Search results for: deep maxout network
6199 Deep Routing Strategy: Deep Learning based Intelligent Routing in Software Defined Internet of Things.
Authors: Zabeehullah, Fahim Arif, Yawar Abbas
Abstract:
Software Defined Network (SDN) is a next genera-tion networking model which simplifies the traditional network complexities and improve the utilization of constrained resources. Currently, most of the SDN based Internet of Things(IoT) environments use traditional network routing strategies which work on the basis of max or min metric value. However, IoT network heterogeneity, dynamic traffic flow and complexity demands intelligent and self-adaptive routing algorithms because traditional routing algorithms lack the self-adaptions, intelligence and efficient utilization of resources. To some extent, SDN, due its flexibility, and centralized control has managed the IoT complexity and heterogeneity but still Software Defined IoT (SDIoT) lacks intelligence. To address this challenge, we proposed a model called Deep Routing Strategy (DRS) which uses Deep Learning algorithm to perform routing in SDIoT intelligently and efficiently. Our model uses real-time traffic for training and learning. Results demonstrate that proposed model has achieved high accuracy and low packet loss rate during path selection. Proposed model has also outperformed benchmark routing algorithm (OSPF). Moreover, proposed model provided encouraging results during high dynamic traffic flow.Keywords: SDN, IoT, DL, ML, DRS
Procedia PDF Downloads 1066198 Document-level Sentiment Analysis: An Exploratory Case Study of Low-resource Language Urdu
Authors: Ammarah Irum, Muhammad Ali Tahir
Abstract:
Document-level sentiment analysis in Urdu is a challenging Natural Language Processing (NLP) task due to the difficulty of working with lengthy texts in a language with constrained resources. Deep learning models, which are complex neural network architectures, are well-suited to text-based applications in addition to data formats like audio, image, and video. To investigate the potential of deep learning for Urdu sentiment analysis, we implemented five different deep learning models, including Bidirectional Long Short Term Memory (BiLSTM), Convolutional Neural Network (CNN), Convolutional Neural Network with Bidirectional Long Short Term Memory (CNN-BiLSTM), and Bidirectional Encoder Representation from Transformer (BERT). In this study, we developed a hybrid deep learning model called BiLSTM-Single Layer Multi Filter Convolutional Neural Network (BiLSTM-SLMFCNN) by fusing BiLSTM and CNN architecture. The proposed and baseline techniques are applied on Urdu Customer Support data set and IMDB Urdu movie review data set by using pre-trained Urdu word embedding that are suitable for sentiment analysis at the document level. Results of these techniques are evaluated and our proposed model outperforms all other deep learning techniques for Urdu sentiment analysis. BiLSTM-SLMFCNN outperformed the baseline deep learning models and achieved 83%, 79%, 83% and 94% accuracy on small, medium and large sized IMDB Urdu movie review data set and Urdu Customer Support data set respectively.Keywords: urdu sentiment analysis, deep learning, natural language processing, opinion mining, low-resource language
Procedia PDF Downloads 636197 Efficient Deep Neural Networks for Real-Time Strawberry Freshness Monitoring: A Transfer Learning Approach
Authors: Mst. Tuhin Akter, Sharun Akter Khushbu, S. M. Shaqib
Abstract:
A real-time system architecture is highly effective for monitoring and detecting various damaged products or fruits that may deteriorate over time or become infected with diseases. Deep learning models have proven to be effective in building such architectures. However, building a deep learning model from scratch is a time-consuming and costly process. A more efficient solution is to utilize deep neural network (DNN) based transfer learning models in the real-time monitoring architecture. This study focuses on using a novel strawberry dataset to develop effective transfer learning models for the proposed real-time monitoring system architecture, specifically for evaluating and detecting strawberry freshness. Several state-of-the-art transfer learning models were employed, and the best performing model was found to be Xception, demonstrating higher performance across evaluation metrics such as accuracy, recall, precision, and F1-score.Keywords: strawberry freshness evaluation, deep neural network, transfer learning, image augmentation
Procedia PDF Downloads 766196 A Comparison of Methods for Neural Network Aggregation
Authors: John Pomerat, Aviv Segev
Abstract:
Recently, deep learning has had many theoretical breakthroughs. For deep learning to be successful in the industry, however, there need to be practical algorithms capable of handling many real-world hiccups preventing the immediate application of a learning algorithm. Although AI promises to revolutionize the healthcare industry, getting access to patient data in order to train learning algorithms has not been easy. One proposed solution to this is data- sharing. In this paper, we propose an alternative protocol, based on multi-party computation, to train deep learning models while maintaining both the privacy and security of training data. We examine three methods of training neural networks in this way: Transfer learning, average ensemble learning, and series network learning. We compare these methods to the equivalent model obtained through data-sharing across two different experiments. Additionally, we address the security concerns of this protocol. While the motivating example is healthcare, our findings regarding multi-party computation of neural network training are purely theoretical and have use-cases outside the domain of healthcare.Keywords: neural network aggregation, multi-party computation, transfer learning, average ensemble learning
Procedia PDF Downloads 1526195 Prediction on Housing Price Based on Deep Learning
Authors: Li Yu, Chenlu Jiao, Hongrun Xin, Yan Wang, Kaiyang Wang
Abstract:
In order to study the impact of various factors on the housing price, we propose to build different prediction models based on deep learning to determine the existing data of the real estate in order to more accurately predict the housing price or its changing trend in the future. Considering that the factors which affect the housing price vary widely, the proposed prediction models include two categories. The first one is based on multiple characteristic factors of the real estate. We built Convolution Neural Network (CNN) prediction model and Long Short-Term Memory (LSTM) neural network prediction model based on deep learning, and logical regression model was implemented to make a comparison between these three models. Another prediction model is time series model. Based on deep learning, we proposed an LSTM-1 model purely regard to time series, then implementing and comparing the LSTM model and the Auto-Regressive and Moving Average (ARMA) model. In this paper, comprehensive study of the second-hand housing price in Beijing has been conducted from three aspects: crawling and analyzing, housing price predicting, and the result comparing. Ultimately the best model program was produced, which is of great significance to evaluation and prediction of the housing price in the real estate industry.Keywords: deep learning, convolutional neural network, LSTM, housing prediction
Procedia PDF Downloads 2986194 Domain-Specific Deep Neural Network Model for Classification of Abnormalities on Chest Radiographs
Authors: Nkechinyere Joy Olawuyi, Babajide Samuel Afolabi, Bola Ibitoye
Abstract:
This study collected a preprocessed dataset of chest radiographs and formulated a deep neural network model for detecting abnormalities. It also evaluated the performance of the formulated model and implemented a prototype of the formulated model. This was with the view to developing a deep neural network model to automatically classify abnormalities in chest radiographs. In order to achieve the overall purpose of this research, a large set of chest x-ray images were sourced for and collected from the CheXpert dataset, which is an online repository of annotated chest radiographs compiled by the Machine Learning Research Group, Stanford University. The chest radiographs were preprocessed into a format that can be fed into a deep neural network. The preprocessing techniques used were standardization and normalization. The classification problem was formulated as a multi-label binary classification model, which used convolutional neural network architecture to make a decision on whether an abnormality was present or not in the chest radiographs. The classification model was evaluated using specificity, sensitivity, and Area Under Curve (AUC) score as the parameter. A prototype of the classification model was implemented using Keras Open source deep learning framework in Python Programming Language. The AUC ROC curve of the model was able to classify Atelestasis, Support devices, Pleural effusion, Pneumonia, A normal CXR (no finding), Pneumothorax, and Consolidation. However, Lung opacity and Cardiomegaly had a probability of less than 0.5 and thus were classified as absent. Precision, recall, and F1 score values were 0.78; this implies that the number of False Positive and False Negative is the same, revealing some measure of label imbalance in the dataset. The study concluded that the developed model is sufficient to classify abnormalities present in chest radiographs into present or absent.Keywords: transfer learning, convolutional neural network, radiograph, classification, multi-label
Procedia PDF Downloads 1126193 Probing Syntax Information in Word Representations with Deep Metric Learning
Authors: Bowen Ding, Yihao Kuang
Abstract:
In recent years, with the development of large-scale pre-trained lan-guage models, building vector representations of text through deep neural network models has become a standard practice for natural language processing tasks. From the performance on downstream tasks, we can know that the text representation constructed by these models contains linguistic information, but its encoding mode and extent are unclear. In this work, a structural probe is proposed to detect whether the vector representation produced by a deep neural network is embedded with a syntax tree. The probe is trained with the deep metric learning method, so that the distance between word vectors in the metric space it defines encodes the distance of words on the syntax tree, and the norm of word vectors encodes the depth of words on the syntax tree. The experiment results on ELMo and BERT show that the syntax tree is encoded in their parameters and the word representations they produce.Keywords: deep metric learning, syntax tree probing, natural language processing, word representations
Procedia PDF Downloads 566192 Neural Network Based Decision Trees Using Machine Learning for Alzheimer's Diagnosis
Authors: P. S. Jagadeesh Kumar, Tracy Lin Huan, S. Meenakshi Sundaram
Abstract:
Alzheimer’s disease is one of the prevalent kind of ailment, expected for impudent reconciliation or an effectual therapy is to be accredited hitherto. Probable detonation of patients in the upcoming years, and consequently an enormous deal of apprehension in early discovery of the disorder, this will conceivably chaperon to enhanced healing outcomes. Complex impetuosity of the brain is an observant symbolic of the disease and a unique recognition of genetic sign of the disease. Machine learning alongside deep learning and decision tree reinforces the aptitude to absorb characteristics from multi-dimensional data’s and thus simplifies automatic classification of Alzheimer’s disease. Susceptible testing was prophesied and realized in training the prospect of Alzheimer’s disease classification built on machine learning advances. It was shrewd that the decision trees trained with deep neural network fashioned the excellent results parallel to related pattern classification.Keywords: Alzheimer's diagnosis, decision trees, deep neural network, machine learning, pattern classification
Procedia PDF Downloads 2906191 Personal Information Classification Based on Deep Learning in Automatic Form Filling System
Authors: Shunzuo Wu, Xudong Luo, Yuanxiu Liao
Abstract:
Recently, the rapid development of deep learning makes artificial intelligence (AI) penetrate into many fields, replacing manual work there. In particular, AI systems also become a research focus in the field of automatic office. To meet real needs in automatic officiating, in this paper we develop an automatic form filling system. Specifically, it uses two classical neural network models and several word embedding models to classify various relevant information elicited from the Internet. When training the neural network models, we use less noisy and balanced data for training. We conduct a series of experiments to test my systems and the results show that our system can achieve better classification results.Keywords: artificial intelligence and office, NLP, deep learning, text classification
Procedia PDF Downloads 1896190 The Detection of Implanted Radioactive Seeds on Ultrasound Images Using Convolution Neural Networks
Authors: Edward Holupka, John Rossman, Tye Morancy, Joseph Aronovitz, Irving Kaplan
Abstract:
A common modality for the treatment of early stage prostate cancer is the implantation of radioactive seeds directly into the prostate. The radioactive seeds are positioned inside the prostate to achieve optimal radiation dose coverage to the prostate. These radioactive seeds are positioned inside the prostate using Transrectal ultrasound imaging. Once all of the planned seeds have been implanted, two dimensional transaxial transrectal ultrasound images separated by 2 mm are obtained through out the prostate, beginning at the base of the prostate up to and including the apex. A common deep neural network, called DetectNet was trained to automatically determine the position of the implanted radioactive seeds within the prostate under ultrasound imaging. The results of the training using 950 training ultrasound images and 90 validation ultrasound images. The commonly used metrics for successful training were used to evaluate the efficacy and accuracy of the trained deep neural network and resulted in an loss_bbox (train) = 0.00, loss_coverage (train) = 1.89e-8, loss_bbox (validation) = 11.84, loss_coverage (validation) = 9.70, mAP (validation) = 66.87%, precision (validation) = 81.07%, and a recall (validation) = 82.29%, where train and validation refers to the training image set and validation refers to the validation training set. On the hardware platform used, the training expended 12.8 seconds per epoch. The network was trained for over 10,000 epochs. In addition, the seed locations as determined by the Deep Neural Network were compared to the seed locations as determined by a commercial software based on a one to three months after implant CT. The Deep Learning approach was within \strikeout off\uuline off\uwave off2.29\uuline default\uwave default mm of the seed locations determined by the commercial software. The Deep Learning approach to the determination of radioactive seed locations is robust, accurate, and fast and well within spatial agreement with the gold standard of CT determined seed coordinates.Keywords: prostate, deep neural network, seed implant, ultrasound
Procedia PDF Downloads 1876189 Deep Reinforcement Learning-Based Computation Offloading for 5G Vehicle-Aware Multi-Access Edge Computing Network
Authors: Ziying Wu, Danfeng Yan
Abstract:
Multi-Access Edge Computing (MEC) is one of the key technologies of the future 5G network. By deploying edge computing centers at the edge of wireless access network, the computation tasks can be offloaded to edge servers rather than the remote cloud server to meet the requirements of 5G low-latency and high-reliability application scenarios. Meanwhile, with the development of IOV (Internet of Vehicles) technology, various delay-sensitive and compute-intensive in-vehicle applications continue to appear. Compared with traditional internet business, these computation tasks have higher processing priority and lower delay requirements. In this paper, we design a 5G-based Vehicle-Aware Multi-Access Edge Computing Network (VAMECN) and propose a joint optimization problem of minimizing total system cost. In view of the problem, a deep reinforcement learning-based joint computation offloading and task migration optimization (JCOTM) algorithm is proposed, considering the influences of multiple factors such as concurrent multiple computation tasks, system computing resources distribution, and network communication bandwidth. And, the mixed integer nonlinear programming problem is described as a Markov Decision Process. Experiments show that our proposed algorithm can effectively reduce task processing delay and equipment energy consumption, optimize computing offloading and resource allocation schemes, and improve system resource utilization, compared with other computing offloading policies.Keywords: multi-access edge computing, computation offloading, 5th generation, vehicle-aware, deep reinforcement learning, deep q-network
Procedia PDF Downloads 1036188 A Case Study of Deep Learning for Disease Detection in Crops
Authors: Felipe A. Guth, Shane Ward, Kevin McDonnell
Abstract:
In the precision agriculture area, one of the main tasks is the automated detection of diseases in crops. Machine Learning algorithms have been studied in recent decades for such tasks in view of their potential for improving economic outcomes that automated disease detection may attain over crop fields. The latest generation of deep learning convolution neural networks has presented significant results in the area of image classification. In this way, this work has tested the implementation of an architecture of deep learning convolution neural network for the detection of diseases in different types of crops. A data augmentation strategy was used to meet the requirements of the algorithm implemented with a deep learning framework. Two test scenarios were deployed. The first scenario implemented a neural network under images extracted from a controlled environment while the second one took images both from the field and the controlled environment. The results evaluated the generalisation capacity of the neural networks in relation to the two types of images presented. Results yielded a general classification accuracy of 59% in scenario 1 and 96% in scenario 2.Keywords: convolutional neural networks, deep learning, disease detection, precision agriculture
Procedia PDF Downloads 2476187 Reconfigurable Intelligent Surfaces (RIS)-Assisted Integrated Leo Satellite and UAV for Non-terrestrial Networks Using a Deep Reinforcement Learning Approach
Authors: Tesfaw Belayneh Abebe
Abstract:
Integrating low-altitude earth orbit (LEO) satellites and unmanned aerial vehicles (UAVs) within a non-terrestrial network (NTN) with the assistance of reconfigurable intelligent surfaces (RIS), we investigate the problem of how to enhance throughput through integrated LEO satellites and UAVs with the assistance of RIS. We propose a method to jointly optimize the associations with the LEO satellite, the 3D trajectory of the UAV, and the phase shifts of the RIS to maximize communication throughput for RIS-assisted integrated LEO satellite and UAV-enabled wireless communications, which is challenging due to the time-varying changes in the position of the LEO satellite, the high mobility of UAVs, an enormous number of possible control actions, and also the large number of RIS elements. Utilizing a multi-agent double deep Q-network (MADDQN), our approach dynamically adjusts LEO satellite association, UAV positioning, and RIS phase shifts. Simulation results demonstrate that our method significantly outperforms baseline strategies in maximizing throughput. Lastly, thanks to the integrated network and the RIS, the proposed scheme achieves up to 65.66x higher peak throughput and 25.09x higher worst-case throughput.Keywords: integrating low-altitude earth orbit (LEO) satellites, unmanned aerial vehicles (UAVs) within a non-terrestrial network (NTN), reconfigurable intelligent surfaces (RIS), multi-agent double deep Q-network (MADDQN)
Procedia PDF Downloads 396186 Remote Sensing through Deep Neural Networks for Satellite Image Classification
Authors: Teja Sai Puligadda
Abstract:
Satellite images in detail can serve an important role in the geographic study. Quantitative and qualitative information provided by the satellite and remote sensing images minimizes the complexity of work and time. Data/images are captured at regular intervals by satellite remote sensing systems, and the amount of data collected is often enormous, and it expands rapidly as technology develops. Interpreting remote sensing images, geographic data mining, and researching distinct vegetation types such as agricultural and forests are all part of satellite image categorization. One of the biggest challenge data scientists faces while classifying satellite images is finding the best suitable classification algorithms based on the available that could able to classify images with utmost accuracy. In order to categorize satellite images, which is difficult due to the sheer volume of data, many academics are turning to deep learning machine algorithms. As, the CNN algorithm gives high accuracy in image recognition problems and automatically detects the important features without any human supervision and the ANN algorithm stores information on the entire network (Abhishek Gupta., 2020), these two deep learning algorithms have been used for satellite image classification. This project focuses on remote sensing through Deep Neural Networks i.e., ANN and CNN with Deep Sat (SAT-4) Airborne dataset for classifying images. Thus, in this project of classifying satellite images, the algorithms ANN and CNN are implemented, evaluated & compared and the performance is analyzed through evaluation metrics such as Accuracy and Loss. Additionally, the Neural Network algorithm which gives the lowest bias and lowest variance in solving multi-class satellite image classification is analyzed.Keywords: artificial neural network, convolutional neural network, remote sensing, accuracy, loss
Procedia PDF Downloads 1506185 Development of Deep Neural Network-Based Strain Values Prediction Models for Full-Scale Reinforced Concrete Frames Using Highly Flexible Sensing Sheets
Authors: Hui Zhang, Sherif Beskhyroun
Abstract:
Structural Health monitoring systems (SHM) are commonly used to identify and assess structural damage. In terms of damage detection, SHM needs to periodically collect data from sensors placed in the structure as damage-sensitive features. This includes abnormal changes caused by the strain field and abnormal symptoms of the structure, such as damage and deterioration. Currently, deploying sensors on a large scale in a building structure is a challenge. In this study, a highly stretchable strain sensors are used in this study to collect data sets of strain generated on the surface of full-size reinforced concrete (RC) frames under extreme cyclic load application. This sensing sheet can be switched freely between the test bending strain and the axial strain to achieve two different configurations. On this basis, the deep neural network prediction model of the frame beam and frame column is established. The training results show that the method can accurately predict the strain value and has good generalization ability. The two deep neural network prediction models will also be deployed in the SHM system in the future as part of the intelligent strain sensor system.Keywords: strain sensing sheets, deep neural networks, strain measurement, SHM system, RC frames
Procedia PDF Downloads 866184 Assessing Performance of Data Augmentation Techniques for a Convolutional Network Trained for Recognizing Humans in Drone Images
Authors: Masood Varshosaz, Kamyar Hasanpour
Abstract:
In recent years, we have seen growing interest in recognizing humans in drone images for post-disaster search and rescue operations. Deep learning algorithms have shown great promise in this area, but they often require large amounts of labeled data to train the models. To keep the data acquisition cost low, augmentation techniques can be used to create additional data from existing images. There are many techniques of such that can help generate variations of an original image to improve the performance of deep learning algorithms. While data augmentation is potentially assumed to improve the accuracy and robustness of the models, it is important to ensure that the performance gains are not outweighed by the additional computational cost or complexity of implementing the techniques. To this end, it is important to evaluate the impact of data augmentation on the performance of the deep learning models. In this paper, we evaluated the most currently available 2D data augmentation techniques on a standard convolutional network which was trained for recognizing humans in drone images. The techniques include rotation, scaling, random cropping, flipping, shifting, and their combination. The results showed that the augmented models perform 1-3% better compared to a base network. However, as the augmented images only contain the human parts already visible in the original images, a new data augmentation approach is needed to include the invisible parts of the human body. Thus, we suggest a new method that employs simulated 3D human models to generate new data for training the network.Keywords: human recognition, deep learning, drones, disaster mitigation
Procedia PDF Downloads 856183 Keyframe Extraction Using Face Quality Assessment and Convolution Neural Network
Authors: Rahma Abed, Sahbi Bahroun, Ezzeddine Zagrouba
Abstract:
Due to the huge amount of data in videos, extracting the relevant frames became a necessity and an essential step prior to performing face recognition. In this context, we propose a method for extracting keyframes from videos based on face quality and deep learning for a face recognition task. This method has two steps. We start by generating face quality scores for each face image based on the use of three face feature extractors, including Gabor, LBP, and HOG. The second step consists in training a Deep Convolutional Neural Network in a supervised manner in order to select the frames that have the best face quality. The obtained results show the effectiveness of the proposed method compared to the methods of the state of the art.Keywords: keyframe extraction, face quality assessment, face in video recognition, convolution neural network
Procedia PDF Downloads 2226182 Hyperspectral Band Selection for Oil Spill Detection Using Deep Neural Network
Authors: Asmau Mukhtar Ahmed, Olga Duran
Abstract:
Hydrocarbon (HC) spills constitute a significant problem that causes great concern to the environment. With the latest technology (hyperspectral images) and state of the earth techniques (image processing tools), hydrocarbon spills can easily be detected at an early stage to mitigate the effects caused by such menace. In this study; a controlled laboratory experiment was used, and clay soil was mixed and homogenized with different hydrocarbon types (diesel, bio-diesel, and petrol). The different mixtures were scanned with HYSPEX hyperspectral camera under constant illumination to generate the hypersectral datasets used for this experiment. So far, the Short Wave Infrared Region (SWIR) has been exploited in detecting HC spills with excellent accuracy. However, the Near-Infrared Region (NIR) is somewhat unexplored with regards to HC contamination and how it affects the spectrum of soils. In this study, Deep Neural Network (DNN) was applied to the controlled datasets to detect and quantify the amount of HC spills in soils in the Near-Infrared Region. The initial results are extremely encouraging because it indicates that the DNN was able to identify features of HC in the Near-Infrared Region with a good level of accuracy.Keywords: hydrocarbon, Deep Neural Network, short wave infrared region, near-infrared region, hyperspectral image
Procedia PDF Downloads 1066181 Emotion Detection in Twitter Messages Using Combination of Long Short-Term Memory and Convolutional Deep Neural Networks
Authors: Bahareh Golchin, Nooshin Riahi
Abstract:
One of the most significant issues as attended a lot in recent years is that of recognizing the sentiments and emotions in social media texts. The analysis of sentiments and emotions is intended to recognize the conceptual information such as the opinions, feelings, attitudes and emotions of people towards the products, services, organizations, people, topics, events and features in the written text. These indicate the greatness of the problem space. In the real world, businesses and organizations are always looking for tools to gather ideas, emotions, and directions of people about their products, services, or events related to their own. This article uses the Twitter social network, one of the most popular social networks with about 420 million active users, to extract data. Using this social network, users can share their information and opinions about personal issues, policies, products, events, etc. It can be used with appropriate classification of emotional states due to the availability of its data. In this study, supervised learning and deep neural network algorithms are used to classify the emotional states of Twitter users. The use of deep learning methods to increase the learning capacity of the model is an advantage due to the large amount of available data. Tweets collected on various topics are classified into four classes using a combination of two Bidirectional Long Short Term Memory network and a Convolutional network. The results obtained from this study with an average accuracy of 93%, show good results extracted from the proposed framework and improved accuracy compared to previous work.Keywords: emotion classification, sentiment analysis, social networks, deep neural networks
Procedia PDF Downloads 1326180 DMBR-Net: Deep Multiple-Resolution Bilateral Networks for Real-Time and Accurate Semantic Segmentation
Authors: Pengfei Meng, Shuangcheng Jia, Qian Li
Abstract:
We proposed a real-time high-precision semantic segmentation network based on a multi-resolution feature fusion module, the auxiliary feature extracting module, upsampling module, and atrous spatial pyramid pooling (ASPP) module. We designed a feature fusion structure, which is integrated with sufficient features of different resolutions. We also studied the effect of side-branch structure on the network and made discoveries. Based on the discoveries about the side-branch of the network structure, we used a side-branch auxiliary feature extraction layer in the network to improve the effectiveness of the network. We also designed upsampling module, which has better results than the original upsampling module. In addition, we also re-considered the locations and number of atrous spatial pyramid pooling (ASPP) modules and modified the network structure according to the experimental results to further improve the effectiveness of the network. The network presented in this paper takes the backbone network of Bisenetv2 as a basic network, based on which we constructed a network structure on which we made improvements. We named this network deep multiple-resolution bilateral networks for real-time, referred to as DMBR-Net. After experimental testing, our proposed DMBR-Net network achieved 81.2% mIoU at 119FPS on the Cityscapes validation dataset, 80.7% mIoU at 109FPS on the CamVid test dataset, 29.9% mIoU at 78FPS on the COCOStuff test dataset. Compared with all lightweight real-time semantic segmentation networks, our network achieves the highest accuracy at an appropriate speed.Keywords: multi-resolution feature fusion, atrous convolutional, bilateral networks, pyramid pooling
Procedia PDF Downloads 1396179 Understanding and Improving Neural Network Weight Initialization
Authors: Diego Aguirre, Olac Fuentes
Abstract:
In this paper, we present a taxonomy of weight initialization schemes used in deep learning. We survey the most representative techniques in each class and compare them in terms of overhead cost, convergence rate, and applicability. We also introduce a new weight initialization scheme. In this technique, we perform an initial feedforward pass through the network using an initialization mini-batch. Using statistics obtained from this pass, we initialize the weights of the network, so the following properties are met: 1) weight matrices are orthogonal; 2) ReLU layers produce a predetermined number of non-zero activations; 3) the output produced by each internal layer has a unit variance; 4) weights in the last layer are chosen to minimize the error in the initial mini-batch. We evaluate our method on three popular architectures, and a faster converge rates are achieved on the MNIST, CIFAR-10/100, and ImageNet datasets when compared to state-of-the-art initialization techniques.Keywords: deep learning, image classification, supervised learning, weight initialization
Procedia PDF Downloads 1256178 Satellite Imagery Classification Based on Deep Convolution Network
Authors: Zhong Ma, Zhuping Wang, Congxin Liu, Xiangzeng Liu
Abstract:
Satellite imagery classification is a challenging problem with many practical applications. In this paper, we designed a deep convolution neural network (DCNN) to classify the satellite imagery. The contributions of this paper are twofold — First, to cope with the large-scale variance in the satellite image, we introduced the inception module, which has multiple filters with different size at the same level, as the building block to build our DCNN model. Second, we proposed a genetic algorithm based method to efficiently search the best hyper-parameters of the DCNN in a large search space. The proposed method is evaluated on the benchmark database. The results of the proposed hyper-parameters search method show it will guide the search towards better regions of the parameter space. Based on the found hyper-parameters, we built our DCNN models, and evaluated its performance on satellite imagery classification, the results show the classification accuracy of proposed models outperform the state of the art method.Keywords: satellite imagery classification, deep convolution network, genetic algorithm, hyper-parameter optimization
Procedia PDF Downloads 2916177 A Custom Convolutional Neural Network with Hue, Saturation, Value Color for Malaria Classification
Authors: Ghazala Hcini, Imen Jdey, Hela Ltifi
Abstract:
Malaria disease should be considered and handled as a potential restorative catastrophe. One of the most challenging tasks in the field of microscopy image processing is due to differences in test design and vulnerability of cell classifications. In this article, we focused on applying deep learning to classify patients by identifying images of infected and uninfected cells. We performed multiple forms, counting a classification approach using the Hue, Saturation, Value (HSV) color space. HSV is used since of its superior ability to speak to image brightness; at long last, for classification, a convolutional neural network (CNN) architecture is created. Clusters of focus were used to deliver the classification. The highlights got to be forbidden, and a few more clamor sorts are included in the information. The suggested method has a precision of 99.79%, a recall value of 99.55%, and provides 99.96% accuracy.Keywords: deep learning, convolutional neural network, image classification, color transformation, HSV color, malaria diagnosis, malaria cells images
Procedia PDF Downloads 826176 Ophthalmic Hashing Based Supervision of Glaucoma and Corneal Disorders Imposed on Deep Graphical Model
Authors: P. S. Jagadeesh Kumar, Yang Yung, Mingmin Pan, Xianpei Li, Wenli Hu
Abstract:
Glaucoma is impelled by optic nerve mutilation habitually represented as cupping and visual field injury frequently with an arcuate pattern of mid-peripheral loss, subordinate to retinal ganglion cell damage and death. Glaucoma is the second foremost cause of blindness and the chief cause of permanent blindness worldwide. Consequently, all-embracing study into the analysis and empathy of glaucoma is happening to escort deep learning based neural network intrusions to deliberate this substantial optic neuropathy. This paper advances an ophthalmic hashing based supervision of glaucoma and corneal disorders preeminent on deep graphical model. Ophthalmic hashing is a newly proposed method extending the efficacy of visual hash-coding to predict glaucoma corneal disorder matching, which is the faster than the existing methods. Deep graphical model is proficient of learning interior explications of corneal disorders in satisfactory time to solve hard combinatoric incongruities using deep Boltzmann machines.Keywords: corneal disorders, deep Boltzmann machines, deep graphical model, glaucoma, neural networks, ophthalmic hashing
Procedia PDF Downloads 2386175 A Survey of Skin Cancer Detection and Classification from Skin Lesion Images Using Deep Learning
Authors: Joseph George, Anne Kotteswara Roa
Abstract:
Skin disease is one of the most common and popular kinds of health issues faced by people nowadays. Skin cancer (SC) is one among them, and its detection relies on the skin biopsy outputs and the expertise of the doctors, but it consumes more time and some inaccurate results. At the early stage, skin cancer detection is a challenging task, and it easily spreads to the whole body and leads to an increase in the mortality rate. Skin cancer is curable when it is detected at an early stage. In order to classify correct and accurate skin cancer, the critical task is skin cancer identification and classification, and it is more based on the cancer disease features such as shape, size, color, symmetry and etc. More similar characteristics are present in many skin diseases; hence it makes it a challenging issue to select important features from a skin cancer dataset images. Hence, the skin cancer diagnostic accuracy is improved by requiring an automated skin cancer detection and classification framework; thereby, the human expert’s scarcity is handled. Recently, the deep learning techniques like Convolutional neural network (CNN), Deep belief neural network (DBN), Artificial neural network (ANN), Recurrent neural network (RNN), and Long and short term memory (LSTM) have been widely used for the identification and classification of skin cancers. This survey reviews different DL techniques for skin cancer identification and classification. The performance metrics such as precision, recall, accuracy, sensitivity, specificity, and F-measures are used to evaluate the effectiveness of SC identification using DL techniques. By using these DL techniques, the classification accuracy increases along with the mitigation of computational complexities and time consumption.Keywords: skin cancer, deep learning, performance measures, accuracy, datasets
Procedia PDF Downloads 1196174 Hysteresis Modeling in Iron-Dominated Magnets Based on a Deep Neural Network Approach
Authors: Maria Amodeo, Pasquale Arpaia, Marco Buzio, Vincenzo Di Capua, Francesco Donnarumma
Abstract:
Different deep neural network architectures have been compared and tested to predict magnetic hysteresis in the context of pulsed electromagnets for experimental physics applications. Modelling quasi-static or dynamic major and especially minor hysteresis loops is one of the most challenging topics for computational magnetism. Recent attempts at mathematical prediction in this context using Preisach models could not attain better than percent-level accuracy. Hence, this work explores neural network approaches and shows that the architecture that best fits the measured magnetic field behaviour, including the effects of hysteresis and eddy currents, is the nonlinear autoregressive exogenous neural network (NARX) model. This architecture aims to achieve a relative RMSE of the order of a few 100 ppm for complex magnetic field cycling, including arbitrary sequences of pseudo-random high field and low field cycles. The NARX-based architecture is compared with the state-of-the-art, showing better performance than the classical operator-based and differential models, and is tested on a reference quadrupole magnetic lens used for CERN particle beams, chosen as a case study. The training and test datasets are a representative example of real-world magnet operation; this makes the good result obtained very promising for future applications in this context.Keywords: deep neural network, magnetic modelling, measurement and empirical software engineering, NARX
Procedia PDF Downloads 1266173 DNpro: A Deep Learning Network Approach to Predicting Protein Stability Changes Induced by Single-Site Mutations
Authors: Xiao Zhou, Jianlin Cheng
Abstract:
A single amino acid mutation can have a significant impact on the stability of protein structure. Thus, the prediction of protein stability change induced by single site mutations is critical and useful for studying protein function and structure. Here, we presented a deep learning network with the dropout technique for predicting protein stability changes upon single amino acid substitution. While using only protein sequence as input, the overall prediction accuracy of the method on a standard benchmark is >85%, which is higher than existing sequence-based methods and is comparable to the methods that use not only protein sequence but also tertiary structure, pH value and temperature. The results demonstrate that deep learning is a promising technique for protein stability prediction. The good performance of this sequence-based method makes it a valuable tool for predicting the impact of mutations on most proteins whose experimental structures are not available. Both the downloadable software package and the user-friendly web server (DNpro) that implement the method for predicting protein stability changes induced by amino acid mutations are freely available for the community to use.Keywords: bioinformatics, deep learning, protein stability prediction, biological data mining
Procedia PDF Downloads 4536172 Investigation on Behavior of Fixed-Ended Reinforced Concrete Deep Beams
Authors: Y. Heyrani Birak, R. Hizaji, J. Shahkarami
Abstract:
Reinforced Concrete (RC) deep beams are special structural elements because of their geometry and behavior under loads. For example, assumption of strain- stress distribution is not linear in the cross section. These types of beams may have simple supports or fixed supports. A lot of research works have been conducted on simply supported deep beams, but little study has been done in the fixed-end RC deep beams behavior. Recently, using of fixed-ended deep beams has been widely increased in structures. In this study, the behavior of fixed-ended deep beams is investigated, and the important parameters in capacity of this type of beams are mentioned.Keywords: deep beam, capacity, reinforced concrete, fixed-ended
Procedia PDF Downloads 3276171 Failure Mechanism in Fixed-Ended Reinforced Concrete Deep Beams under Cyclic Load
Authors: A. Aarabzadeh, R. Hizaji
Abstract:
Reinforced Concrete (RC) deep beams are a special type of beams due to their geometry, boundary conditions, and behavior compared to ordinary shallow beams. For example, assumption of a linear strain-stress distribution in the cross section is not valid. Little study has been dedicated to fixed-end RC deep beams. Also, most experimental studies are carried out on simply supported deep beams. Regarding recent tendency for application of deep beams, possibility of using fixed-ended deep beams has been widely increased in structures. Therefore, it seems necessary to investigate the aforementioned structural element in more details. In addition to experimental investigation of a concrete deep beam under cyclic load, different failure mechanisms of fixed-ended deep beams under this type of loading have been evaluated in the present study. The results show that failure mechanisms of deep beams under cyclic loads are quite different from monotonic loads.Keywords: deep beam, cyclic load, reinforced concrete, fixed-ended
Procedia PDF Downloads 3526170 A Deep Learning Model with Greedy Layer-Wise Pretraining Approach for Optimal Syngas Production by Dry Reforming of Methane
Authors: Maryam Zarabian, Hector Guzman, Pedro Pereira-Almao, Abraham Fapojuwo
Abstract:
Dry reforming of methane (DRM) has sparked significant industrial and scientific interest not only as a viable alternative for addressing the environmental concerns of two main contributors of the greenhouse effect, i.e., carbon dioxide (CO₂) and methane (CH₄), but also produces syngas, i.e., a mixture of hydrogen (H₂) and carbon monoxide (CO) utilized by a wide range of downstream processes as a feedstock for other chemical productions. In this study, we develop an AI-enable syngas production model to tackle the problem of achieving an equivalent H₂/CO ratio [1:1] with respect to the most efficient conversion. Firstly, the unsupervised density-based spatial clustering of applications with noise (DBSAN) algorithm removes outlier data points from the original experimental dataset. Then, random forest (RF) and deep neural network (DNN) models employ the error-free dataset to predict the DRM results. DNN models inherently would not be able to obtain accurate predictions without a huge dataset. To cope with this limitation, we employ reusing pre-trained layers’ approaches such as transfer learning and greedy layer-wise pretraining. Compared to the other deep models (i.e., pure deep model and transferred deep model), the greedy layer-wise pre-trained deep model provides the most accurate prediction as well as similar accuracy to the RF model with R² values 1.00, 0.999, 0.999, 0.999, 0.999, and 0.999 for the total outlet flow, H₂/CO ratio, H₂ yield, CO yield, CH₄ conversion, and CO₂ conversion outputs, respectively.Keywords: artificial intelligence, dry reforming of methane, artificial neural network, deep learning, machine learning, transfer learning, greedy layer-wise pretraining
Procedia PDF Downloads 80