Search results for: feature pyramid networks

Authors: Soheila Sadeghi

Abstract:

Construction projects are complex and often subject to significant cost overruns due to the multifaceted nature of the activities involved. Accurate cost estimation is crucial for effective budget planning and resource allocation. Traditional methods for predicting overruns often rely on expert judgment or analysis of historical data, which can be time-consuming, subjective, and may fail to consider important factors. However, with the increasing availability of data from construction projects, machine learning techniques can be leveraged to improve the accuracy of overrun predictions. This study applied machine learning algorithms to enhance the prediction of cost overruns in a case study of a construction project. The methodology involved the development and evaluation of two machine learning models: Random Forest and Neural Networks. Random Forest can handle high-dimensional data, capture complex relationships, and provide feature importance estimates. Neural Networks, particularly Deep Neural Networks (DNNs), are capable of automatically learning and modeling complex, non-linear relationships between input features and the target variable. These models can adapt to new data, reduce human bias, and uncover hidden patterns in the dataset. The findings of this study demonstrate that both Random Forest and Neural Networks can significantly improve the accuracy of cost overrun predictions compared to traditional methods. The Random Forest model also identified key cost drivers and risk factors, such as changes in the scope of work and delays in material delivery, which can inform better project risk management. However, the study acknowledges several limitations. First, the findings are based on a single construction project, which may limit the generalizability of the results to other projects or contexts. Second, the dataset, although comprehensive, may not capture all relevant factors influencing cost overruns, such as external economic conditions or political factors. Third, the study focuses primarily on cost overruns, while schedule overruns are not explicitly addressed. Future research should explore the application of machine learning techniques to a broader range of projects, incorporate additional data sources, and investigate the prediction of both cost and schedule overruns simultaneously.

Keywords: cost prediction, machine learning, project management, random forest, neural networks

Procedia PDF Downloads 51

Authors: Myriam Ferreira

Abstract:

Websites and social networks allow museums to divulge their works by new and attractive means. Besides, these technologies provide tools to generate a new history of art’s contents and promote visits to their installations. At the same time, museums are proposing more and more activities to families, children and young people. However, these activities usually take place in the museum’s physical installations, while websites and social networks seem to be mainly targeted to adults. The problem is that being children and young people digital natives, they feel apart from museums, so they need a presence of museums in digital means to feel attracted to them. Some institutions are making efforts to fill this vacuum. In this paper, resources designed specifically for children and teenagers have been selected from websites and social networks of five Spanish Museums: Prado Museum, Thyssen Museum, Guggenheim Museum, America Museum and Cerralbo Museum. After that, we have carried out an investigation in a school with children and teenagers between 11 and 15 years old. Those young people have been asked about their valuation of those web pages and social networks, with quantitative-qualitative questions. The results show that the least rated resources were videos and social networks because they were considered ‘too serious’, while the most rated were games and augmented reality. These ratings confirm theoretical papers that affirm that the future of technologies applied to museums is edutainment and interaction.

Keywords: children, museums, social networks, teenagers, websites

Procedia PDF Downloads 149

Authors: Pei-Jyun Hsieh, Cheng-Hsuan Li, Bor-Chen Kuo

Abstract:

For hyperspectral image classification, feature reduction is an important pre-processing for avoiding the Hughes phenomena due to the difficulty for collecting training samples. Hence, lots of researches developed feature selection methods such as F-score, HSIC (Hilbert-Schmidt Independence Criterion), and etc., to improve hyperspectral image classification. However, most of them only consider the class separability in the original space, i.e., a linear class separability. In this study, we proposed a nonlinear class separability measure based on kernel trick for selecting an appropriate feature subset. The proposed nonlinear class separability was formed by a generalized RBF kernel with different bandwidths with respect to different features. Moreover, it considered the within-class separability and the between-class separability. A genetic algorithm was applied to tune these bandwidths such that the smallest with-class separability and the largest between-class separability simultaneously. This indicates the corresponding feature space is more suitable for classification. In addition, the corresponding nonlinear classification boundary can separate classes very well. These optimal bandwidths also show the importance of bands for hyperspectral image classification. The reciprocals of these bandwidths can be viewed as weights of bands. The smaller bandwidth, the larger weight of the band, and the more importance for classification. Hence, the descending order of the reciprocals of the bands gives an order for selecting the appropriate feature subsets. In the experiments, three hyperspectral image data sets, the Indian Pine Site data set, the PAVIA data set, and the Salinas A data set, were used to demonstrate the selected feature subsets by the proposed nonlinear feature selection method are more appropriate for hyperspectral image classification. Only ten percent of samples were randomly selected to form the training dataset. All non-background samples were used to form the testing dataset. The support vector machine was applied to classify these testing samples based on selected feature subsets. According to the experiments on the Indian Pine Site data set with 220 bands, the highest accuracies by applying the proposed method, F-score, and HSIC are 0.8795, 0.8795, and 0.87404, respectively. However, the proposed method selects 158 features. F-score and HSIC select 168 features and 217 features, respectively. Moreover, the classification accuracies increase dramatically only using first few features. The classification accuracies with respect to feature subsets of 10 features, 20 features, 50 features, and 110 features are 0.69587, 0.7348, 0.79217, and 0.84164, respectively. Furthermore, only using half selected features (110 features) of the proposed method, the corresponding classification accuracy (0.84168) is approximate to the highest classification accuracy, 0.8795. For other two hyperspectral image data sets, the PAVIA data set and Salinas A data set, we can obtain the similar results. These results illustrate our proposed method can efficiently find feature subsets to improve hyperspectral image classification. One can apply the proposed method to determine the suitable feature subset first according to specific purposes. Then researchers can only use the corresponding sensors to obtain the hyperspectral image and classify the samples. This can not only improve the classification performance but also reduce the cost for obtaining hyperspectral images.

Keywords: hyperspectral image classification, nonlinear feature selection, kernel trick, support vector machine

Procedia PDF Downloads 262

Authors: Chung-Yuan Huang, Yu-Hsiang Fu, Chuen-Tsai Sun

Abstract:

Community detection represents an important data-mining tool for analyzing and understanding real-world complex network structures and functions. We believe that at least four criteria determine the appropriateness of a community detection algorithm: (a) it produces useable normalized mutual information (NMI) and modularity results for social networks, (b) it overcomes resolution limitation problems associated with synthetic networks, (c) it produces good NMI results and performance efficiency for Lancichinetti-Fortunato-Radicchi (LFR) benchmark networks, and (d) it produces good modularity and performance efficiency for large-scale real-world complex networks. To our knowledge, no existing community detection algorithm meets all four criteria. In this paper, we describe a simple hierarchical arc-merging (HAM) algorithm that uses network topologies and rule-based arc-merging strategies to identify community structures that satisfy the criteria. We used five well-studied social network datasets and eight sets of LFR benchmark networks to validate the ground-truth community correctness of HAM, eight large-scale real-world complex networks to measure its performance efficiency, and two synthetic networks to determine its susceptibility to resolution limitation problems. Our results indicate that the proposed HAM algorithm is capable of providing satisfactory performance efficiency and that HAM-identified communities were close to ground-truth communities in social and LFR benchmark networks while overcoming resolution limitation problems.

Keywords: complex network, social network, community detection, network hierarchy

Procedia PDF Downloads 226

Authors: Arwa Alzughaibi

Abstract:

Human Motion detection is a challenging task due to a number of factors including variable appearance, posture and a wide range of illumination conditions and background. So, the first need of such a model is a reliable feature set that can discriminate between a human and a non-human form with a fair amount of confidence even under difficult conditions. By having richer representations, the classification task becomes easier and improved results can be achieved. The Aim of this paper is to investigate the reliable and accurate human motion detection models that are able to detect the human motions accurately under varying illumination levels and backgrounds. Different sets of features are tried and tested including Histogram of Oriented Gradients (HOG), Deformable Parts Model (DPM), Local Decorrelated Channel Feature (LDCF) and Aggregate Channel Feature (ACF). However, we propose an efficient and reliable human motion detection approach by combining Histogram of oriented gradients (HOG) and local phase quantization (LPQ) as the feature set, and implementing search pruning algorithm based on optical flow to reduce the number of false positive. Experimental results show the effectiveness of combining local phase quantization descriptor and the histogram of gradient to perform perfectly well for a large range of illumination conditions and backgrounds than the state-of-the-art human detectors. Areaunder th ROC Curve (AUC) of the proposed method achieved 0.781 for UCF dataset and 0.826 for CDW dataset which indicates that it performs comparably better than HOG, DPM, LDCF and ACF methods.

Keywords: human motion detection, histograms of oriented gradient, local phase quantization, local phase quantization

Procedia PDF Downloads 256

Authors: Rahul Malhotra, Anurag Sharma

Abstract:

Fiber based access networks can deliver performance that can support the increasing demands for high speed connections. One of the new technologies that have emerged in recent years is Passive Optical Networks. This paper is targeted to show the simultaneous delivery of triple play service (data, voice, and video). The comparative investigation and suitability of various data rates is presented. It is demonstrated that as we increase the data rate, number of users to be accommodated decreases due to increase in bit error rate.

Keywords: FTTH, quad play, play service, access networks, data rate

Procedia PDF Downloads 411

Authors: Si-Gwan Kim

Abstract:

In Delay Tolerant Networks (DTN), there may not exist an end-to-end path between source and destination at the time of message transmission. Employing ‘Store Carry and Forward’ delivery mechanism for message transmission in such networks usually incurs long message delays. In this paper, we present the modified Binary Spray and Wait (BSW) routing protocol that enhances the performance of the original one. Our proposed algorithm adjusts the number of forward messages depending on the number of neighbor nodes. By using beacon messages periodically, the number of neighbor nodes can be managed. The simulation using ONE simulator results shows that our modified version gives higher delivery ratio and less latency as compared to BSW.

Keywords: delay tolerant networks, store carry and forward, one simulator, binary spray and wait

Procedia PDF Downloads 123

Authors: Paulo Cesar Pereira Junior, Alexandre Monteiro, Rafael da Luz Ribeiro, Antonio Carlos Sobieranski, Aldo von Wangenheim

Abstract:

In this paper, we present a comparison between convolutional neural networks and classical computer vision approaches, for the specific precision agriculture problem of weed mapping on sugarcane fields aerial images. A systematic literature review was conducted to find which computer vision methods are being used on this specific problem. The most cited methods were implemented, as well as four models of convolutional neural networks. All implemented approaches were tested using the same dataset, and their results were quantitatively and qualitatively analyzed. The obtained results were compared to a human expert made ground truth for validation. The results indicate that the convolutional neural networks present better precision and generalize better than the classical models.

Keywords: convolutional neural networks, deep learning, digital image processing, precision agriculture, semantic segmentation, unmanned aerial vehicles

Procedia PDF Downloads 258

Authors: Daham Owaid Matrood, Naqaa Hussein Raheem

Abstract:

Evolutionary algorithms and Artificial neural networks (ANN) are two relatively young research areas that were subject to a steadily growing interest during the past years. This paper examines the use of Particle Swarm Optimization (PSO) to train a multi-layer feed forward neural network for demand forecasting. We use in this paper weekly demand data for packed cement and towels, which have been outfitted by the Northern General Company for Cement and General Company of prepared clothes respectively. The results showed superiority of trained neural networks using particle swarm optimization on neural networks trained using error back propagation because their ability to escape from local optima.

Keywords: artificial neural network, demand forecasting, particle swarm optimization, weight optimization

Procedia PDF Downloads 450

Authors: RAM PAL SINGH, VIKASH CHAUDHARY, MONIKA VERMA

Abstract:

In this paper, we proposed the implementation of optimization based Extreme Learning Machine (ELM) for watermarking of B-channel of color image in discrete wavelet transform (DWT) domain. ELM, a regularization algorithm, works based on generalized single-hidden-layer feed-forward neural networks (SLFNs). However, hidden layer parameters, generally called feature mapping in context of ELM need not to be tuned every time. This paper shows the embedding and extraction processes of watermark with the help of ELM and results are compared with already used machine learning models for watermarking.Here, a cover image is divide into suitable numbers of non-overlapping blocks of required size and DWT is applied to each block to be transformed in low frequency sub-band domain. Basically, ELM gives a unified leaning platform with a feature mapping, that is, mapping between hidden layer and output layer of SLFNs, is tried for watermark embedding and extraction purpose in a cover image. Although ELM has widespread application right from binary classification, multiclass classification to regression and function estimation etc. Unlike SVM based algorithm which achieve suboptimal solution with high computational complexity, ELM can provide better generalization performance results with very small complexity. Efficacy of optimization method based ELM algorithm is measured by using quantitative and qualitative parameters on a watermarked image even though image is subjected to different types of geometrical and conventional attacks.

Keywords: BER, DWT, extreme leaning machine (ELM), PSNR

Procedia PDF Downloads 310

Authors: Jonathan Gong

Abstract:

Purpose AI-driven solutions are at the forefront of many pathology and medical imaging methods. Using algorithms designed to better the experience of medical professionals within their respective fields, the efficiency and accuracy of diagnosis can improve. In particular, X-rays are a fast and relatively inexpensive test that can diagnose diseases. In recent years, X-rays have not been widely used to detect and diagnose COVID-19. The under use of Xrays is mainly due to the low diagnostic accuracy and confounding with pneumonia, another respiratory disease. However, research in this field has expressed a possibility that artificial neural networks can successfully diagnose COVID-19 with high accuracy. Models and Data The dataset used is the COVID-19 Radiography Database. This dataset includes images and masks of chest X-rays under the labels of COVID-19, normal, and pneumonia. The classification model developed uses an autoencoder and a pre-trained convolutional neural network (DenseNet201) to provide transfer learning to the model. The model then uses a deep neural network to finalize the feature extraction and predict the diagnosis for the input image. This model was trained on 4035 images and validated on 807 separate images from the ones used for training. The images used to train the classification model include an important feature: the pictures are cropped beforehand to eliminate distractions when training the model. The image segmentation model uses an improved U-Net architecture. This model is used to extract the lung mask from the chest X-ray image. The model is trained on 8577 images and validated on a validation split of 20%. These models are calculated using the external dataset for validation. The models’ accuracy, precision, recall, f1-score, IOU, and loss are calculated. Results The classification model achieved an accuracy of 97.65% and a loss of 0.1234 when differentiating COVID19-infected, pneumonia-infected, and normal lung X-rays. The segmentation model achieved an accuracy of 97.31% and an IOU of 0.928. Conclusion The models proposed can detect COVID-19, pneumonia, and normal lungs with high accuracy and derive the lung mask from a chest X-ray with similarly high accuracy. The hope is for these models to elevate the experience of medical professionals and provide insight into the future of the methods used.

Keywords: artificial intelligence, convolutional neural networks, deep learning, image processing, machine learning

Procedia PDF Downloads 129

Authors: Belkacem Soundes, Guezouli Larbi

Abstract:

This paper presents a robust and accurate method for text detection and localization over lecture videos. Frame regions are classified into text or background based on visual feature analysis. However, lecture video shows significant degradation mainly related to acquisition conditions, camera motion and environmental changes resulting in low quality videos. Hence, affecting feature extraction and description efficiency. Moreover, traditional text detection methods cannot be directly applied to lecture videos. Therefore, robust feature extraction methods dedicated to this specific video genre are required for robust and accurate text detection and extraction. Method consists of a three-step process: Slide region detection and segmentation; Feature extraction and non-text filtering. For robust and effective features extraction moment functions are used. Two distinct types of moments are used: orthogonal and non-orthogonal. For orthogonal Zernike Moments, both Pseudo Zernike moments are used, whereas for non-orthogonal ones Hu moments are used. Expressivity and description efficiency are given and discussed. Proposed approach shows that in general, orthogonal moments show high accuracy in comparison to the non-orthogonal one. Pseudo Zernike moments are more effective than Zernike with better computation time.

Keywords: text detection, text localization, lecture videos, pseudo zernike moments

Procedia PDF Downloads 149

Authors: Faisal Al Yahmadi, Muhammad R. Ahmed

Abstract:

Electric power is a fundamental necessity in the 21^st century. Consequently, any break in electric power is probably going to affect the general activity. To make the power supply smooth and efficient, a smart grid network is introduced which uses communication technology. In any communication network, security is essential. It has been observed from several recent incidents that adversary causes an interruption to the operation of networks. In order to resolve the issues, it is vital to understand the threats and vulnerabilities associated with the smart grid networks. In this paper, we have investigated the threats and vulnerabilities in Smart Grid Networks (SGN) and the few solutions in the literature. Proposed solutions showed developments in electricity theft countermeasures, Denial of services attacks (DoS) and malicious injection attacks detection model, as well as malicious nodes detection using watchdog like techniques and other solutions.

Keywords: smart grid network, security, threats, vulnerabilities

Procedia PDF Downloads 138

Authors: Ardavan Zamanpour, Somaieh Yassari

Abstract:

ZigBee protocol network was developed in industries and MIT laboratory in 1997. ZigBee is a wireless networking technology by alliance ZigBee which is designed to low board and low data rate applications. It is a Protocol which connects between electrical devises with very low energy and cost. The first version of IEEE 802.15.4 which was formed ZigBee was based on 2.4GHZ MHZ 912MHZ 868 frequency band. The name of system is often reminded random directions that bees (BEES) traversing during pollination of products. Such as alloy of the ways in which information packets are traversed within the mesh network. This paper aims to study the performance and effectiveness of this protocol in wireless networks.

Keywords: ZigBee, protocol, wireless, networks

Procedia PDF Downloads 368

Authors: W. Shin, Y. Kim

Abstract:

This study focuses on disconfirming that wide-area networks can be made mobile, highly-available, and wireless. This methodological test shows that IPv7 and context-free grammar are mismatched. In the cases of robots, a similar tendency is also revealed. Further, we also prove that public-private key pairs could be built embedded, adaptive, and wireless. Finally, we disconfirm that although hash tables can be made distributed, interposable, and autonomous, XML and DNS can interfere to realize this purpose. Our experiments soon proved that exokernelizing our replicated Knesis keyboards was more significant than interrupting them. Our experiments exhibited degraded average sampling rate.

Keywords: collaborative communication, DNS, local-area networks, XML

Procedia PDF Downloads 185

Authors: Huayuan Lu, Chunfang Yang, Ma Zhu, Baojun Qi, Yaqiong Qiao, Jiangqian Xu

Abstract:

Image geolocalization has great application prospects in fields such as autonomous driving and virtual/augmented reality. In practical application scenarios, the size of the image to be located is not fixed; it is impractical to train different networks for all possible sizes. When its size does not match the size of the input of the descriptor extraction model, existing image geolocalization methods usually directly scale or crop the image in some common ways. This will result in the loss of some information important to the geolocalization task, thus affecting the performance of the image geolocalization method. For example, excessive down-sampling can lead to blurred building contour, and inappropriate cropping can lead to the loss of key semantic elements, resulting in incorrect geolocation results. To address this problem, this paper designs a learnable image resizer and proposes an arbitrary-sized image geolocation method. (1) The designed learnable image resizer employs the self-attention mechanism to enhance the geometric features of the resized image. Firstly, it applies bilinear interpolation to the input image and its feature maps to obtain the initial resized image and the resized feature maps. Then, SKNet (selective kernel net) is used to approximate the best receptive field, thus keeping the geometric shapes as the original image. And SENet (squeeze and extraction net) is used to automatically select the feature maps with strong contour information, enhancing the geometric features. Finally, the enhanced geometric features are fused with the initial resized image, to obtain the final resized images. (2) The proposed image geolocalization method embeds the above image resizer as a fronting layer of the descriptor extraction network. It not only enables the network to be compatible with arbitrary-sized input images but also enhances the geometric features that are crucial to the image geolocalization task. Moreover, the triplet attention mechanism is added after the first convolutional layer of the backbone network to optimize the utilization of geometric elements extracted by the first convolutional layer. Finally, the local features extracted by the backbone network are aggregated to form image descriptors for image geolocalization. The proposed method was evaluated on several mainstream datasets, such as Pittsburgh30K, Tokyo24/7, and Places365. The results show that the proposed method has excellent size compatibility and compares favorably to recently mainstream geolocalization methods.

Keywords: image geolocalization, self-attention mechanism, image resizer, geometric feature

Procedia PDF Downloads 213

Authors: Bhalke D. G., Bormane D. S., Kharate G. K.

Abstract:

This paper presents classification of musical instrument using machine learning techniques. The classification has been carried out using temporal, spectral, cepstral and wavelet features. Detail feature analysis is carried out using separate and combined features. Further, instrument model has been developed using K-Nearest Neighbor and Support Vector Machine (SVM). Benchmarked McGill university database has been used to test the performance of the system. Experimental result shows that SVM performs better as compared to KNN classifier.

Keywords: feature extraction, SVM, KNN, musical instruments

Procedia PDF Downloads 479

Authors: Edward Guillén, Jeisson Sánchez, Carlos Omar Ramos

Abstract:

In recent years, a wide variety of applications are developed with Support Vector Machines -SVM- methods and Artificial Neural Networks -ANN-. In general, these methods depend on intrusion knowledge databases such as KDD99, ISCX, and CAIDA among others. New classes of detectors are generated by machine learning techniques, trained and tested over network databases. Thereafter, detectors are employed to detect anomalies in network communication scenarios according to user’s connections behavior. The first detector based on training dataset is deployed in different real-world networks with mobile and non-mobile devices to analyze the performance and accuracy over static detection. The vulnerabilities are based on previous work in telemedicine apps that were developed on the research group. This paper presents the differences on detections results between some network scenarios by applying traditional detectors deployed with artificial neural networks and support vector machines.

Keywords: anomaly detection, back-propagation neural networks, network intrusion detection systems, support vector machines

Procedia PDF Downloads 356

Authors: K. Bresilla, L. Manfrini, B. Morandi, A. Boini, G. Perulli, L. C. Grappadelli

Abstract:

Image/video processing for fruit in the tree using hard-coded feature extraction algorithms have shown high accuracy during recent years. While accurate, these approaches even with high-end hardware are computationally intensive and too slow for real-time systems. This paper details the use of deep convolution neural networks (CNNs), specifically an algorithm (YOLO - You Only Look Once) with 24+2 convolution layers. Using deep-learning techniques eliminated the need for hard-code specific features for specific fruit shapes, color and/or other attributes. This CNN is trained on more than 5000 images of apple and pear fruits on 960 cores GPU (Graphical Processing Unit). Testing set showed an accuracy of 90%. After this, trained data were transferred to an embedded device (Raspberry Pi gen.3) with camera for more portability. Based on correlation between number of visible fruits or detected fruits on one frame and the real number of fruits on one tree, a model was created to accommodate this error rate. Speed of processing and detection of the whole platform was higher than 40 frames per second. This speed is fast enough for any grasping/harvesting robotic arm or other real-time applications.

Keywords: artificial intelligence, computer vision, deep learning, fruit recognition, harvesting robot, precision agriculture

Procedia PDF Downloads 420

Authors: Ali Isapour, Ramin Nateghi

Abstract:

— Markov parameters are unique parameters of the system and remain unchanged under similarity transformations. Markov parameters from a power series that is convergent only if the system matrix’s eigenvalues are inside the unity circle. Therefore, Markov parameters of a stable discrete-time system are convergent. In this study, we aim to realize the system based on Markov parameters by using Artificial Neural Networks (ANN), and this end, we use Flexible Neural Networks. Realization means determining the elements of matrices A, B, C, and D.

Keywords: Markov parameters, realization, activation function, flexible neural network

Procedia PDF Downloads 193

Authors: Majid Habib Khan, Jinwei Zhao, Xinhong Hei, Liu Jiedong, Rana Shahzad Noor, Muhammad Imran

Abstract:

Deep learning is increasingly employed in image classification, yet the scarcity and high cost of labeled data for training remain a challenge. Limited samples often lead to overfitting due to biased sample distribution. This paper introduces a dynamic distribution calibration method for few-shot learning. Initially, base and new class samples undergo normalization to mitigate disparate feature magnitudes. A pre-trained model then extracts feature vectors from both classes. The method dynamically selects distribution characteristics from base classes (both adjacent and remote) in the embedding space, using a threshold value approach for new class samples. Given the propensity of similar classes to share feature distributions like mean and variance, this research assumes a Gaussian distribution for feature vectors. Subsequently, distributional features of new class samples are calibrated using a corrected hyperparameter, derived from the distribution features of both adjacent and distant base classes. This calibration augments the new class sample set. The technique demonstrates significant improvements, with up to 4% accuracy gains in few-shot classification challenges, as evidenced by tests on miniImagenet and CUB datasets.

Keywords: deep learning, computer vision, image classification, few-shot learning, threshold

Procedia PDF Downloads 64

Authors: Zhifeng Kong

Abstract:

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

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

Procedia PDF Downloads 142

Authors: Shervin Alaei, Reza Moradi

Abstract:

Trading in financial markets is subject to risks due to their high volatilities. Here, using an LSTM neural network, and by doing some risk-based feature engineering tasks, we developed a method that can accurately predict trends of the Tehran stock exchange market index from a few days ago. Our test results have shown that the proposed method with an average prediction accuracy of more than 94% is superior to the other common machine learning algorithms. To the best of our knowledge, this is the first work incorporating deep learning and risk factors to accurately predict market trends.

Keywords: deep learning, LSTM, trend prediction, risk management, artificial neural networks

Procedia PDF Downloads 153

Authors: Frodouard Minani

Abstract:

Since last decade, wireless sensor networks (WSNs) have been used in many areas like health care, agriculture, defense, military, disaster hit areas and so on. Wireless Sensor Networks consist of a Base Station (BS) and more number of wireless sensors in order to monitor temperature, pressure, motion in different environment conditions. The key parameter that plays a major role in designing a protocol for Wireless Sensor Networks is energy efficiency which is a scarcest resource of sensor nodes and it determines the lifetime of sensor nodes. Maximizing sensor node’s lifetime is an important issue in the design of applications and protocols for Wireless Sensor Networks. Clustering sensor nodes mechanism is an effective topology control approach for helping to achieve the goal of this research. In this paper, the researcher presents an energy efficiency protocol to prolong the network lifetime based on Energy efficient clustering algorithm. The Low Energy Adaptive Clustering Hierarchy (LEACH) is a routing protocol for clusters which is used to lower the energy consumption and also to improve the lifetime of the Wireless Sensor Networks. Maximizing energy dissipation and network lifetime are important matters in the design of applications and protocols for wireless sensor networks. Proposed system is to maximize the lifetime of the Wireless Sensor Networks by choosing the farthest cluster head (CH) instead of the closest CH and forming the cluster by considering the following parameter metrics such as Node’s density, residual-energy and distance between clusters (inter-cluster distance). In this paper, comparisons between the proposed protocol and comparative protocols in different scenarios have been done and the simulation results showed that the proposed protocol performs well over other comparative protocols in various scenarios.

Keywords: base station, clustering algorithm, energy efficient, sensors, wireless sensor networks

Procedia PDF Downloads 143

Authors: Claudio Di Benedetto, Irene Bengo

Abstract:

The objective of the paper is to study how large European companies develop social business (SB) at the base of the economic pyramid (BoP). BoP markets are defined as the four billions people living with an annual income below $3,260 in local purchasing power. Despite they are heterogeneous in terms of geographic range they present some common characteristics: the presence of significant unmet (social) needs, high level of informal economy and the so-called ‘poverty penalty’. As a result, most people living at BoP are excluded from the value created by the global market economy. But it is worth noting, that BoP population with an aggregate purchasing power of around $5 trillion a year, represent a huge opportunity for companies that want to enhance their long-term profitability perspective. We suggest that in this context, the development of SB is, for companies, an innovative and promising way to satisfy unmet social needs and to experience new forms of value creation. Indeed, SB can be considered a strategic model to develop CSR programs that fully integrate the social dimension into the business to create economic and social value simultaneously. Despite in literature many studies have been conducted on social business, only few have explicitly analyzed such phenomenon from a company perspective and their role in the development of such initiatives remains understudied with fragmented results. To fill this gap the paper analyzes the key characteristics of the social business initiatives developed by European companies at BoP. The study was performed analyzing 1475 European companies participating in the United Nation Global Compact, the world’s leading corporate social responsibility program. Through the analysis of the corporate websites the study identifies companies that actually do SB at BoP. For SB initiatives identified, information were collected according to a framework adapted from the SB model developed by preliminary results show that more than one hundred European companies have already implemented social businesses at BoP accounting for the 6,5% of the total. This percentage increases to 15% if the focus is on companies with more than 10.440 employees. In terms of geographic distribution 80% of companies doing SB at BoP are located in western and southern Europe. The companies more active in promoting SB belong to financial sector (20%), energy sector (17%) and food and beverage sector (12%). In terms of social needs addressed almost 30% of the companies develop SB to provide access to energy and WASH, 25% of companies develop SB to reduce local unemployment or to promote local entrepreneurship and 21% of companies develop SB to promote financial inclusion of poor. In developing SB companies implement different social business configurations ranging from forms of outsourcing to internal development models. The study identifies seven main configurations through which company develops social business and each configuration present distinguishing characteristics respect to the involvement of the company in the management, the resources provided and the benefits achieved. By performing different analysis on data collected the paper provides detailed insights on how European companies develop SB at BoP.

Keywords: base of the economic pyramid, corporate social responsibility, social business, social enterprise

Procedia PDF Downloads 226

Authors: Sathapath Kilaso

Abstract:

Nowadays, Today, wireless sensor networks are an important technology that works with Internet of Things. It is receiving various data from many sensor. Then sent to processing or storing. By wireless network or through the Internet. The devices around us are intelligent, can receiving/transmitting and processing data and communicating through the system. There are many applications of wireless sensor networks, such as smart city, smart farm, environmental management, weather. This article will explore the use of wireless sensor networks in Thailand and collect data from Thai Thesis database in 2012-2017. How to Implementing Wireless Sensor Network Technology. Advantage from this study To know the usage wireless technology in many fields. This will be beneficial for future research. In this study was found the most widely used wireless sensor network in agriculture field. Especially for smart farms. And the second is the adoption of the environment. Such as weather stations and water inspection.

Keywords: wireless sensor network, smart city, survey, Adhoc Network

Procedia PDF Downloads 206

Authors: Anca Radulescu, Danae Evans

Abstract:

The relationship between a network’s hardwiring and its emergent dynamics are central to neuroscience. We study the principles of this correspondence in a canonical setup (in which network nodes exhibit well-studied complex quadratic dynamics), then test their universality in biological networks. By extending methods from discrete dynamics, we study the effects of network connectivity on temporal patterns, encapsulating long-term behavior into the rich topology of network Mandelbrot sets. Then elements of fractal geometry can be used to predict and classify network behavior.

Keywords: canonical model, complex dynamics, dynamic networks, fractals, Mandelbrot set, network connectivity

Procedia PDF Downloads 307

Authors: Ayadi Aya, Ghorbel Oussama, M. Obeid Abdulfattah, Abid Mohamed

Abstract:

Nowadays, many wireless sensor networks have been distributed in the real world to collect valuable raw sensed data. The challenge is to extract high-level knowledge from this huge amount of data. However, the identification of outliers can lead to the discovery of useful and meaningful knowledge. In the field of wireless sensor networks, an outlier is defined as a measurement that deviates from the normal behavior of sensed data. Many detection techniques of outliers in WSNs have been extensively studied in the past decade and have focused on classic based algorithms. These techniques identify outlier in the real transaction dataset. This survey aims at providing a structured and comprehensive overview of the existing researches on classification based outlier detection techniques as applicable to WSNs. Thus, we have identified key hypotheses, which are used by these approaches to differentiate between normal and outlier behavior. In addition, this paper tries to provide an easier and a succinct understanding of the classification based techniques. Furthermore, we identified the advantages and disadvantages of different classification based techniques and we presented a comparative guide with useful paradigms for promoting outliers detection research in various WSN applications and suggested further opportunities for future research.

Keywords: bayesian networks, classification-based approaches, KPCA, neural networks, one-class SVM, outlier detection, wireless sensor networks

Procedia PDF Downloads 496

Authors: Shivani Singh, Shantanu Nakhare, Kalyani Nair, Rohan Shetty

Abstract:

As social networking is growing at a rapid pace today it is vital that we work on improving its management. Research has shown that the content present in online social networks may have significant influence on impressionable minds. If such platforms are misused, it will lead to negative consequences. Detecting insults or inappropriate messages continues to be one of the most challenging aspects of Online Social Networks (OSNs) today. We address this problem through a Machine Learning Based Soft Text Classifier approach using Support Vector Machine algorithm. The proposed system acts as a screening mechanism the alerts the user about such messages. The messages are classified according to their subject matter and each comment is labeled for the presence of profanity and insults.

Keywords: machine learning, online social networks, soft text classifier, support vector machine

Procedia PDF Downloads 507

Authors: Sabri Arik

Abstract:

In this paper, we study the global asymptotic robust stability of delayed neural networks with norm-bounded uncertainties. By employing the Lyapunov stability theory and Homeomorphic mapping theorem, we derive some new types of sufficient conditions ensuring the existence, uniqueness and global asymptotic stability of the equilibrium point for the class of neural networks with discrete time delays under parameter uncertainties and with respect to continuous and slopebounded activation functions. An important aspect of our results is their low computational complexity as the reported results can be verified by checking some properties symmetric matrices associated with the uncertainty sets of network parameters. The obtained results are shown to be generalization of some of the previously published corresponding results. Some comparative numerical examples are also constructed to compare our results with some closely related existing literature results.

Keywords: neural networks, delayed systems, lyapunov functionals, stability analysis

Procedia PDF Downloads 525