabilistic native vegetation map that has uncertainty and subjectivity as its intrinsic characteristics, thus its use can lead to legal queries, corruption, and an unfair benefit application. But why this decision was made even after the scientific advice was vastly divulgated? We raised some possible reasons to explain it. First, the decision was made during a government transition, showing that circumstantial political events can overshadow scientific arguments. Second, the debate about the NVPL in SP was not pacified and powerful stakeholders could benefit from the confusion created by this decision. Finally, the native vegetation protection mechanism is a complex issue, with many technical aspects that can be hard to understand for a non-specialized courtroom, such as the one that made the final decision at SP. This example shows that science and decision-makers still have a long way ahead to improve their way to interact and that science needs to find its way to be heard above the political buzz.

Keywords: Brazil, forest act, science-based dialogue, science-policy interface

Procedia PDF Downloads 122

Authors: Mpofu Vongai, Vhurumuku Elaosi

Abstract:

Many African countries, such as Zimbabwe and South Africa, have curricula reform agendas that include incorporation of Indigenous Knowledge and Nature of Science (NOS) into school Science, Technology, Engineering and Mathematics (STEM) education. It is argued that at high school level, STEM learning, which incorporates understandings of indigenization science and NOS, has the potential to provide a strong foundation for a culturally embedded scientific knowledge essential for their advancement in Science and Technology. Globally, investment in STEM education is recognized as essential for economic development. For this reason, developing countries such as Zimbabwe and South Africa have been investing into training specialized teachers in natural sciences and technology. However, in many cases this training has been detached from the cultural realities and contexts of indigenous learners. For this reason, the STEM curricula reform has provided implementation challenges to teachers. An issue of major concern is the teachers’ pedagogical content knowledge (PCK), which is essential for effective implementation of these STEM curricula. Well-developed Teacher PCK include an understanding of both the nature of indigenous knowledge (NOIK) and of NOS. This paper reports the results of a study that investigated the development of 3 South African and 3 Zimbabwean in-service teachers’ abilities to integrate NOS and NOIK as part of their PCK. A participatory action research design was utilized. The main focus was on capturing, determining and developing teachers STEM knowledge for integrating NOIK and NOS in science classrooms. Their use of indigenous games was used to determine how their subject knowledge for STEM and pedagogical abilities could be developed. Qualitative data were gathered through the use dialogues between the researchers and the in-service teachers, as well as interviewing the participating teachers. Analysis of the data provides a methodological window through which in-service teachers’ PCK can be STEMITIZED and their abilities to integrate NOS and NOIK developed. Implications are raised for developing teachers’ STEM education in universities and teacher training colleges.

Keywords: indigenous knowledge, nature of science, pedagogical content knowledge, STEM education

Procedia PDF Downloads 279

Authors: Luís F. da Silva, Ariadne C. Catto, Osmando F. Lopes, Khalifa Aguir, Valmor R. Mastelaro, Caue Ribeiro, Elson Longo

Abstract:

Gas detection is important for controlling industrial, and vehicle emissions, agricultural residues, and environmental control. In last decades, several semiconducting oxides have been used to detect dangerous or toxic gases. The excellent gas-sensing performance of these devices have been observed at high temperatures (~250 °C), which forbids the use for the detection of flammable and explosive gases. In this way, ultraviolet light activated gas sensors have been a simple and promising alternative to achieve room temperature sensitivity. Among the semiconductor oxides which exhibit a good performance as gas sensor, the zinc oxide (ZnO) and tin oxide (SnO2) have been highlighted. Nevertheless, their poor selectivity is the main disadvantage for application as gas sensor devices. Recently, heterostructures combining these two semiconductors (ZnO-SnO2) have been studied as an alternative way to enhance the gas sensor performance (sensitivity, selectivity, and stability). In this work, we investigated the influence of mass ratio Zn:Sn on the properties of ZnO-SnO2 nanocomposites prepared by hydrothermal treatment for 4 hours at 200 °C. The crystalline phase, surface, and morphological features were characterized by X-ray diffraction (XRD), high-resolution transmission electron (HR-TEM), and X-ray photoelectron spectroscopy (XPS) measurements. The gas sensor measurements were carried out at room-temperature under ultraviolet (UV) light irradiation using different ozone levels (0.06 to 0.61 ppm). The XRD measurements indicate the presence of ZnO and SnO2 crystalline phases, without the evidence of solid solution formation. HR-TEM analysis revealed that a good contact between the SnO2 nanoparticles and the ZnO nanorods, which are very important since interface characteristics between nanostructures are considered as challenge to development new and efficient heterostructures. Electrical measurements proved that the best ozone gas-sensing performance is obtained for ZnO:SnO2 (50:50) nanocomposite under UV light irradiation. Its sensitivity was around 6 times higher when compared to SnO2 pure, a traditional ozone gas sensor. These results demonstrate the potential of ZnO-SnO2 heterojunctions for the detection of ozone gas at room-temperature when irradiated with UV light irradiation.

Keywords: hydrothermal, zno-sno2, ozone sensor, uv-activation, room-temperature

Procedia PDF Downloads 284

Authors: Isomar Lima da Silva, Alcilene Batalha Pontes, Aristeu Jonatas Leite de Oliveira, Roberto Maia Augusto

Abstract:

Context: Mechanical ventilators are medical devices that help provide oxygen and ventilation to patients with respiratory difficulties. This equipment consists of a manual breathing unit that can be operated by a doctor or nurse and a mechanical ventilator that controls the airflow and pressure in the patient's respiratory system. This type of ventilator is commonly used in emergencies and intensive care units where it is necessary to provide breathing support to critically ill or injured patients. Objective: In this context, this work aims to develop a reliable and low-cost mechanical ventilator to meet the demand of hospitals in treating people affected by Covid-19 and other severe respiratory diseases, offering a chance of treatment as an alternative to mechanical ventilators currently available in the market. Method: The project presents the development of a low-cost auxiliary ventilator with a controlled ventilatory system assisted by integrated hardware and firmware for respiratory cycle control in non-invasive mechanical ventilation treatments using a manual artificial respiration unit. The hardware includes pressure sensors capable of identifying positive expiratory pressure, peak inspiratory flow, and injected air volume. The embedded system controls the data sent by the sensors. It ensures efficient patient breathing through the operation of the sensors, microcontroller, and actuator, providing patient data information to the healthcare professional (system operator) through the graphical interface and enabling clinical parameter adjustments as needed. Results: The test data of the developed mechanical ventilator presented satisfactory results in terms of performance and reliability, showing that the equipment developed can be a viable alternative to commercial mechanical ventilators currently available, offering a low-cost solution to meet the increasing demand for respiratory support equipment.

Keywords: mechanical fans, breathing, medical equipment, COVID-19, intensive care units

Procedia PDF Downloads 71

Authors: David A. Harness

Abstract:

Natural evolution of ATP cognitive systems is to meet AI peer review standards. ATP process of axiom selection from Mizar to prove a conjecture would be further refined, as in all human and machine learning, by solving the real world problem of the proposed AI peer review challenge: Determine which conjecture forms the higher confidence level constructive proof between Standard Model of Physics SU(n) lattice gauge group operation vs. present non-standard 4D GEM EOS SU(n) lattice gauge group spatially extended operation in which the photon and electron are the first two trace angular momentum invariants of a gravitoelectromagnetic (GEM) energy momentum density tensor wavetrain integration spin-stress pressure-volume equation of state (EOS), initiated via 32 lines of Mathematica code. Resulting gravitoelectromagnetic spectrum ranges from compressive through rarefactive of the central cosmological constant vacuum energy density in units of pascals. Said self-adjoint group operation exclusively operates on the stress energy momentum tensor of the Einstein field equations, introducing quantization directly on the 4D spacetime level, essentially reformulating the Yang-Mills virtual superpositioned particle compounded lattice gauge groups quantization of the vacuum—into a single hyper-complex multi-valued GEM U(1) × SU(1,3) lattice gauge group Planck spacetime mesh quantization of the vacuum. Thus the Mizar corpus already contains all of the axioms required for relevant DeepMath premise selection and unambiguous formal natural language parsing in context deep learning.

Keywords: automated theorem proving, constructive quantum field theory, information theory, neural networks

Procedia PDF Downloads 180

Authors: J. DeBoard, R. Dietrich, J. Hughes, K. Yurko, G. Harms

Abstract:

Alzheimer’s disease (AD) is a predominant type of dementia and is likely a major cause of neural network impairment. The pathogenesis of this neurodegenerative disorder has yet to be fully elucidated. There are currently no known cures for the disease, and the best hope is to be able to detect it early enough to impede its progress. Beyond age and genetics, another prevalent risk factor for AD might be traumatic brain injury (TBI), which has similar neurodegenerative hallmarks. Our research focuses on obtaining information and methods to be able to predict when neurodegenerative effects might occur at a clinical level by observation of events at a cellular and molecular level in model mice. First, we wish to introduce our evidence that brain damage can be observed via brain imaging prior to the noticeable loss of neuromuscular control in model mice of AD. We then show our evidence that some blood biomarkers might be able to be early predictors of AD in the same model mice. Thus, we were interested to see if we might be able to predict which mice might show long-term neurodegenerative effects due to differing degrees of TBI and what level of TBI causes further damage and earlier death to the AD model mice. Upon application of TBIs via an apparatus to effectively induce extremely mild to mild TBIs, wild-type (WT) mice and AD mouse models were tested for cognition, neuromuscular control, olfactory ability, blood biomarkers, and brain imaging. Experiments are currently still in process, and more results are therefore forthcoming. Preliminary data suggest that neuromotor control diminishes as well as olfactory function for both AD and WT mice after the administration of five consecutive mild TBIs. Also, seizure activity increases significantly for both AD and WT after the administration of the five TBI treatment. If future data supports these findings, important implications about the effect of TBI on those at risk for AD might be possible.

Keywords: Alzheimer's disease, blood biomarker, neurodegeneration, neuromuscular control, olfaction, traumatic brain injury

Procedia PDF Downloads 141

Authors: Imad Khan, Imran Shafi, Sarmad Farooq

Abstract:

Large diameter pipes, barrels, tubes, and ducts are used in a variety of applications covering civil and defense-related technologies. This may include heating/cooling networks, sign poles, bracing, casing, and artillery and tank gun barrels. These large diameter assemblies require regular inspection and cleaning to increase their life and reduce replacement costs. This paper describes the design, development, and testing results of an efficient yet simplified, low maintenance mechanical design controlled with minimal essential electronics using an electric motor for a non-technical staff. The proposed solution provides a simplified user interface and an automated cleaning mechanism that requires a single user to optimally clean pipes and barrels in the range of 105 mm to 203 mm caliber. The proposed system employs linear motion of specially designed brush along the barrel using a chain of specific strength and a pulley anchor attached to both ends of the barrel. A specially designed and manufactured gearbox is coupled with an AC motor to allow movement of contact brush with high torque to allow efficient cleaning. A suitably powered AC motor is fixed to the front adapter mounted on the muzzle side whereas the rear adapter has a pulley-based anchor mounted towards the breach block in case of a gun barrel. A mix of soft nylon and hard copper bristles-based large surface brush is connected through a strong steel chain to motor and anchor pulley. The system is equipped with limit switches to auto switch the direction when one end is reached on its operation. The testing results based on carefully established performance indicators indicate the superiority of the proposed user-friendly cleaning mechanism vis-à-vis its life cycle cost.

Keywords: pipe cleaning mechanism, limiting switch, pipe cleaning robot, large pipes

Procedia PDF Downloads 111

Authors: Merzak Laribi, Abdelmadjid Kasser

Abstract:

Thermal spraying processes are widely used to produce coatings on original constructions as well as in repair and maintenance of long standing structures. A lot of efforts forwarding to develop thermal spray coatings technology have been focused on improving mechanical characteristics, minimizing residual stress level and reducing porosity of the coatings. The specific aim of this paper is to determine either residual stresses distribution or wear resistance of stainless steel coating thermally sprayed on a carbon steel substrate. Internal stresses determination was performed using an extensometric method in combination with a simultaneous progressive electrolytic polishing. The procedure consists of measuring micro-deformations using a bi-directional extensometric gauges glued on the substrate side of the materials. Very thin layers of the deposits are removed by electrochemical polishing across the sample surface. Micro-deformations are instantaneously measured, leading to residual stresses calculation after each removal. Wear resistance of the coating has been determined using a ball-on-plate tribometer. Friction coefficient is instantaneously measured during the tribological test. Attention was particularly focused on the influence of a post-annealing at 850 °C for one hour in vacuum either on the residual stresses distribution or on the wear resistance behavior under specific wear and lubrication conditions. The obtained results showed that the microstructure of the obtained arc sprayed stainless steel coating is classical. It is homogeneous and contains un-melted particles, metallic oxides and also pores and micro-cracks. The internal stresses are in compression in the coating. They are more or less scattered between -50 and -270 MPa on the surface and decreased more at the interface. The value at the surface of the substrate is about –700 MPa, partially due to the molten particles impact with the substrate. The post annealing has reduced the residual stresses in both coating and surface of the steel substrate so that the hole material becomes more relaxed. Friction coefficient has an average value of 0.3 and 0.4 respectively for non annealed and annealed specimen. It is rather oil lubrication which is really benefit so that friction coefficient is decreased to about 0.06.

Keywords: residual stresses, wear resistance, stainless steel, coating, thermal spraying, annealing, lubrication

Procedia PDF Downloads 126

Authors: M. Bilal Ishfaq, Adnan N. Qureshi

Abstract:

COVID-19 is the most infectious disease these days, it was first reported in Wuhan, the capital city of Hubei in China then it spread rapidly throughout the whole world. Later on 11 March 2020, the World Health Organisation (WHO) declared it a pandemic. Since COVID-19 is highly contagious, it has affected approximately 219M people worldwide and caused 4.55M deaths. It has brought the importance of accurate diagnosis of respiratory diseases such as pneumonia and COVID-19 to the forefront. In this paper, we propose a hybrid approach for the automated detection of COVID-19 using medical imaging. We have presented the hybridization of manually extracted and convolutional features. Our approach combines Haralick texture features and convolutional features extracted from chest X-rays and CT scans. We also employ a minimum redundancy maximum relevance (MRMR) feature selection algorithm to reduce computational complexity and enhance classification performance. The proposed model is evaluated on four publicly available datasets, including Chest X-ray Pneumonia, COVID-19 Pneumonia, COVID-19 CTMaster, and VinBig data. The results demonstrate high accuracy and effectiveness, with 0.9925 on the Chest X-ray pneumonia dataset, 0.9895 on the COVID-19, Pneumonia and Normal Chest X-ray dataset, 0.9806 on the Covid CTMaster dataset, and 0.9398 on the VinBig dataset. We further evaluate the effectiveness of the proposed model using ROC curves, where the AUC for the best-performing model reaches 0.96. Our proposed model provides a promising tool for the early detection and accurate diagnosis of COVID-19, which can assist healthcare professionals in making informed treatment decisions and improving patient outcomes. The results of the proposed model are quite plausible and the system can be deployed in a clinical or research setting to assist in the diagnosis of COVID-19.

Keywords: COVID-19, feature engineering, artificial neural networks, radiology images

Procedia PDF Downloads 76

Authors: Animish Jain

Abstract:

This study deals with using Multiplicative Weight Update within artificial intelligence and machine learning to create models that can diagnose skin cancer using microscopic images of cancer samples. In this study, the multiplicative weight update method is used to take the predictions of multiple models to try and acquire more accurate results. Logistic Regression, Convolutional Neural Network (CNN), and Support Vector Machine Classifier (SVMC) models are employed within the Multiplicative Weight Update system. These models are trained on pictures of skin cancer from the ISIC-Archive, to look for patterns to label unseen scans as either benign or malignant. These models are utilized in a multiplicative weight update algorithm which takes into account the precision and accuracy of each model through each successive guess to apply weights to their guess. These guesses and weights are then analyzed together to try and obtain the correct predictions. The research hypothesis for this study stated that there would be a significant difference in the accuracy of the three models and the Multiplicative Weight Update system. The SVMC model had an accuracy of 77.88%. The CNN model had an accuracy of 85.30%. The Logistic Regression model had an accuracy of 79.09%. Using Multiplicative Weight Update, the algorithm received an accuracy of 72.27%. The final conclusion that was drawn was that there was a significant difference in the accuracy of the three models and the Multiplicative Weight Update system. The conclusion was made that using a CNN model would be the best option for this problem rather than a Multiplicative Weight Update system. This is due to the possibility that Multiplicative Weight Update is not effective in a binary setting where there are only two possible classifications. In a categorical setting with multiple classes and groupings, a Multiplicative Weight Update system might become more proficient as it takes into account the strengths of multiple different models to classify images into multiple categories rather than only two categories, as shown in this study. This experimentation and computer science project can help to create better algorithms and models for the future of artificial intelligence in the medical imaging field.

Keywords: artificial intelligence, machine learning, multiplicative weight update, skin cancer

Procedia PDF Downloads 80

Authors: Nicola Menga, Giuseppe Carbone

Abstract:

Although contact mechanics has been widely focused on the study of contacts between half-space, it has been recently pointed out that in presence of finite thickness elastic layers the results of the contact problem show significant difference in terms of the main contact quantities (e.g. contact area, penetration, mean pressure, etc.). Actually, there exist a wide range of industrial application demanding for this kind of studies, such as seals leakage prediction or pressure-sensitive coatings for electrical applications. In this work, we focus on the contact between a rigid profile and an elastic layer of thickness h confined under two different configurations: rigid constrain and applied uniform pressure. The elastic problem at hand has been formalized following Green’s function method and then numerically solved by means of a matrix inversion. We study different contact conditions, both considering and neglecting adhesive interactions at the interface. This leads to different solution techniques: Adhesive contacts equilibrium solution is found, in term of contact area for given penetration, making stationary the total free energy of the system; whereas, adhesiveless contacts are addressed defining an equilibrium criterion, again on the contact area, relying on the fracture mechanics stress intensity factor KI. In particular, we make the KI vanish at the edges of the contact area, as peculiar for adhesiveless elastic contacts. The results are obtained in terms of contact area, penetration, and mean pressure for both adhesive and adhesiveless contact conditions. As expected, in the case of a uniform applied pressure the slab turns out much more compliant than the rigidly constrained one. Indeed, we have observed that the peak value of the contact pressure, for both the adhesive and adhesiveless condition, is much higher for the rigidly constrained configuration than in the case of applied uniform pressure. Furthermore, we observed that, for little contact area, both systems behave the same and the pull-off occurs at approximately the same contact area and mean contact pressure. This is an expected result since in this condition the ratio between the layers thickness and the contact area is very high and both layer configurations recover the half-space behavior where the pull-off occurrence is mainly controlled by the adhesive interactions, which are kept constant among the cases.

Keywords: contact mechanics, adhesion, friction, thick layer

Procedia PDF Downloads 513

Authors: Susan Diamond

Abstract:

Machine Learning workloads have traditionally been run in high-performance computing (HPC) environments, where users log in to dedicated machines and utilize the attached GPUs to run training jobs on huge datasets. Training of large neural network models is very resource intensive, and even after exploiting parallelism and accelerators such as GPUs, a single training job can still take days. Consequently, the cost of hardware is a barrier to entry. Even when upfront cost is not a concern, the lead time to set up such an HPC environment takes months from acquiring hardware to set up the hardware with the right set of firmware, software installed and configured. Furthermore, scalability is hard to achieve in a rigid traditional lab environment. Therefore, it is slow to react to the dynamic change in the artificial intelligent industry. Watson Deep Learning as a service, a cloud-based deep learning platform that mitigates the long lead time and high upfront investment in hardware. It enables robust and scalable sharing of resources among the teams in an organization. It is designed for on-demand cloud environments. Providing a similar user experience in a multi-tenant cloud environment comes with its own unique challenges regarding fault tolerance, performance, and security. Watson Deep Learning as a service tackles these challenges and present a deep learning stack for the cloud environments in a secure, scalable and fault-tolerant manner. It supports a wide range of deep-learning frameworks such as Tensorflow, PyTorch, Caffe, Torch, Theano, and MXNet etc. These frameworks reduce the effort and skillset required to design, train, and use deep learning models. Deep Learning as a service is used at IBM by AI researchers in areas including machine translation, computer vision, and healthcare. 

Keywords: deep learning, machine learning, cognitive computing, model training

Procedia PDF Downloads 209