Search results for: time delay neural network

Authors: Zhijie Li, Wenhui Peng, Zelong Yuan, Jianchun Wang

Abstract:

Turbulence is a complex phenomenon that plays a crucial role in various fields, such as engineering, atmospheric science, and fluid dynamics. Predicting and understanding its behavior over long time scales have been challenging tasks. Traditional methods, such as large-eddy simulation (LES), have provided valuable insights but are computationally expensive. In the past few years, machine learning methods have experienced rapid development, leading to significant improvements in computational speed. However, ensuring stable and accurate long-term predictions remains a challenging task for these methods. In this study, we introduce the implicit U-net enhanced Fourier neural operator (IU-FNO) as a solution for stable and efficient long-term predictions of the nonlinear dynamics in three-dimensional (3D) turbulence. The IU-FNO model combines implicit re-current Fourier layers to deepen the network and incorporates the U-Net architecture to accurately capture small-scale flow structures. We evaluate the performance of the IU-FNO model through extensive large-eddy simulations of three types of 3D turbulence: forced homogeneous isotropic turbulence (HIT), temporally evolving turbulent mixing layer, and decaying homogeneous isotropic turbulence. The results demonstrate that the IU-FNO model outperforms other FNO-based models, including vanilla FNO, implicit FNO (IFNO), and U-net enhanced FNO (U-FNO), as well as the dynamic Smagorinsky model (DSM), in predicting various turbulence statistics. Specifically, the IU-FNO model exhibits improved accuracy in predicting the velocity spectrum, probability density functions (PDFs) of vorticity and velocity increments, and instantaneous spatial structures of the flow field. Furthermore, the IU-FNO model addresses the stability issues encountered in long-term predictions, which were limitations of previous FNO models. In addition to its superior performance, the IU-FNO model offers faster computational speed compared to traditional large-eddy simulations using the DSM model. It also demonstrates generalization capabilities to higher Taylor-Reynolds numbers and unseen flow regimes, such as decaying turbulence. Overall, the IU-FNO model presents a promising approach for long-term dynamics prediction in 3D turbulence, providing improved accuracy, stability, and computational efficiency compared to existing methods.

Keywords: data-driven, Fourier neural operator, large eddy simulation, fluid dynamics

Procedia PDF Downloads 74

Authors: Mr. P. Suryaprasad, R. Lalitha

Abstract:

This paper presents an efficient area and delay architectures for the implementation of one dimensional and two dimensional discrete cosine transform (DCT). These are supported to different lengths (4, 8, 16, and 32). DCT blocks are used in the different video coding standards for the image compression. The 2D- DCT calculation is made using the 2D-DCT separability property, such that the whole architecture is divided into two 1D-DCT calculations by using a transpose buffer. Based on the existing 1D-DCT architecture two different types of 2D-DCT architectures, folded and parallel types are implemented. Both of these two structures use the same transpose buffer. Proposed transpose buffer occupies less area and high speed than existing transpose buffer. Hence the area, low power and delay of both the 2D-DCT architectures are reduced.

Keywords: transposition buffer, video compression, discrete cosine transform, high efficiency video coding, two dimensional picture

Procedia PDF Downloads 521

Authors: Semachew M. Kassa, Africa M Geremew, Tezera F. Azmatch, Nandyala Darga Kumar

Abstract:

Frequency ratio (FR) and analytical hierarchy process (AHP) methods are developed based on past landslide failure points to identify the landslide susceptibility mapping because landslides can seriously harm both the environment and society. However, it is still difficult to select the most efficient method and correctly identify the main driving factors for particular regions. In this study, we used fourteen landslide conditioning factors (LCFs) and five soft computing algorithms, including Random Forest (RF), Support Vector Machine (SVM), Logistic Regression (LR), Artificial Neural Network (ANN), and Naïve Bayes (NB), to predict the landslide susceptibility at 12.5 m spatial scale. The performance of the RF (F1-score: 0.88, AUC: 0.94), ANN (F1-score: 0.85, AUC: 0.92), and SVM (F1-score: 0.82, AUC: 0.86) methods was significantly better than the LR (F1-score: 0.75, AUC: 0.76) and NB (F1-score: 0.73, AUC: 0.75) method, according to the classification results based on inventory landslide points. The findings also showed that around 35% of the study region was made up of places with high and very high landslide risk (susceptibility greater than 0.5). The very high-risk locations were primarily found in the western and southeastern regions, and all five models showed good agreement and similar geographic distribution patterns in landslide susceptibility. The towns with the highest landslide risk include Amhara Saint Town's western part, the Northern part, and St. Gebreal Church villages, with mean susceptibility values greater than 0.5. However, rainfall, distance to road, and slope were typically among the top leading factors for most villages. The primary contributing factors to landslide vulnerability were slightly varied for the five models. Decision-makers and policy planners can use the information from our study to make informed decisions and establish policies. It also suggests that various places should take different safeguards to reduce or prevent serious damage from landslide events.

Keywords: artificial neural network, logistic regression, landslide susceptibility, naïve Bayes, random forest, support vector machine

Procedia PDF Downloads 82

Authors: Keisuke Fujioka, Yuta Itoh, Masaru Minagawa, Shunji Kusayanagi

Abstract:

In this study, building information models of the underground temporary structures and adjacent embedded pipes were constructed to show the importance of the information on underground pipes adjacent to the structures to enhance the productivity of execution of construction. Next, the bar chart used in actual construction process were employed to make the Gantt chart, and the critical pass analysis was carried out to show that accurate information on the arrangement of underground existing pipes can be used for the enhancement of the productivity of the construction of underground structures. In the analyzed project, significant construction delay was not caused by unforeseeable existence of underground pipes by the management ability of the construction manager. However, in many cases of construction executions in the developing countries, the existence of unforeseeable embedded pipes often causes substantial delay of construction. Design change based on uncertainty on the position information of embedded pipe can be also important risk for contractors in domestic construction. So CPM analyses were performed by a project-management-software to the situation that influence of the tasks causing construction delay was assumed more significant. Through the analyses, the efficiency of information management on underground pipes and BIM analysis in the design stage for workability improvement was indirectly confirmed.

Keywords: building-information modelling, construction information modelling, design improvement, infrastructure

Procedia PDF Downloads 308

Authors: Pierre Tize Mha, Mohammad Jahazi, Amèvi Togne, Olivier Pantalé

Abstract:

Large-size forged blocks made of medium carbon high-strength steels are extensively used in the automotive industry as dies for the production of bumpers and dashboards through the plastic injection process. The manufacturing process of the large blocks starts with ingot casting, followed by open die forging and a quench and temper heat treatment process to achieve the desired mechanical properties and numerical simulation is widely used nowadays to predict these properties before the experiment. But the temperature gradient inside the specimen remains challenging in the sense that the temperature before loading inside the material is not the same, but during the simulation, constant temperature is used to simulate the experiment because it is assumed that temperature is homogenized after some holding time. Therefore to be close to the experiment, real distribution of the temperature through the specimen is needed before the mechanical loading. Thus, We present here a robust algorithm that allows the calculation of the temperature gradient within the specimen, thus representing a real temperature distribution within the specimen before deformation. Indeed, most numerical simulations consider a uniform temperature gradient which is not really the case because the surface and core temperatures of the specimen are not identical. Another feature that influences the mechanical properties of the specimen is recrystallization which strongly depends on the deformation conditions and the type of deformation like Upsetting, Cogging...etc. Indeed, Upsetting and Cogging are the stages where the greatest deformations are observed, and a lot of microstructural phenomena can be observed, like recrystallization, which requires in-depth characterization. Complete dynamic recrystallization plays an important role in the final grain size during the process and therefore helps to increase the mechanical properties of the final product. Thus, the identification of the conditions for the initiation of dynamic recrystallization is still relevant. Also, the temperature distribution within the sample and strain rate influence the recrystallization initiation. So the development of a technique allowing to predict the initiation of this recrystallization remains challenging. In this perspective, we propose here, in addition to the algorithm allowing to get the temperature distribution before the loading stage, an analytical model leading to determine the initiation of this recrystallization. These two techniques are implemented into the Abaqus finite element software via the UAMP and VUHARD subroutines for comparison with a simulation where an isothermal temperature is imposed. The Artificial Neural Network (ANN) model to describe the plastic behavior of the material is also implemented via the VUHARD subroutine. From the simulation, the temperature distribution inside the material and recrystallization initiation is properly predicted and compared to the literature models.

Keywords: dynamic recrystallization, finite element modeling, artificial neural network, numerical implementation

Procedia PDF Downloads 80

Authors: Kavyadevi N., Kaviya G., Gowsalya P., Janani M., Mohanraj S.

Abstract:

Hevea brasiliensis, also known as the rubber tree, is one of the foremost assets of crops in the world. One of the most significant advantages of the Rubber Plant in terms of air oxygenation is its capacity to reduce the likelihood of an individual developing respiratory allergies like asthma. To construct such a system that can properly identify crop diseases and pests and then create a database of insecticides for each pest and disease, we must first give treatment for the illness that has been detected. We shall primarily examine three major leaf diseases since they are economically deficient in this article, which is Bird's eye spot, algal spot and powdery mildew. And the recommended work focuses on disease identification on rubber tree leaves. It will be accomplished by employing one of the superior algorithms. Input, Preprocessing, Image Segmentation, Extraction Feature, and Classification will be followed by the processing technique. We will use time-consuming procedures that they use to detect the sickness. As a consequence, the main ailments, underlying causes, and signs and symptoms of diseases that harm the rubber tree are covered in this study.

Keywords: image processing, python, convolution neural network (CNN), machine learning

Procedia PDF Downloads 76

Authors: Jessica Lopez Espejel, Mahaman Sanoussi Yahaya Alassan, Walid Dahhane, El Hassane Ettifouri

Abstract:

Pretrained transformer-based models have shown high performance in natural language generation tasks. However, a new wave of interest has surged: automatic programming language code generation. This task consists of translating natural language instructions to a source code. Despite the fact that well-known pre-trained models on language generation have achieved good performance in learning programming languages, effort is still needed in automatic code generation. In this paper, we introduce JaCoText, a model based on Transformer neural network. It aims to generate java source code from natural language text. JaCoText leverages the advantages of both natural language and code generation models. More specifically, we study some findings from state of the art and use them to (1) initialize our model from powerful pre-trained models, (2) explore additional pretraining on our java dataset, (3) lead experiments combining the unimodal and bimodal data in training, and (4) scale the input and output length during the fine-tuning of the model. Conducted experiments on CONCODE dataset show that JaCoText achieves new state-of-the-art results.

Keywords: java code generation, natural language processing, sequence-to-sequence models, transformer neural networks

Procedia PDF Downloads 284

Authors: Ogwueleka Francisca Nonyelum, Jiya Muhammad

Abstract:

Poor communication between the victims of the burglaries, road and fire accidents and the agencies, and lack of quick emergency response by the agencies is solved through Machine-to-Machine (M2M) communication. A distress caller is expected to make a call through a network to the respective agency for emergency response but due to some challenges, this often becomes arduous and futile. This research puts forth an Internet-based architecture for Machine-to-Machine (M2M) communication to enhance information dissemination in National Public Security Communication System (NPSCS) network. M2M enables the flow of data between machines and machines and ultimately machines and people with information flowing from a machine over a network, and then through a gateway to a system where it is reviewed and acted on. The research findings showed that Internet-based architecture for M2M communication is most suitable for deployment of a public security network which will allow machines to use Internet to talk to each other.

Keywords: machine-to-machine (M2M), internet-based architecture, network, gateway

Procedia PDF Downloads 484

Authors: Muslim Malik, Avadhesh Kumar

Abstract:

A stronger concept of exact controllability which is called Total Controllability is introduced in this manuscript. Sufficient conditions have been established for the total controllability of a control problem, governed by second order nonlinear differential equation with delay and non-instantaneous impulses in a Banach space X. The results are obtained using the strongly continuous cosine family and Banach fixed point theorem. Also, the total controllability of an integrodifferential problem is investigated. At the end, some numerical examples are provided to illustrate the analytical findings.

Keywords: Banach fixed point theorem, non-instantaneous impulses, strongly continuous cosine family, total controllability

Procedia PDF Downloads 298

Authors: Z. Gharaee

Abstract:

Recognizing actions performed by others is important in our daily lives since it is necessary for communicating with others in a proper way. We perceive an action by observing the kinematics of motions involved in the performance. We use our experience and concepts to make a correct recognition of the actions. Although building the action concepts is a life-long process, which is repeated throughout life, we are very efficient in applying our learned concepts in analyzing motions and recognizing actions. Experiments on the subjects observing the actions performed by an actor show that an action is recognized after only about two hundred milliseconds of observation. In this study, hierarchical action recognition architecture is proposed by using growing grid layers. The first-layer growing grid receives the pre-processed data of consecutive 3D postures of joint positions and applies some heuristics during the growth phase to allocate areas of the map by inserting new neurons. As a result of training the first-layer growing grid, action pattern vectors are generated by connecting the elicited activations of the learned map. The ordered vector representation layer receives action pattern vectors to create time-invariant vectors of key elicited activations. Time-invariant vectors are sent to second-layer growing grid for categorization. This grid creates the clusters representing the actions. Finally, one-layer neural network developed by a delta rule labels the action categories in the last layer. System performance has been evaluated in an experiment with the publicly available MSR-Action3D dataset. There are actions performed by using different parts of human body: Hand Clap, Two Hands Wave, Side Boxing, Bend, Forward Kick, Side Kick, Jogging, Tennis Serve, Golf Swing, Pick Up and Throw. The growing grid architecture was trained by applying several random selections of generalization test data fed to the system during on average 100 epochs for each training of the first-layer growing grid and around 75 epochs for each training of the second-layer growing grid. The average generalization test accuracy is 92.6%. A comparison analysis between the performance of growing grid architecture and self-organizing map (SOM) architecture in terms of accuracy and learning speed show that the growing grid architecture is superior to the SOM architecture in action recognition task. The SOM architecture completes learning the same dataset of actions in around 150 epochs for each training of the first-layer SOM while it takes 1200 epochs for each training of the second-layer SOM and it achieves the average recognition accuracy of 90% for generalization test data. In summary, using the growing grid network preserves the fundamental features of SOMs, such as topographic organization of neurons, lateral interactions, the abilities of unsupervised learning and representing high dimensional input space in the lower dimensional maps. The architecture also benefits from an automatic size setting mechanism resulting in higher flexibility and robustness. Moreover, by utilizing growing grids the system automatically obtains a prior knowledge of input space during the growth phase and applies this information to expand the map by inserting new neurons wherever there is high representational demand.

Keywords: action recognition, growing grid, hierarchical architecture, neural networks, system performance

Procedia PDF Downloads 157

Authors: Precious Ehiomogue, Ifechukwude Israel Ahuchaogu, Isiguzo Edwin Ahaneku

Abstract:

Vermicompost is the product of the decomposition process using various species of worms, to create a mixture of decomposing vegetable or food waste, bedding materials, and vemicast. This process is called vermicomposting, while the rearing of worms for this purpose is called vermiculture. Several works have verified the adsorption of toxic metals using vermicompost but the application is still scarce for the retention of organic compounds. This research brings to knowledge the effectiveness of earthworm waste (vermicompost) for the remediation of crude oil contaminated soils. The remediation methods adopted in this study were two soil washing methods namely, batch and column process which represent laboratory and in-situ remediation. Characterization of the vermicompost and crude oil contaminated soil were performed before and after the soil washing using Fourier transform infrared (FTIR), scanning electron microscopy (SEM), X-ray fluorescence (XRF), X-ray diffraction (XRD) and Atomic adsorption spectrometry (AAS). The optimization of washing parameters, using response surface methodology (RSM) based on Box-Behnken Design was performed on the response from the laboratory experimental results. This study also investigated the application of machine learning models [Artificial neural network (ANN), Adaptive neuro fuzzy inference system (ANFIS). ANN and ANFIS were evaluated using the coefficient of determination (R²) and mean square error (MSE)]. Removal efficiency obtained from the Box-Behnken design experiment ranged from 29% to 98.9% for batch process remediation. Optimization of the experimental factors carried out using numerical optimization techniques by applying desirability function method of the response surface methodology (RSM) produce the highest removal efficiency of 98.9% at absorbent dosage of 34.53 grams, adsorbate concentration of 69.11 (g/ml), contact time of 25.96 (min), and pH value of 7.71, respectively. Removal efficiency obtained from the multilevel general factorial design experiment ranged from 56% to 92% for column process remediation. The coefficient of determination (R²) for ANN was (0.9974) and (0.9852) for batch and column process, respectively, showing the agreement between experimental and predicted results. For batch and column precess, respectively, the coefficient of determination (R²) for RSM was (0.9712) and (0.9614), which also demonstrates agreement between experimental and projected findings. For the batch and column processes, the ANFIS coefficient of determination was (0.7115) and (0.9978), respectively. It can be concluded that machine learning models can predict the removal of crude oil from polluted soil using vermicompost. Therefore, it is recommended to use machines learning models to predict the removal of crude oil from contaminated soil using vermicompost.

Keywords: ANFIS, ANN, crude-oil, contaminated soil, remediation and vermicompost

Procedia PDF Downloads 111

Authors: Faribaebrahimi, Mehdi Ghorbani, Mohsen Mohsenisaravi

Abstract:

Ecosystem management is complex because of natural and human issues. To cope with this complexity water governance is recommended since it involves all stakeholders including people, governmental and non-governmental organization who related to environmental systems. Water governance emphasizes on water co-management through consideration of all the stakeholders in the form of social and organizational network. In this research, to illustrate indicators of water governance in Dorood watershed, in Shemiranat region of Iran, social network analysis had been applied. The results revealed that social cohesion among pastoralists in Dorood is medium because of trust links, while link sustainability is weak to medium. According to the results, some pastoralists have high social power and therefore are key actors in the utilization network, regarding to centrality index and trust links. The results also demonstrated that Agricultural Development Office and (Shemshak-Darbandsar Islamic) Council are key actors in rangeland co-management, based on centrality index in rangeland institutional network at regional scale in Shemiranat district.

Keywords: social network analysis, water governance, organizational network, water co-management

Procedia PDF Downloads 351

Authors: Adewusi Adeoluwawale, Oronti Iyabosola Busola, Akinola Iretiayo, Komolafe Olusola Aderibigbe

Abstract:

The complex and radial structure of the low voltage distribution network (415V) makes it vulnerable to faults which are due to system and the environmental related factors. Besides these, the protective scheme employed on the low voltage network which is the fuse cannot be monitored remotely such that in the event of sustained fault, the utility will have to rely solely on the complaint brought by customers for loss of supply and this tends to increase the length of outages. A microcontroller based fault detection scheme is hereby developed to detect low voltage and high voltage fault conditions which are common faults on this network. Voltages below 198V and above 242V on the distribution feeders are classified and detected as low voltage and high voltages respectively. Results shows that the developed scheme produced a good response time in the detection of these faults.

Keywords: fault detection, low voltage distribution feeders, outage times, sustained faults

Procedia PDF Downloads 543

Authors: O. Mohamed Amine, A. Khireddine

Abstract:

Telecoms operators are always looking to meet their share of the other customers, they try to gain optimum utilization of the deployed equipment and network optimization has become essential. This project consists of optimizing UMTS network, and the study area is an urban area situated in the center of Algiers. It was initially questions to become familiar with the different communication systems (3G) and the optimization technique, its main components, and its fundamental characteristics radios were introduced.

Keywords: UMTS, UTRAN, WCDMA, optimization

Procedia PDF Downloads 383

Authors: R. Sudhesh Solomon, P. Y. K. L. Srinivas, Abhay Narayan, Amitava Das

Abstract:

Users in social networks either unicast or broadcast their messages. At mention is the popular way of unicasting for Twitter whereas general tweeting could be considered as broadcasting method. Understanding the information flow and dynamics within a Social Network and modeling the same is a promising and an open research area called Information Diffusion. This paper seeks an answer to a fundamental question - understanding if the at-mention network or the unicasting pattern in social media is purely random in nature or is there any user specific selectional preference? To answer the question we present an empirical analysis to understand the sociological aspects of Twitter mentions network within a social network community. To understand the sociological behavior we analyze the values (Schwartz model: Achievement, Benevolence, Conformity, Hedonism, Power, Security, Self-Direction, Stimulation, Traditional and Universalism) of all the users. Empirical results suggest that values traits are indeed salient cue to understand how the mention-based communication network functions. For example, we notice that individuals possessing similar values unicast among themselves more often than with other value type people. We also observe that traditional and self-directed people do not maintain very close relationship in the network with the people of different values traits.

Keywords: information diffusion, personality and values, social network analysis, twitter mentions network

Procedia PDF Downloads 374

Authors: Patrick J. Kerpan, Ryan C. Koop, Margaret M. Walker, Chris P. Swan

Abstract:

The User Datagram Protocol (UDP) multicast is a vital part of data center networking that is being left out of major cloud computing providers' network infrastructure. Enterprise users rely on multicast, and particularly UDP multicast to create and connect vital business operations. For example, UPD makes a variety of business functions possible from simultaneous content media updates, High-Performance Computing (HPC) grids, and video call routing for massive open online courses (MOOCs). Essentially, UDP multicast's technological slight is causing a huge effect on whether companies choose to use (or not to use) public cloud infrastructure as a service (IaaS). Allowing the ‘chatty’ UDP multicast protocol inside a cloud network could have a serious impact on the performance of the cloud as a whole. Cloud IaaS providers solve the issue by disallowing all UDP multicast. But what about enterprise use cases for multicast applications in organizations that want to move to the cloud? To re-allow multicast traffic, enterprises can build a layer 3 - 7 network over the top of a data center, private cloud, or public cloud. An overlay network simply creates a private, sealed network on top of the existing network. Overlays give complete control of the network back to enterprise cloud users the freedom to manage their network beyond the control of the cloud provider’s firewall conditions. The same logic applies if for users who wish to use IPsec or BGP network protocols inside or connected into an overlay network in cloud IaaS.

Keywords: cloud computing, protocols, UDP multicast, virtualization

Procedia PDF Downloads 590

Authors: Tolulope Aremu

Abstract:

This paper is based on the idea of using deep learning methodology for optimizing production yield by tuning a few key process parameters in a manufacturing environment. The study was explicitly on how to maximize production yield and minimize operational costs by utilizing advanced neural network models, specifically Long Short-Term Memory and Convolutional Neural Networks. These models were implemented using Python-based frameworks—TensorFlow and Keras. The targets of the research are the precision molding processes in which temperature ranges between 150°C and 220°C, the pressure ranges between 5 and 15 bar, and the material flow rate ranges between 10 and 50 kg/h, which are critical parameters that have a great effect on yield. A dataset of 1 million production cycles has been considered for five continuous years, where detailed logs are present showing the exact setting of parameters and yield output. The LSTM model would model time-dependent trends in production data, while CNN analyzed the spatial correlations between parameters. Models are designed in a supervised learning manner. For the model's loss, an MSE loss function is used, optimized through the Adam optimizer. After running a total of 100 training epochs, 95% accuracy was achieved by the models recommending optimal parameter configurations. Results indicated that with the use of RSM and DOE traditional methods, there was an increase in production yield of 12%. Besides, the error margin was reduced by 8%, hence consistent quality products from the deep learning models. The monetary value was annually around $2.5 million, the cost saved from material waste, energy consumption, and equipment wear resulting from the implementation of optimized process parameters. This system was deployed in an industrial production environment with the help of a hybrid cloud system: Microsoft Azure, for data storage, and the training and deployment of their models were performed on Google Cloud AI. The functionality of real-time monitoring of the process and automatic tuning of parameters depends on cloud infrastructure. To put it into perspective, deep learning models, especially those employing LSTM and CNN, optimize the production yield by fine-tuning process parameters. Future research will consider reinforcement learning with a view to achieving further enhancement of system autonomy and scalability across various manufacturing sectors.

Keywords: production yield optimization, deep learning, tuning of process parameters, LSTM, CNN, precision manufacturing, TensorFlow, Keras, cloud infrastructure, cost saving

Procedia PDF Downloads 31

Authors: Anoop T. R., Otman Basir, Robert F. Hess, Eileen E. Birch, Brooke A. Koritala, Reed M. Jost, Becky Luu, David Stager, Ben Thompson

Abstract:

Strabismus refers to a misalignment of the eyes. Early detection and treatment of strabismus in childhood can prevent the development of permanent vision loss due to abnormal development of visual brain areas. We developed a two-stage method for strabismus detection and classification based on photographs of the face. The first stage detects the presence or absence of strabismus, and the second stage classifies the type of strabismus. The first stage comprises face detection using Haar cascade, facial landmark estimation, face alignment, aligned face landmark detection, segmentation of the eye region, and detection of strabismus using VGG 16 convolution neural networks. Face alignment transforms the face to a canonical pose to ensure consistency in subsequent analysis. Using facial landmarks, the eye region is segmented from the aligned face and fed into a VGG 16 CNN model, which has been trained to classify strabismus. The CNN determines whether strabismus is present and classifies the type of strabismus (exotropia, esotropia, and vertical deviation). If stage 1 detects strabismus, the eye region image is fed into stage 2, which starts with the estimation of pupil center coordinates using mask R-CNN deep neural networks. Then, the distance between the pupil coordinates and eye landmarks is calculated along with the angle that the pupil coordinates make with the horizontal and vertical axis. The distance and angle information is used to characterize the degree and direction of the strabismic eye misalignment. This model was tested on 100 clinically labeled images of children with (n = 50) and without (n = 50) strabismus. The True Positive Rate (TPR) and False Positive Rate (FPR) of the first stage were 94% and 6% respectively. The classification stage has produced a TPR of 94.73%, 94.44%, and 100% for esotropia, exotropia, and vertical deviations, respectively. This method also had an FPR of 5.26%, 5.55%, and 0% for esotropia, exotropia, and vertical deviation, respectively. The addition of one more feature related to the location of corneal light reflections may reduce the FPR, which was primarily due to children with pseudo-strabismus (the appearance of strabismus due to a wide nasal bridge or skin folds on the nasal side of the eyes).

Keywords: strabismus, deep neural networks, face detection, facial landmarks, face alignment, segmentation, VGG 16, mask R-CNN, pupil coordinates, angle deviation, horizontal and vertical deviation

Procedia PDF Downloads 93

Authors: Khalid Ali, Manar Jammal

Abstract:

In order to meet the stringent latency and reliability requirements of the upcoming 5G networks, Open Radio Access Networks (O-RAN) have been proposed. The virtualization of O-RAN has allowed it to be treated as a Network Function Virtualization (NFV) architecture, while its components are considered Virtualized Network Functions (VNFs). Hence, intelligent Machine Learning (ML) based solutions can be utilized to apply different resource management and allocation techniques on O-RAN. However, intelligently allocating resources for O-RAN VNFs can prove challenging due to the dynamicity of traffic in mobile networks. Network providers need to dynamically scale the allocated resources in response to the incoming traffic. Elastically allocating resources can provide a higher level of flexibility in the network in addition to reducing the OPerational EXpenditure (OPEX) and increasing the resources utilization. Most of the existing elastic solutions are reactive in nature, despite the fact that proactive approaches are more agile since they scale instances ahead of time by predicting the incoming traffic. In this work, we propose and evaluate traffic forecasting models based on the ML algorithm. The algorithms aim at predicting future O-RAN traffic by using previous traffic data. Detailed analysis of the traffic data was carried out to validate the quality and applicability of the traffic dataset. Hence, two ML models were proposed and evaluated based on their prediction capabilities.

Keywords: O-RAN, traffic forecasting, NFV, ARIMA, LSTM, elasticity

Procedia PDF Downloads 226

Authors: Ghazwan Al-Haji

Abstract:

Most accidents occur in urban areas, and the most related casualties are vulnerable road users (pedestrians and cyclists). The traffic calming measures (TCMs) are widely used and considered to be successful in reducing speed and traffic volume. However, TCMs create unwanted effects include: noise, emissions, energy consumption, vehicle delays and emergency response time (ERT). Different vertical and horizontal TCMs have been already applied nationally (Sweden) and internationally with different impacts. It is a big challenge among traffic engineers, planners, and policy-makers to choose and priorities the best TCMs to be implemented. This study will assess the existing guidelines for TCMs in relation to safety and ERT with focus on data from Norrkoping city in Sweden. The expected results will save lives, time, and money on particularly Swedish Roads. The study will also review newly technologies and how they can improve safety and reduce ERT.

Keywords: traffic calming measures, traffic safety, delay time, vulnerable road users

Procedia PDF Downloads 140

Authors: Jade Martínez-Llinàs, Pere Colet

Abstract:

By exploring the possible dynamical regimes in a prototypical model for mutually delay-coupled OEOs, here it is shown that two mutually coupled non-identical OEOs, besides in- and out-of-phase square-waves, can generate stable square-wave pulses synchronized at a quarter of the period (T/4) in a broad parameter region. The key point to obtain T/4 solutions is that the two OEO operate with mixed feedback, namely with negative feedback in one and positive in the other. Furthermore, the coexistence of multiple solutions provides a large degree of flexibility for tuning the frequency in the GHz range without changing any parameter. As a result the two coupled OEOs system is good candidate to be implemented for information encoding as a high-capacity memory device.

Keywords: nonlinear optics, optoelectronic oscillators, square waves, synchronization

Procedia PDF Downloads 370

Authors: Sunil Chandra, Himanshu Rawat, Vikas Gusain, Triparna Barman

Abstract:

Forest fires are a recurrent phenomenon in the Himalayan region owing to the presence of vulnerable forest types, topographical gradients, climatic weather conditions, and anthropogenic pressure. The present study focuses on the identification of forest fire-affected areas in a small part of the West Himalayan region using a differential normalized burnt ratio method and spectral unmixing methods. The study area has a rugged terrain with the presence of sub-tropical pine forest, montane temperate forest, and sub-alpine forest and scrub. The major reason for fires in this region is anthropogenic in nature, with the practice of human-induced fires for getting fresh leaves, scaring wild animals to protect agricultural crops, grazing practices within reserved forests, and igniting fires for cooking and other reasons. The fires caused by the above reasons affect a large area on the ground, necessitating its precise estimation for further management and policy making. In the present study, two approaches have been used for carrying out a burnt area analysis. The first approach followed for burnt area analysis uses a differenced normalized burnt ratio (dNBR) index approach that uses the burnt ratio values generated using the Short-Wave Infrared (SWIR) band and Near Infrared (NIR) bands of the Sentinel-2 image. The results of the dNBR have been compared with the outputs of the spectral mixing methods. It has been found that the dNBR is able to create good results in fire-affected areas having homogenous forest stratum and with slope degree <5 degrees. However, in a rugged terrain where the landscape is largely influenced by the topographical variations, vegetation types, tree density, the results may be largely influenced by the effects of topography, complexity in tree composition, fuel load composition, and soil moisture. Hence, such variations in the factors influencing burnt area assessment may not be effectively carried out using a dNBR approach which is commonly followed for burnt area assessment over a large area. Hence, another approach that has been attempted in the present study utilizes a spectral mixing method where the individual pixel is tested before assigning an information class to it. The method uses a neural network approach utilizing Sentinel-2 bands. The training and testing data are generated from the Sentinel-2 data and the national field inventory, which is further used for generating outputs using ML tools. The analysis of the results indicates that the fire-affected regions and their severity can be better estimated using spectral unmixing methods, which have the capability to resolve the noise in the data and can classify the individual pixel to the precise burnt/unburnt class.

Keywords: categorical data, log linear modeling, neural network, shifting cultivation

Procedia PDF Downloads 54

Authors: Longfei Wang, Selçuk Köse

Abstract:

Hardware Trojan becomes great concern as integrated circuit (IC) technology advances and not all manufacturing steps of an IC are accomplished within one company. Real-time hardware Trojan detection is proven to be a feasible way to detect randomly activated Trojans that cannot be detected at testing stage. On-chip sensors serve as a great candidate to implement real-time hardware Trojan detection, however, the optimization of on-chip sensors has not been thoroughly investigated and the location of Trojan has not been carefully explored. On-chip sensor ellipse distribution method and equivalent mapping technique are proposed based on the characteristics of on-chip power delivery network in this paper to address the optimization and distribution of on-chip sensors for real-time hardware Trojan detection as well as to estimate the location and current consumption of hardware Trojan. Simulation results verify that hardware Trojan activation can be effectively detected and the location of a hardware Trojan can be efficiently estimated with less than 5% error for a realistic power grid using our proposed methods. The proposed techniques therefore lay a solid foundation for isolation and even deactivation of hardware Trojans through accurate location of Trojans.

Keywords: hardware trojan, on-chip sensor, power distribution network, power/ground noise

Procedia PDF Downloads 391

Authors: Farina Riaz, Shahab Abdulla, Srinjoy Ganguly, Hajime Suzuki, Ravinesh C. Deo, Susan Hopkins

Abstract:

Recently, quantum machine learning has received significant attention. For various types of data, including text and images, numerous quantum machine learning (QML) models have been created and are being tested. Images are exceedingly complex data components that demand more processing power. Despite being mature, classical machine learning still has difficulties with big data applications. Furthermore, quantum technology has revolutionized how machine learning is thought of, by employing quantum features to address optimization issues. Since quantum hardware is currently extremely noisy, it is not practicable to run machine learning algorithms on it without risking the production of inaccurate results. To discover the advantages of quantum versus classical approaches, this research has concentrated on colored image data. Deep learning classification models are currently being created on Quantum platforms, but they are still in a very early stage. Black and white benchmark image datasets like MNIST and Fashion MINIST have been used in recent research. MNIST and CIFAR-10 were compared for binary classification, but the comparison showed that MNIST performed more accurately than colored CIFAR-10. This research will evaluate the performance of the QML algorithm on the colored benchmark dataset CIFAR-10 to advance QML's real-time applicability. However, deep learning classification models have not been developed to compare colored images like Quantum Convolutional Neural Network (QCNN) to determine how much it is better to classical. Only a few models, such as quantum variational circuits, take colored images. The methodology adopted in this research is a hybrid approach by using penny lane as a simulator. To process the 10 classes of CIFAR-10, the image data has been translated into grey scale and the 28 × 28-pixel image containing 10,000 test and 50,000 training images were used. The objective of this work is to determine how much the quantum approach can outperform a classical approach for a comprehensive dataset of color images. After pre-processing 50,000 images from a classical computer, the QCNN model adopted a hybrid method and encoded the images into a quantum simulator for feature extraction using quantum gate rotations. The measurements were carried out on the classical computer after the rotations were applied. According to the results, we note that the QCNN approach is ~12% more effective than the traditional classical CNN approaches and it is possible that applying data augmentation may increase the accuracy. This study has demonstrated that quantum machine and deep learning models can be relatively superior to the classical machine learning approaches in terms of their processing speed and accuracy when used to perform classification on colored classes.

Keywords: CIFAR-10, quantum convolutional neural networks, quantum deep learning, quantum machine learning

Procedia PDF Downloads 129

Authors: Ma Lei-Lei, Zhou You

Abstract:

Aiming at the problems of low detection accuracy of deep learning algorithm in traffic sign detection, this paper proposes improved EfficientDet based traffic sign detection algorithm. Multi-head self-attention is introduced in the minimum resolution layer of the backbone of EfficientDet to achieve effective aggregation of local and global depth information, and this study proposes an improved feature fusion pyramid with increased vertical cross-layer connections, which improves the performance of the model while introducing a small amount of complexity, the Balanced L1 Loss is introduced to replace the original regression loss function Smooth L1 Loss, which solves the problem of balance in the loss function. Experimental results show, the algorithm proposed in this study is suitable for the task of traffic sign detection. Compared with other models, the improved EfficientDet has the best detection accuracy. Although the test speed is not completely dominant, it still meets the real-time requirement.

Keywords: convolutional neural network, transformer, feature pyramid networks, loss function

Procedia PDF Downloads 97

Authors: Elaheh Vaezpour

Abstract:

Data sensing in wireless sensor networks is done by query deceleration the network by the users. In many applications of the wireless sensor networks, many users send queries to the network simultaneously. If the queries are processed separately, the network’s energy consumption will increase significantly. Therefore, it is very important to aggregate the queries before sending them to the network. In this paper, we propose a multiple query optimization framework based on sensors physical and temporal correlation. In the proposed method, queries are merged and sent to network by considering correlation among the sensors in order to reduce the communication cost between the sensors and the base station.

Keywords: wireless sensor networks, multiple query optimization, data correlation, reducing energy consumption

Procedia PDF Downloads 334

Authors: Mina Beheshti, Vida Shams, Mojtaba Esfandiari, Farzaneh Abdollahi, Abdolreza Ohadi

Abstract:

Most of the people who survive a stroke need rehabilitation tools to regain their mobility. The core function of these devices is a brake actuator. The goal of this study is to design and control a magnetorheological brake which can be used as a rehabilitation tool. In fact, the fluid used in this brake is called magnetorheological fluid or MR that properties can change by variation of the magnetic field. The braking properties can be set as control by using this feature of the fluid. In this research, different MR brake designs are first introduced in each design, and the dimensions of the brake have been determined based on the required torque for foot movement. To calculate the brake dimensions, it is assumed that the shear stress distribution in the fluid is uniform and the fluid is in its saturated state. After designing the rehabilitation brake, the mathematical model of the healthy movement of a healthy person is extracted. Due to the nonlinear nature of the system and its variability, various adaptive controllers, neural networks, and robust have been implemented to estimate the parameters and control the system. After calculating torque and control current, the best type of controller in terms of error and control current has been selected. Finally, this controller is implemented on the experimental data of the patient's movements, and the control current is calculated to achieve the desired torque and motion.

Keywords: rehabilitation, magnetorheological fluid, knee, brake, adaptive control, robust control, neural network control, torque control

Procedia PDF Downloads 151

Authors: S. S. Kenny Lee, C. C. Kong, S. K. Ting

Abstract:

Teaching engineering lab provides opportunity for students to practice theories learned through physical experiment in the laboratory. However, building laboratories to accommodate increased number of students are expensive, making it impossible for an educational institution to afford the high expenses. In this paper, we develop a low cost and remote platform to aid teaching undergraduate students. The platform is constructed where the real experiment setting up in laboratory can be reconfigure and accessed remotely, the aim is to increase student’s desire to learn at which they can interact with the physical experiment using network enabled devices at anywhere in the campus. The platform is constructed with Raspberry Pi as a main control board that provides communication between computer interfaces to the actual experiment preset in the laboratory. The interface allows real-time remote viewing and triggering the physical experiment in the laboratory and also provides instructions and learning guide about the experimental.

Keywords: engineering lab, low cost, network, remote platform, reconfigure, real-time

Procedia PDF Downloads 308

Authors: Jessica Liqin Kong

Abstract:

Female engineers and scientists face unique challenges in their careers that make the development of professional networks crucial, but also more difficult. Working to overcome these challenges, ‘The Network’ was established in 2013 at the Queensland University of Technology (QUT) in Australia as an alumni chapter with the purpose of evoking continuous positive change for female participation and retention in science, technology, engineering and mathematics (STEM). ‘The Network’ adopts an innovative model for a Women in STEM alumni chapter which was inspired by the cradle to cradle approach to engagement, and the concept of growing and harvesting individual and collective social capital through a variety of initiatives. ‘The Network’ fosters an environment where the values exchanged in social and professional relationships can be capitalized for both current and future women in STEM. The model of ‘The Network’ acts as a simulation and opportunity for participants to further develop their leadership and other soft skills through learning, building and experimenting with ‘The Network’.

Keywords: women in STEM, engagement, Cradle-to-Cradle, social capital

Procedia PDF Downloads 284

Authors: Rachid Dehini, Brahim Berbaoui

Abstract:

The main objectives of shunt active power filter (SAPF) is to preserve the power system from unwanted harmonic currents produced by nonlinear loads, as well as to compensate the reactive power. The aim of this paper is to present a (PAPF) supplied by the Photovoltaic cells ,in such a way that the (PAPF) feeds the linear and nonlinear loads by harmonics currents and the excess of the energy is injected into the power system. In order to improve the performances of conventional (PAPF) This paper also proposes artificial neural networks (ANN) for harmonics identification and DC link voltage control. The simulation study results of the new (SAPF) identification technique are found quite satisfactory by assuring good filtering characteristics and high system stability.

Keywords: SAPF, harmonics current, photovoltaic cells, MPPT, artificial neural networks (ANN)

Procedia PDF Downloads 331