Search results for: deep learning image model

Authors: Tingwei Shu, Dong Zhou, Chengjun Guo

Abstract:

Semantic communication is an emerging form of communication that realize intelligent communication by extracting semantic information of data at the source and transmitting it, and recovering the data at the receiving end. It can effectively solve the problem of data transmission under the situation of large data volume, low SNR and restricted bandwidth. With the development of Deep Learning, semantic communication further matures and is gradually applied in the fields of the Internet of Things, Uumanned Air Vehicle cluster communication, remote sensing scenarios, etc. We propose an improved semantic communication system for the situation where the data volume is huge and the spectrum resources are limited during the transmission of remote sensing images. At the transmitting, we need to extract the semantic information of remote sensing images, but there are some problems. The traditional semantic communication system based on Convolutional Neural Network cannot take into account the global semantic information and local semantic information of the image, which results in less-than-ideal image recovery at the receiving end. Therefore, we adopt the improved vision-Transformer-based structure as the semantic encoder instead of the mainstream one using CNN to extract the image semantic features. In this paper, we first perform pre-processing operations on remote sensing images to improve the resolution of the images in order to obtain images with more semantic information. We use wavelet transform to decompose the image into high-frequency and low-frequency components, perform bilinear interpolation on the high-frequency components and bicubic interpolation on the low-frequency components, and finally perform wavelet inverse transform to obtain the preprocessed image. We adopt the improved Vision-Transformer structure as the semantic coder to extract and transmit the semantic information of remote sensing images. The Vision-Transformer structure can better train the huge data volume and extract better image semantic features, and adopt the multi-layer self-attention mechanism to better capture the correlation between semantic features and reduce redundant features. Secondly, to improve the coding efficiency, we reduce the quadratic complexity of the self-attentive mechanism itself to linear so as to improve the image data processing speed of the model. We conducted experimental simulations on the RSOD dataset and compared the designed system with a semantic communication system based on CNN and image coding methods such as BGP and JPEG to verify that the method can effectively alleviate the problem of excessive data volume and improve the performance of image data communication.

Keywords: semantic communication, transformer, wavelet transform, data processing

Procedia PDF Downloads 60

Authors: Ivy Rose Mathew

Abstract:

The purpose of this study is to outline the efficacy of technology and the opportunities it can bring to implement a successful delivery model in Distance Learning. Distance Learning has proliferated over the past few years across the world. Some of the current challenges faced by current students of distance education include lack of motivation, a sense of isolation and a need for greater and improved communication. Hence the author proposes a creative technology supported model for distance learning exactly mirrored on the traditional face to face learning that can be adopted by distance learning providers. This model suggests the usage of a range of technologies and social networking facilities, with the aim of creating a more engaging and sustaining learning environment to help overcome the isolation often noted by distance learners. While discussing the possibilities, the author also highlights the complexity and practical challenges of implementing such a model. Design/methodology/approach: Theoretical issues from previous research related to successful models for distance learning providers will be considered. And also the analysis of a case study from one of the largest private tertiary institution in Botswana, Botho University will be included. This case study illustrates important aspects of the distance learning delivery model and provides insights on how curriculum development is planned, quality assurance is done, and learner support is assured for successful distance learning experience. Research limitations/implications: While some of the aspects of this study may not be applicable to other contexts, a number of new providers of distance learning can adapt the key principles of this delivery model.

Keywords: distance learning, efficacy, learning experience, technology supported model

Procedia PDF Downloads 227

Authors: Golnush Masghati-Amoli, Paul Chin

Abstract:

Over the past recent years, with the rapid increases in data availability and computing power, Machine Learning (ML) techniques have been called on in a range of different industries for their strong predictive capability. However, the use of Machine Learning in commercial banking has been limited due to a special challenge imposed by numerous regulations that require lenders to be able to explain their analytic models, not only to regulators but often to consumers. In other words, although Machine Leaning techniques enable better prediction with a higher level of accuracy, in comparison with other industries, they are adopted less frequently in commercial banking especially for scoring purposes. This is due to the fact that Machine Learning techniques are often considered as a black box and fail to provide information on why a certain risk score is given to a customer. In order to bridge this gap between the explain-ability and performance of Machine Learning techniques, a Hybrid Model is developed at Dun and Bradstreet that is focused on blending Machine Learning algorithms with traditional approaches such as scorecards. The Hybrid Model maximizes efficiency of traditional scorecards by merging its practical benefits, such as explain-ability and the ability to input domain knowledge, with the deep insights of Machine Learning techniques which can uncover patterns scorecard approaches cannot. First, through development of Machine Learning models, engineered features and latent variables and feature interactions that demonstrate high information value in the prediction of customer risk are identified. Then, these features are employed to introduce observed non-linear relationships between the explanatory and dependent variables into traditional scorecards. Moreover, instead of directly computing the Weight of Evidence (WoE) from good and bad data points, the Hybrid Model tries to match the score distribution generated by a Machine Learning algorithm, which ends up providing an estimate of the WoE for each bin. This capability helps to build powerful scorecards with sparse cases that cannot be achieved with traditional approaches. The proposed Hybrid Model is tested on different portfolios where a significant gap is observed between the performance of traditional scorecards and Machine Learning models. The result of analysis shows that Hybrid Model can improve the performance of traditional scorecards by introducing non-linear relationships between explanatory and target variables from Machine Learning models into traditional scorecards. Also, it is observed that in some scenarios the Hybrid Model can be almost as predictive as the Machine Learning techniques while being as transparent as traditional scorecards. Therefore, it is concluded that, with the use of Hybrid Model, Machine Learning algorithms can be used in the commercial banking industry without being concerned with difficulties in explaining the models for regulatory purposes.

Keywords: machine learning algorithms, scorecard, commercial banking, consumer risk, feature engineering

Procedia PDF Downloads 117

Authors: Atakan Aral Ormancı, Tuğçe Arslantaş, Murat Özcü

Abstract:

This study presents the design and implementation of an experimental chassis-level system for various control applications. Specifically, the height level of the chassis is controlled using proportional integrated derivative, fuzzy logic, and deep learning control methods. Real-time data obtained from height and pressure sensors installed in a 6x2 truck chassis, in combination with pulse-width modulation signal values, are utilized during the tests. A prototype pneumatic system of a 6x2 truck is added to the setup, which enables the Smart Pneumatic Actuators to function as if they were in a real-world setting. To obtain real-time signal data from height sensors, an Arduino Nano is utilized, while a Raspberry Pi processes the data using Matlab/Simulink and provides the correct output signals to control the Smart Pneumatic Actuator in the truck chassis. The objective of this research is to optimize the time it takes for the chassis to level down and up under various loads. To achieve this, proportional integrated derivative control, fuzzy logic control, and deep learning techniques are applied to the system. The results show that the deep learning method is superior in optimizing time for a non-linear system. Fuzzy logic control with a triangular membership function as the rule base achieves better outcomes than proportional integrated derivative control. Traditional proportional integrated derivative control improves the time it takes to level the chassis down and up compared to an uncontrolled system. The findings highlight the superiority of deep learning techniques in optimizing the time for a non-linear system, and the potential of fuzzy logic control. The proposed approach and the experimental results provide a valuable contribution to the field of control, automation, and systems engineering.

Keywords: automotive, chassis level control, control systems, pneumatic system control

Procedia PDF Downloads 60

Authors: Yayun Hsu, Henry Horng-Shing Lu

Abstract:

This paper proposes a data-driven, biology-inspired neural segmentation method of 3D drosophila Brainbow images. We use Bayesian Sequential Partitioning algorithm for probabilistic modeling, which can be used to detect somas and to eliminate cross talk effects. This work attempts to develop an automatic methodology for neuron image segmentation, which nowadays still lacks a complete solution due to the complexity of the image. The proposed method does not need any predetermined, risk-prone thresholds since biological information is inherently included in the image processing procedure. Therefore, it is less sensitive to variations in neuron morphology; meanwhile, its flexibility would be beneficial for tracing the intertwining structure of neurons.

Keywords: brainbow, 3D imaging, image segmentation, neuron morphology, biological data mining, non-parametric learning

Procedia PDF Downloads 467

Authors: Hilal Naimi, Amelbahahouda Adamou-Mitiche, Lahcene Mitiche

Abstract:

The main problem in the area of medical imaging has been image denoising. The most defying for image denoising is to secure data carrying structures like surfaces and edges in order to achieve good visual quality. Different algorithms with different denoising performances have been proposed in previous decades. More recently, models focused on deep learning have shown a great promise to outperform all traditional approaches. However, these techniques are limited to the necessity of large sample size training and high computational costs. This research proposes a denoising approach basing on LDTCWT (Lifting Dual Tree Complex Wavelet Transform) using Hybrid Thresholding with Wiener filter to enhance the quality image. This research describes the LDTCWT as a type of lifting wavelets remodeling that produce complex coefficients by employing a dual tree of lifting wavelets filters to get its real part and imaginary part. Permits the remodel to produce approximate shift invariance, directionally selective filters and reduces the computation time (properties lacking within the classical wavelets transform). To develop this approach, a hybrid thresholding function is modeled by integrating the Wiener filter into the thresholding function.

Keywords: lifting wavelet transform, image denoising, dual tree complex wavelet transform, wavelet shrinkage, wiener filter

Procedia PDF Downloads 140

Authors: Selim M. Khan

Abstract:

Indoor air quality is strongly influenced by the presence of radioactive radon (222Rn) gas. Indeed, exposure to high 222Rn concentrations is unequivocally linked to DNA damage and lung cancer and is a worsening issue in North American and European built environments, having increased over time within newer housing stocks as a function of as yet unclear variables. Indoor air radon concentration can be influenced by a wide range of environmental, structural, and behavioral factors. As some of these factors are quantitative while others are qualitative, no single statistical model can determine indoor radon level precisely while simultaneously considering all these variables across a complex and highly diverse dataset. The ability of AI- machine (deep) learning to simultaneously analyze multiple quantitative and qualitative features makes it suitable to predict radon with a high degree of precision. Using Canadian and Swedish long-term indoor air radon exposure data, we are using artificial deep neural network models with random weights and polynomial statistical models in MATLAB to assess and predict radon health risk to human as a function of geospatial, human behavioral, and built environmental metrics. Our initial artificial neural network with random weights model run by sigmoid activation tested different combinations of variables and showed the highest prediction accuracy (>96%) within the reasonable iterations. Here, we present details of these emerging methods and discuss strengths and weaknesses compared to the traditional artificial neural network and statistical methods commonly used to predict indoor air quality in different countries. We propose an artificial deep neural network with random weights as a highly effective method for assessing and predicting indoor radon.

Keywords: radon, radiation protection, lung cancer, aI-machine deep learnng, risk assessment, risk prediction, Europe, North America

Procedia PDF Downloads 80

Authors: Mohammed Alhawiti

Abstract:

Assessment of Learning Management Systems has been of less importance than its due share. This paper investigates the evaluation of learning management systems (LMS) within educational setting as both an online learning system as well as a helpful tool for multidisciplinary learning environment. This study suggests a theoretical e-learning evaluation model, studying a multi-dimensional methods for evaluation through LMS system, service and content quality, learner`s perspective and attitudes of the instructor. A survey was conducted among 105 e-learners. The sample consisted of students at both undergraduate and master’s levels. Content validity, reliability were tested through the instrument, Findings suggested the suitability of the proposed model in evaluation for the satisfaction of learners through LMS. The results of this study would be valuable for both instructors and users of e-learning systems.

Keywords: e-learning, LMS, higher education, management systems

Procedia PDF Downloads 383

Authors: Ulas Vural, M. Ergun Okay, E. Mesut Yildiz

Abstract:

Telecommunication service providers demand accurate and precise prediction of customer churn probabilities to increase the effectiveness of their customer relation services. The large amount of customer data owned by the service providers is suitable for analysis by machine learning methods. In this study, expenditure data of customers are analyzed by using an artificial neural network (ANN). The ANN model is applied to the data of customers with different billing duration. The proposed model successfully predicts the churn probabilities at 83% accuracy for only three months expenditure data and the prediction accuracy increases up to 89% when the nine month data is used. The experiments also show that the accuracy of ANN model increases on an extended feature set with information of the changes on the bill amounts.

Keywords: customer relationship management, churn prediction, telecom industry, deep learning, artificial neural networks

Procedia PDF Downloads 125

Authors: Bo-Bin Xiaoa, Chia-Yi Liua

Abstract:

In the task of searching for anomalous radiation sources, personnel holding radiation detectors to search for radiation sources may be exposed to unnecessary radiation risk, and automated search using machines becomes a required project. The research uses various sophisticated algorithms, which are double Q learning, dueling network, and NoisyNet, of deep reinforcement learning to search for radiation sources. The simulation environment, which is a 10*10 grid and one shielding wall setting in it, improves the development of the AI model by training 1 million episodes. In each episode of training, the radiation source position, the radiation source intensity, agent position, shielding wall position, and shielding wall length are all set randomly. The three algorithms are applied to run AI model training in four environments where the training shielding wall is a full-shielding wall, a lead wall, a concrete wall, and a lead wall or a concrete wall appearing randomly. The 12 best performance AI models are selected by observing the reward value during the training period and are evaluated by comparing these AI models with the gradient search algorithm. The results show that the performance of the AI model, no matter which one algorithm, is far better than the gradient search algorithm. In addition, the simulation environment becomes more complex, the AI model which applied Double DQN combined Dueling and NosiyNet algorithm performs better.

Keywords: double Q learning, dueling network, NoisyNet, source searching

Procedia PDF Downloads 90

Authors: Fatemeh Keivanifard

Abstract:

E-learning is becoming increasingly prominent in higher education, with universities increasing provision and more students signing up. This paper examines factors that predict students' attitudes to adapt e-learning at the Khuzestan province Iran. Understanding the nature of these factors may assist these universities in promoting the use of information and communication technology in teaching and learning. The main focus of the paper is on the university students, whose decision supports effective implementation of e-learning. Data was collected through a survey of 300 post graduate students at the University of dezful, shooshtar and chamran in Khuzestan. The technology adoption model put forward by Davis is utilized in this study. Two more independent variables are added to the original model, namely, the pressure to act and resources availability. The results show that there are five factors that can be used in modeling students' attitudes to adapt e-learning. These factors are intention toward e-learning, perceived usefulness of e-learning, perceived ease of e-learning use, pressure to use e-learning, and the availability of resources needed to use e-learning.

Keywords: e-learning, intention, ease of use, pressure to use, usefulness

Procedia PDF Downloads 345

Authors: Somojit Saha, Rohit K. Chatterjee, Sarit K. Das, Avijit Kar

Abstract:

A new force field is designed for propagation of the parametric contour into deep narrow cortical fold in the application of knowledge based reconstruction of cerebral cortex from MR image of brain. Designing of this force field is highly inspired by the Generalized Gradient Vector Flow (GGVF) model and markedly differs in manipulation of image information in order to determine the direction of propagation of the contour. While GGVF uses edge map as its main driving force, the newly designed force field uses the map of distance between zero valued pixels and their nearest non-zero valued pixel as its main driving force. Hence, it is called Zero-Non-Zero Distance (ZNZD) force field. The objective of this force field is forceful propagation of the contour beyond spurious convergence due to partial volume effect (PVE) in to narrow sulcal fold. Being function of the corresponding non-zero pixel value, the force field has got an inherent property to determine spuriousness of the edge automatically. It is effectively applied along with some morphological processing in the application of cortical reconstruction to breach the hindrance of PVE in narrow sulci where conventional GGVF fails.

Keywords: deformable model, external force field, partial volume effect, cortical reconstruction, MR image of brain

Procedia PDF Downloads 375

Authors: Adeep Hande, Shubham Agarwal

Abstract:

This paper presents a study on identifying sexism in online texts using various state-of-the-art deep learning models based on BERT. We experimented with different feature sets and model architectures and evaluated their performance using precision, recall, F1 score, and accuracy metrics. We also explored the use of pseudolabeling technique to improve model performance. Our experiments show that the best-performing models were based on BERT, and their multilingual model achieved an F1 score of 0.83. Furthermore, the use of pseudolabeling significantly improved the performance of the BERT-based models, with the best results achieved using the pseudolabeling technique. Our findings suggest that BERT-based models with pseudolabeling hold great promise for identifying sexism in online texts with high accuracy.

Keywords: large language models, semi-supervised learning, sexism detection, data sparsity

Procedia PDF Downloads 49

Authors: Qianwei Cheng, A. K. M. Mahbubur Rahman, Anis Sarker, Abu Bakar Siddik Nayem, Ovi Paul, Amin Ahsan Ali, M. Ashraful Amin, Ryosuke Shibasaki, Moinul Zaber

Abstract:

Thomas Friedman, in his famous book, argued that the world in this 21st century is flat and will continue to be flatter. This is attributed to rapid globalization and the interdependence of humanity that engendered tremendous in-flow of human migration towards the urban spaces. In order to keep the urban environment sustainable, policy makers need to plan based on extensive analysis of the urban environment. With the advent of high definition satellite images, high resolution data, computational methods such as deep neural network analysis, and hardware capable of high-speed analysis; urban planning is seeing a paradigm shift. Legacy data on urban environments are now being complemented with high-volume, high-frequency data. However, the first step of understanding urban space lies in useful categorization of the space that is usable for data collection, analysis, and visualization. In this paper, we propose a pragmatic categorization method that is readily usable for machine analysis and show applicability of the methodology on a developing world setting. Categorization to plan sustainable urban spaces should encompass the buildings and their surroundings. However, the state-of-the-art is mostly dominated by classification of building structures, building types, etc. and largely represents the developed world. Hence, these methods and models are not sufficient for developing countries such as Bangladesh, where the surrounding environment is crucial for the categorization. Moreover, these categorizations propose small-scale classifications, which give limited information, have poor scalability and are slow to compute in real time. Our proposed method is divided into two steps-categorization and automation. We categorize the urban area in terms of informal and formal spaces and take the surrounding environment into account. 50 km × 50 km Google Earth image of Dhaka, Bangladesh was visually annotated and categorized by an expert and consequently a map was drawn. The categorization is based broadly on two dimensions-the state of urbanization and the architectural form of urban environment. Consequently, the urban space is divided into four categories: 1) highly informal area; 2) moderately informal area; 3) moderately formal area; and 4) highly formal area. In total, sixteen sub-categories were identified. For semantic segmentation and automatic categorization, Google’s DeeplabV3+ model was used. The model uses Atrous convolution operation to analyze different layers of texture and shape. This allows us to enlarge the field of view of the filters to incorporate larger context. Image encompassing 70% of the urban space was used to train the model, and the remaining 30% was used for testing and validation. The model is able to segment with 75% accuracy and 60% Mean Intersection over Union (mIoU). In this paper, we propose a pragmatic categorization method that is readily applicable for automatic use in both developing and developed world context. The method can be augmented for real-time socio-economic comparative analysis among cities. It can be an essential tool for the policy makers to plan future sustainable urban spaces.

Keywords: semantic segmentation, urban environment, deep learning, urban building, classification

Procedia PDF Downloads 160

Authors: Joseph Lillington, Tom Gout, Mike Harrison, Ian Farnan

Abstract:

The vitrification of high-level nuclear waste within borosilicate glass and its incorporation within a multi-barrier repository deep underground is widely accepted as the preferred disposal method. However, for this to happen, any safety case will require validation that the initially localized radionuclides will not be considerably released into the near/far-field. Therefore, accurate mechanistic models are necessary to predict glass dissolution, and these should be robust to a variety of incorporated waste species and leaching test conditions, particularly given substantial variations across international waste-streams. Here, machine learning is used to predict glass material properties (viscosity, density) and glass leaching model parameters from large-scale industrial data. A variety of different machine learning algorithms have been compared to assess performance. Density was predicted solely from composition, whereas viscosity additionally considered temperature. To predict suitable glass leaching model parameters, a large simulated dataset was created by coupling MATLAB and the chemical reactive-transport code HYTEC, considering the state-of-the-art GRAAL model (glass reactivity in allowance of the alteration layer). The trained models were then subsequently applied to the large-scale industrial, experimental data to identify potentially appropriate model parameters. Results indicate that ensemble methods can accurately predict viscosity as a function of temperature and composition across all three industrial datasets. Glass density prediction shows reliable learning performance with predictions primarily being within the experimental uncertainty of the test data. Furthermore, machine learning can predict glass dissolution model parameters behavior, demonstrating potential value in GRAAL model development and in assessing suitable model parameters for large-scale industrial glass dissolution data.

Keywords: machine learning, predictive modelling, pattern recognition, radioactive waste glass

Procedia PDF Downloads 98

Authors: Sumitra Nuanmeesri

Abstract:

The research focus on study, analyze and design the model of the infrastructure and computer networks for distance education, online learning via new media, e-learning and blended learning. The collected information from study and analyze process that information was evaluated by the index of item objective congruence (IOC) by 9 specialists to design model. The results of evaluate the model with the mean and standard deviation by the sample of 9 specialists value is 3.85. The results showed that the infrastructure and computer networks are designed to be appropriate to a great extent appropriate to a great extent.

Keywords: blended learning, new media, infrastructure and computer network, tele-education, online learning

Procedia PDF Downloads 386

Authors: Hameed Olalekan Bolaji

Abstract:

The digital natives are driving the trends of literacy in the use of electronic devices for learning purposes. This has reconfigured the context of learning in the exploration of knowledge in a social learning environment. This study explores the impact of Ubiquitous Collaborative Mobile Learning (UCML) instructional design model in a quantitative designed-based research approach. The UCML model was a synergetic blend of four models that are relevant to the design of instructional content for a social learning environment. The UCML model serves as the treatment and instructions were transmitted via mobile device based on the principle of ‘bring your own device’ (BYOD) to promote social learning. Three research questions and two hypotheses were raised to guide the conduct of this study. A researcher-designed questionnaire was used to collate data and the it was subjected to reliability of Cronbach Alpha which yielded 0.91. Descriptive statistics of mean and standard deviation were used to answer research questions while inferential statistics of independent sample t-test was used to analyze the hypotheses. The findings reveal that the UCML model was adequately evolved and it promotes social learning its design principles through the use of mobile devices.

Keywords: collaboration, mobile device, social learning, ubiquitous

Procedia PDF Downloads 130

Authors: Shannon Steinmetz, Ellen Gethner

Abstract:

In this paper, a theoretical foundation is developed for patterned segmentation of audio using the geometry of music and statistical manifold. We demonstrate image content clustering using conic space sonification. The algorithm takes a geodesic curve as a model estimator of the three-parameter Gamma distribution. The random variable is parameterized by musical centricity and centric velocity. Model parameters predict audio segmentation in the form of duration and frame count based on the likelihood of musical geometry transition. We provide an example using a database of randomly selected images, resulting in statistically significant clusters of similar image content.

Keywords: sonification, musical information geometry, image, content extraction, automated quantification, audio segmentation, pattern recognition

Procedia PDF Downloads 202

Authors: M. Ramesh Kanthan, S. Naga Nandini Sujatha

Abstract:

The goal of haze removal algorithms is to enhance and recover details of scene from foggy image. In enhancement the proposed method focus into two main categories: (i) image enhancement based on Adaptive contrast Histogram equalization, and (ii) image edge strengthened Gradient model. Many circumstances accurate haze removal algorithms are needed. The de-fog feature works through a complex algorithm which first determines the fog destiny of the scene, then analyses the obscured image before applying contrast and sharpness adjustments to the video in real-time to produce image the fusion strategy is driven by the intrinsic properties of the original image and is highly dependent on the choice of the inputs and the weights. Then the output haze free image has reconstructed using fusion methodology. In order to increase the accuracy, interpolation method has used in the output reconstruction. A promising retrieval performance is achieved especially in particular examples.

Keywords: single image, fusion, dehazing, multi-scale fusion, per-pixel, weight map

Procedia PDF Downloads 445

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 288

Authors: Zhaoxin Luo, Michael Zhu

Abstract:

In natural languages, there are always complex semantic hierarchies. Obtaining the feature representation based on these complex semantic hierarchies becomes the key to the success of the model. Several RNN models have recently been proposed to use latent indicators to obtain the hierarchical structure of documents. However, the model that only uses a single-layer latent indicator cannot achieve the true hierarchical structure of the language, especially a complex language like Chinese. In this paper, we propose a deep layered model that stacks arbitrarily many RNN layers equipped with latent indicators. After using EM and training it hierarchically, our model solves the computational problem of stacking RNN layers and makes it possible to stack arbitrarily many RNN layers. Our deep hierarchical model not only achieves comparable results to large pre-trained models on the Chinese short text classification problem but also achieves state of art results on the Chinese long text classification problem.

Keywords: nature language processing, recurrent neural network, hierarchical structure, document classification, Chinese

Procedia PDF Downloads 46

Authors: Arpit Rai

Abstract:

In this work, we propose PatchMix, a semi-supervised method for pre-training visual representations. PatchMix mixes patches of two images and then solves an auxiliary task of predicting the label of each patch in the mixed image. Our experiments on the CIFAR-10, 100 and the SVHN dataset show that the representations learned by this method encodes useful information for transfer to new tasks and outperform the baseline Residual Network encoders by on CIFAR 10 by 12% on ResNet 101 and 2% on ResNet-56, by 4% on CIFAR-100 on ResNet101 and by 6% on SVHN dataset on the ResNet-101 baseline model.

Keywords: self-supervised learning, representation learning, computer vision, generalization

Procedia PDF Downloads 68

Authors: Payel Datta, Abhishek Das, Abhishek Roychoudhury, Dhiman Chattopadhyay, Tanushyam Chattopadhyay

Abstract:

Machine learning (ML) and deep learning (DL) is most predominantly used in image/video processing, natural language processing (NLP), audio and speech recognition but not that much used in system performance evaluation. In this paper, authors are going to describe the architecture of an abstraction layer constructed using ML/DL to detect the system failure. This proposed system is used to detect the system failure by evaluating the performance metrics of an IoT service deployment under constrained infrastructure environment. This system has been tested on the manually annotated data set containing different metrics of the system, like number of threads, throughput, average response time, CPU usage, memory usage, network input/output captured in different hardware environments like edge (atom based gateway) and cloud (AWS EC2). The main challenge of developing such system is that the accuracy of classification should be 100% as the error in the system has an impact on the degradation of the service performance and thus consequently affect the reliability and high availability which is mandatory for an IoT system. Proposed ML/DL classifiers work with 100% accuracy for the data set of nearly 4,000 samples captured within the organization.

Keywords: machine learning, system performance, performance metrics, IoT, edge

Procedia PDF Downloads 177

Authors: Vikrant Gupta, Amrit Goswami

Abstract:

The fixed income market forms the basis of the modern financial market. All other assets in financial markets derive their value from the bond market. Owing to its over-the-counter nature, corporate bonds have relatively less data publicly available and thus is researched upon far less compared to Equities. Bond price prediction is a complex financial time series forecasting problem and is considered very crucial in the domain of finance. The bond prices are highly volatile and full of noise which makes it very difficult for traditional statistical time-series models to capture the complexity in series patterns which leads to inefficient forecasts. To overcome the inefficiencies of statistical models, various machine learning techniques were initially used in the literature for more accurate forecasting of time-series. However, simple machine learning methods such as linear regression, support vectors, random forests fail to provide efficient results when tested on highly complex sequences such as stock prices and bond prices. hence to capture these intricate sequence patterns, various deep learning-based methodologies have been discussed in the literature. In this study, a recurrent neural network-based deep learning model using long short term networks for prediction of corporate bond prices has been discussed. Long Short Term networks (LSTM) have been widely used in the literature for various sequence learning tasks in various domains such as machine translation, speech recognition, etc. In recent years, various studies have discussed the effectiveness of LSTMs in forecasting complex time-series sequences and have shown promising results when compared to other methodologies. LSTMs are a special kind of recurrent neural networks which are capable of learning long term dependencies due to its memory function which traditional neural networks fail to capture. In this study, a simple LSTM, Stacked LSTM and a Masked LSTM based model has been discussed with respect to varying input sequences (three days, seven days and 14 days). In order to facilitate faster learning and to gradually decompose the complexity of bond price sequence, an Empirical Mode Decomposition (EMD) has been used, which has resulted in accuracy improvement of the standalone LSTM model. With a variety of Technical Indicators and EMD decomposed time series, Masked LSTM outperformed the other two counterparts in terms of prediction accuracy. To benchmark the proposed model, the results have been compared with traditional time series models (ARIMA), shallow neural networks and above discussed three different LSTM models. In summary, our results show that the use of LSTM models provide more accurate results and should be explored more within the asset management industry.

Keywords: bond prices, long short-term memory, time series forecasting, empirical mode decomposition

Procedia PDF Downloads 117

Authors: Bharatendra Rai

Abstract:

The sequence of words in text data has long-term dependencies and is known to suffer from vanishing gradient problems when developing deep learning models. Although recurrent networks such as long short-term memory networks help to overcome this problem, achieving high text classification performance is a challenging problem. Convolutional recurrent networks that combine the advantages of long short-term memory networks and convolutional neural networks can be useful for text classification performance improvements. However, arriving at suitable hyperparameter values for convolutional recurrent networks is still a challenging task where fitting a model requires significant computing resources. This paper illustrates the advantages of using convolutional recurrent networks for text classification with the help of statistically planned computer experiments for hyperparameter tuning.

Keywords: long short-term memory networks, convolutional recurrent networks, text classification, hyperparameter tuning, Tukey honest significant differences

Procedia PDF Downloads 99

Authors: Mohammed A. Zidan, Alaa Sagheer, Nasser Metwally

Abstract:

Real-time learning is an important goal that most of artificial intelligence researches try to achieve it. There are a lot of problems and applications which require low cost learning such as learn a robot to be able to classify and recognize patterns in real time and real-time recall. In this contribution, we suggest a model of quantum competitive learning based on a series of quantum gates and additional operator. The proposed model enables to recognize any incomplete patterns, where we can increase the probability of recognizing the pattern at the expense of the undesired ones. Moreover, these undesired ones could be utilized as new patterns for the system. The proposed model is much better compared with classical approaches and more powerful than the current quantum competitive learning approaches.

Keywords: competitive learning, quantum gates, quantum gates, winner-take-all

Procedia PDF Downloads 444

Authors: Wei Zhang

Abstract:

With the rapid development of deep learning, neural network and deep learning algorithms play a significant role in various practical applications. Due to the high accuracy and good performance, Convolutional Neural Networks (CNNs) especially have become a research hot spot in the past few years. However, the size of the networks becomes increasingly large scale due to the demands of the practical applications, which poses a significant challenge to construct a high-performance implementation of deep learning neural networks. Meanwhile, many of these application scenarios also have strict requirements on the performance and low-power consumption of hardware devices. Therefore, it is particularly critical to choose a moderate computing platform for hardware acceleration of CNNs. This article aimed to survey the recent advance in Field Programmable Gate Array (FPGA)-based acceleration of CNNs. Various designs and implementations of the accelerator based on FPGA under different devices and network models are overviewed, and the versions of Graphic Processing Units (GPUs), Application Specific Integrated Circuits (ASICs) and Digital Signal Processors (DSPs) are compared to present our own critical analysis and comments. Finally, we give a discussion on different perspectives of these acceleration and optimization methods on FPGA platforms to further explore the opportunities and challenges for future research. More helpfully, we give a prospect for future development of the FPGA-based accelerator.

Keywords: deep learning, field programmable gate array, FPGA, hardware accelerator, convolutional neural networks, CNN

Procedia PDF Downloads 103

Authors: Okike Benjamin, Garba E J. D.

Abstract:

Encryption is used to conceal information from prying eyes. Presently, information and data encryption are common due to the volume of data and information in transit across the globe on daily basis. Image encryption is yet to receive the attention of the researchers as deserved. In other words, video and multimedia documents are exposed to unauthorized accessors. The authors propose image encryption using matrix transpose. An algorithm that would allow image encryption is developed. In this proposed image encryption technique, the image to be encrypted is split into parts based on the image size. Each part is encrypted separately using matrix transpose. The actual encryption is on the picture elements (pixel) that make up the image. After encrypting each part of the image, the positions of the encrypted images are swapped before transmission of the image can take place. Swapping the positions of the images is carried out to make the encrypted image more robust for any cryptanalyst to decrypt.

Keywords: image encryption, matrices, pixel, matrix transpose

Procedia PDF Downloads 400

Authors: Junling Li, Fang Meng, Yichun Zhang

Abstract:

Image saliency detection has been long studied, while several challenging problems are still unsolved, such as detecting saliency inaccurately in complex scenes or suppressing salient objects in the image borders. In this paper, we propose a new saliency detection algorithm in order to solving these problems. We represent the image as a graph with superixels as nodes. By considering appearance similarity between the boundary and the background, the proposed method chooses non-saliency boundary nodes as background priors to construct the background probability model. The probability that each node belongs to the model is computed, which measures its similarity with backgrounds. Thus we can calculate saliency by the transformed probability as a metric. We compare our algorithm with ten-state-of-the-art salient detection methods on the public database. Experimental results show that our simple and effective approach can attack those challenging problems that had been baffling in image saliency detection.

Keywords: visual saliency, background probability, boundary knowledge, background priors

Procedia PDF Downloads 407

Authors: Nicholas V. Scott, Jack McCarthy

Abstract:

Military, law enforcement, and counter terrorism organizations are often tasked with target detection and image characterization of scenes containing explosive materials in various types of environments where light scattering intensity is high. Mitigation of this photonic noise using classical digital filtration and signal processing can be difficult. This is partially due to the lack of robust image processing methods for photonic noise removal, which strongly influence high resolution target detection and machine learning-based pattern recognition. Such analysis is crucial to the delivery of reliable intelligence. Polarization filters are a possible method for ambient glare reduction by allowing only certain modes of the electromagnetic field to be captured, providing strong scene contrast. An experiment was carried out utilizing a polarization lens attached to a hyperspectral imagery camera for the purpose of exploring the degree to which an imaged polarized scene of potassium, phosphorus, and nitrogen mixture allows for improved target detection and image segmentation. Preliminary imagery results based on the application of machine learning algorithms, including competitive leaky learning and distance metric analysis, to polarized hyperspectral imagery, suggest that polarization filters provide a slight advantage in image segmentation. The results of this work have implications for understanding the presence of explosive material in dry, desert areas where reflective glare is a significant impediment to scene characterization.

Keywords: explosive material, hyperspectral imagery, image segmentation, machine learning, polarization

Procedia PDF Downloads 121