Search results for: EGY-BCD dataset
Commenced in January 2007
Frequency: Monthly
Edition: International
Paper Count: 1095

Search results for: EGY-BCD dataset

675 FLEX: A Backdoor Detection and Elimination Method in Federated Scenario

Authors: Shuqi Zhang

Abstract:

Federated learning allows users to participate in collaborative model training without sending data to third-party servers, reducing the risk of user data privacy leakage, and is widely used in smart finance and smart healthcare. However, the distributed architecture design of federation learning itself and the existence of secure aggregation protocols make it inherently vulnerable to backdoor attacks. To solve this problem, the federated learning backdoor defense framework FLEX based on group aggregation, cluster analysis, and neuron pruning is proposed, and inter-compatibility with secure aggregation protocols is achieved. The good performance of FLEX is verified by building a horizontal federated learning framework on the CIFAR-10 dataset for experiments, which achieves 98% success rate of backdoor detection and reduces the success rate of backdoor tasks to 0% ~ 10%.

Keywords: federated learning, secure aggregation, backdoor attack, cluster analysis, neuron pruning

Procedia PDF Downloads 73
674 Generating Music with More Refined Emotions

Authors: Shao-Di Feng, Von-Wun Soo

Abstract:

To generate symbolic music with specific emotions is a challenging task due to symbolic music datasets that have emotion labels are scarce and incomplete. This research aims to generate more refined emotions based on the training datasets that are only labeled with four quadrants in Russel’s 2D emotion model. We focus on the theory of Music Fadernet and map arousal and valence to the low-level attributes, and build a symbolic music generation model by combining transformer and GM-VAE. We adopt an in-attention mechanism for the model and improve it by allowing modulation by conditional information. And we show the music generation model could control the generation of music according to the emotions specified by users in terms of high-level linguistic expression and by manipulating their corresponding low-level musical attributes. Finally, we evaluate the model performance using a pre-trained emotion classifier against a pop piano midi dataset called EMOPIA, and by subjective listening evaluation, we demonstrate that the model could generate music with more refined emotions correctly.

Keywords: music generation, music emotion controlling, deep learning, semi-supervised learning

Procedia PDF Downloads 63
673 Determining Antecedents of Employee Turnover: A Study on Blue Collar vs White Collar Workers on Marco Level

Authors: Evy Rombaut, Marie-Anne Guerry

Abstract:

Predicting voluntary turnover of employees is an important topic of study, both in academia and industry. Researchers try to uncover determinants for a broader understanding and possible prevention of turnover. In the current study, we use a data set based approach to reveal determinants for turnover, differing for blue and white collar workers. Our data set based approach made it possible to study actual turnover for more than 500000 employees in 15692 Belgian corporations. We use logistic regression to calculate individual turnover probabilities and test the goodness of our model with the AUC (area under the ROC-curve) method. The results of the study confirm the relationship of known determinants to employee turnover such as age, seniority, pay and work distance. In addition, the study unravels unknown and verifies known differences between blue and white collar workers. It shows opposite relationships to turnover for gender, marital status, the number of children, nationality, and pay.

Keywords: employee turnover, blue collar, white collar, dataset analysis

Procedia PDF Downloads 254
672 A Mutually Exclusive Task Generation Method Based on Data Augmentation

Authors: Haojie Wang, Xun Li, Rui Yin

Abstract:

In order to solve the memorization overfitting in the model-agnostic meta-learning MAML algorithm, a method of generating mutually exclusive tasks based on data augmentation is proposed. This method generates a mutex task by corresponding one feature of the data to multiple labels so that the generated mutex task is inconsistent with the data distribution in the initial dataset. Because generating mutex tasks for all data will produce a large number of invalid data and, in the worst case, lead to an exponential growth of computation, this paper also proposes a key data extraction method that only extract part of the data to generate the mutex task. The experiments show that the method of generating mutually exclusive tasks can effectively solve the memorization overfitting in the meta-learning MAML algorithm.

Keywords: mutex task generation, data augmentation, meta-learning, text classification.

Procedia PDF Downloads 114
671 Evolution under Length Constraints for Convolutional Neural Networks Architecture Design

Authors: Ousmane Youme, Jean Marie Dembele, Eugene Ezin, Christophe Cambier

Abstract:

In recent years, the convolutional neural networks (CNN) architectures designed by evolution algorithms have proven to be competitive with handcrafted architectures designed by experts. However, these algorithms need a lot of computational power, which is beyond the capabilities of most researchers and engineers. To overcome this problem, we propose an evolution architecture under length constraints. It consists of two algorithms: a search length strategy to find an optimal space and a search architecture strategy based on a genetic algorithm to find the best individual in the optimal space. Our algorithms drastically reduce resource costs and also keep good performance. On the Cifar-10 dataset, our framework presents outstanding performance with an error rate of 5.12% and only 4.6 GPU a day to converge to the optimal individual -22 GPU a day less than the lowest cost automatic evolutionary algorithm in the peer competition.

Keywords: CNN architecture, genetic algorithm, evolution algorithm, length constraints

Procedia PDF Downloads 106
670 Printed Thai Character Recognition Using Particle Swarm Optimization Algorithm

Authors: Phawin Sangsuvan, Chutimet Srinilta

Abstract:

This Paper presents the applications of Particle Swarm Optimization (PSO) Method for Thai optical character recognition (OCR). OCR consists of the pre-processing, character recognition and post-processing. Before enter into recognition process. The Character must be “Prepped” by pre-processing process. The PSO is an optimization method that belongs to the swarm intelligence family based on the imitation of social behavior patterns of animals. Route of each particle is determined by an individual data among neighborhood particles. The interaction of the particles with neighbors is the advantage of Particle Swarm to determine the best solution. So PSO is interested by a lot of researchers in many difficult problems including character recognition. As the previous this research used a Projection Histogram to extract printed digits features and defined the simple Fitness Function for PSO. The results reveal that PSO gives 67.73% for testing dataset. So in the future there can be explored enhancement the better performance of PSO with improve the Fitness Function.

Keywords: character recognition, histogram projection, particle swarm optimization, pattern recognition techniques

Procedia PDF Downloads 446
669 A Comparative Assessment Method For Map Alignment Techniques

Authors: Rema Daher, Theodor Chakhachiro, Daniel Asmar

Abstract:

In the era of autonomous robot mapping, assessing the goodness of the generated maps is important, and is usually performed by aligning them to ground truth. Map alignment is difficult for two reasons: first, the query maps can be significantly distorted from ground truth, and second, establishing what constitutes ground truth for different settings is challenging. Most map alignment techniques to this date have addressed the first problem, while paying too little importance to the second. In this paper, we propose a benchmark dataset, which consists of synthetically transformed maps with their corresponding displacement fields. Furthermore, we propose a new system for comparison, where the displacement field of any map alignment technique can be computed and compared to the ground truth using statistical measures. The local information in displacement fields renders the evaluation system applicable to any alignment technique, whether it is linear or not. In our experiments, the proposed method was applied to different alignment methods from the literature, allowing for a comparative assessment between them all.

Keywords: assessment methods, benchmark, image deformation, map alignment, robot mapping, robot motion

Procedia PDF Downloads 99
668 Data-Centric Anomaly Detection with Diffusion Models

Authors: Sheldon Liu, Gordon Wang, Lei Liu, Xuefeng Liu

Abstract:

Anomaly detection, also referred to as one-class classification, plays a crucial role in identifying product images that deviate from the expected distribution. This study introduces Data-centric Anomaly Detection with Diffusion Models (DCADDM), presenting a systematic strategy for data collection and further diversifying the data with image generation via diffusion models. The algorithm addresses data collection challenges in real-world scenarios and points toward data augmentation with the integration of generative AI capabilities. The paper explores the generation of normal images using diffusion models. The experiments demonstrate that with 30% of the original normal image size, modeling in an unsupervised setting with state-of-the-art approaches can achieve equivalent performances. With the addition of generated images via diffusion models (10% equivalence of the original dataset size), the proposed algorithm achieves better or equivalent anomaly localization performance.

Keywords: diffusion models, anomaly detection, data-centric, generative AI

Procedia PDF Downloads 63
667 Diversity and Phylogenetic Placement of Seven Inocybe (Inocybaceae, Fungi) from Benin

Authors: Hyppolite Aignon, Souleymane Yorou, Martin Ryberg, Anneli Svanholm

Abstract:

Climate change and human actions cause the extinction of wild mushrooms. In Benin, the diversity of fungi is large and may still contain species new to science but the inventory effort remains low and focuses on particularly edible species (Russula, Lactarius, Lactifluus, and also Amanita). In addition, inventories have started recently and some groups of fungi are not sufficiently sampled, however, the degradation of fungal habitat continues to increase and some species are already disappearing. (Yorou and De Kesel, 2011), however, the degradation of fungi habitat continues to increase and some species may disappear without being known. This genus (Inocybe) overlooked has a worldwide distribution and includes more than 700 species with many undiscovered or poorly known species worldwide and particularly in tropical Africa. It is therefore important to orient the inventory to other genera or important families such as Inocybe (Fungi, Agaricales) in order to highlight their diversity and also to know their phylogenetic positions with a combined approach of gene regions. This study aims to evaluate the species richness and phylogenetic position of Inocybe species and affiliated taxa in West Africa. Thus, in North Benin, we visited the Forest Reserve of Ouémé Supérieur, the Okpara forest and the Alibori Supérieur Forest Reserve. In the center, we targeted the Forest Reserve of Toui-Kilibo. The surveys have been carried during the raining season in the study area meaning from June to October. A total of 24 taxa were collected, photographed and described. The DNA was extracted, the Polymerase Chain Reaction was carried out using primers (ITS1-F, ITS4-B) for Internal transcribed spacer (ITS), (LROR, LWRB, LR7, LR5) for nuclear ribosomal (LSU), (RPB2-f5F, RPB2-b6F, RPB2- b6R2, RPB2-b7R) for RNA polymerase II gene (RPB2) and sequenced. The ITS sequences of the 24 collections of Inocybaceae were edited in Staden and all the sequences were aligned and edited with Aliview v1.17. The sequences were examined by eye for sufficient similarity to be considered the same species. 13 different species were present in the collections. In addition, sequences similar to the ITS sequences of the thirteen final species were searched using BLAST. The nLSU and RPB2 markers for these species have been inserted in a complete alignment, where species from all major Inocybaceae clades as well as from all continents except Antarctica are present. Our new sequences for nLSU and RPB2 have been manually aligned in this dataset. Phylogenetic analysis was performed using the RAxML v7.2.6 maximum likelihood software. Bootstrap replications have been set to 100 and no partitioning of the dataset has been performed. The resulting tree was viewed and edited with FigTree v1.4.3. The preliminary tree resulting from the analysis of maximum likelihood shows us that these species coming from Benin are much diversified and are distributed in four different clades (Inosperma, Inocybe, Mallocybe and Pseudosperma) on the seven clades of Inocybaceae but the phylogeny position of 7 is currently known. This study marks the diversity of Inocybe in Benin and the investigations will continue and a protection plan will be developed in the coming years.

Keywords: Benin, diversity, Inocybe, phylogeny placement

Procedia PDF Downloads 120
666 Investigating the Impacts of Climate Change on Soil Erosion: A Case Study of Kasilian Watershed, Northern Iran

Authors: Mohammad Zare, Mahbubeh Sheikh

Abstract:

Many of the impact of climate change will material through change in soil erosion which were rarely addressed in Iran. This paper presents an investigation of the impacts of climate change soil erosin for the Kasilian basin. LARS-WG5 was used to downscale the IPCM4 and GFCM21 predictions of the A2 scenarios for the projected periods of 1985-2030 and 2080-2099. This analysis was carried out by means of the dataset the International Centre for Theoretical Physics (ICTP) of Trieste. Soil loss modeling using Revised Universal Soil Loss Equation (RUSLE). Results indicate that soil erosion increase or decrease, depending on which climate scenarios are considered. The potential for climate change to increase soil loss rate, soil erosion in future periods was established, whereas considerable decreases in erosion are projected when land use is increased from baseline periods.

Keywords: Kasilian watershed, climatic change, soil erosion, LARS-WG5 Model, RUSLE

Procedia PDF Downloads 489
665 Clinical Validation of an Automated Natural Language Processing Algorithm for Finding COVID-19 Symptoms and Complications in Patient Notes

Authors: Karolina Wieczorek, Sophie Wiliams

Abstract:

Introduction: Patient data is often collected in Electronic Health Record Systems (EHR) for purposes such as providing care as well as reporting data. This information can be re-used to validate data models in clinical trials or in epidemiological studies. Manual validation of automated tools is vital to pick up errors in processing and to provide confidence in the output. Mentioning a disease in a discharge letter does not necessarily mean that a patient suffers from this disease. Many of them discuss a diagnostic process, different tests, or discuss whether a patient has a certain disease. The COVID-19 dataset in this study used natural language processing (NLP), an automated algorithm which extracts information related to COVID-19 symptoms, complications, and medications prescribed within the hospital. Free-text patient clinical patient notes are rich sources of information which contain patient data not captured in a structured form, hence the use of named entity recognition (NER) to capture additional information. Methods: Patient data (discharge summary letters) were exported and screened by an algorithm to pick up relevant terms related to COVID-19. Manual validation of automated tools is vital to pick up errors in processing and to provide confidence in the output. A list of 124 Systematized Nomenclature of Medicine (SNOMED) Clinical Terms has been provided in Excel with corresponding IDs. Two independent medical student researchers were provided with a dictionary of SNOMED list of terms to refer to when screening the notes. They worked on two separate datasets called "A” and "B”, respectively. Notes were screened to check if the correct term had been picked-up by the algorithm to ensure that negated terms were not picked up. Results: Its implementation in the hospital began on March 31, 2020, and the first EHR-derived extract was generated for use in an audit study on June 04, 2020. The dataset has contributed to large, priority clinical trials (including International Severe Acute Respiratory and Emerging Infection Consortium (ISARIC) by bulk upload to REDcap research databases) and local research and audit studies. Successful sharing of EHR-extracted datasets requires communicating the provenance and quality, including completeness and accuracy of this data. The results of the validation of the algorithm were the following: precision (0.907), recall (0.416), and F-score test (0.570). Percentage enhancement with NLP extracted terms compared to regular data extraction alone was low (0.3%) for relatively well-documented data such as previous medical history but higher (16.6%, 29.53%, 30.3%, 45.1%) for complications, presenting illness, chronic procedures, acute procedures respectively. Conclusions: This automated NLP algorithm is shown to be useful in facilitating patient data analysis and has the potential to be used in more large-scale clinical trials to assess potential study exclusion criteria for participants in the development of vaccines.

Keywords: automated, algorithm, NLP, COVID-19

Procedia PDF Downloads 80
664 Mobile Platform’s Attitude Determination Based on Smoothed GPS Code Data and Carrier-Phase Measurements

Authors: Mohamed Ramdani, Hassen Abdellaoui, Abdenour Boudrassen

Abstract:

Mobile platform’s attitude estimation approaches mainly based on combined positioning techniques and developed algorithms; which aim to reach a fast and accurate solution. In this work, we describe the design and the implementation of an attitude determination (AD) process, using only measurements from GPS sensors. The major issue is based on smoothed GPS code data using Hatch filter and raw carrier-phase measurements integrated into attitude algorithm based on vectors measurement using least squares (LSQ) estimation method. GPS dataset from a static experiment is used to investigate the effectiveness of the presented approach and consequently to check the accuracy of the attitude estimation algorithm. Attitude results from GPS multi-antenna over short baselines are introduced and analyzed. The 3D accuracy of estimated attitude parameters using smoothed measurements is over 0.27°.

Keywords: attitude determination, GPS code data smoothing, hatch filter, carrier-phase measurements, least-squares attitude estimation

Procedia PDF Downloads 138
663 3D Building Model Utilizing Airborne LiDAR Dataset and Terrestrial Photographic Images

Authors: J. Jasmee, I. Roslina, A. Mohammed Yaziz & A.H Juazer Rizal

Abstract:

The need of an effective building information collection method is vital to support a diversity of land development activities. At present, advances in remote sensing such as airborne LiDAR (Light Detection and Ranging) is an established technology for building information collection, location, and elevation of the reflecting laser points towards the construction of 3D building models. In this study, LiDAR datasets and terrestrial photographic images of buildings towards the construction of 3D building models is explored. It is found that, the quantitative accuracy of the constructed 3D building model, namely in the horizontal and vertical components were ± 0.31m (RMSEx,y) and ± 0.145m (RMSEz) respectively. The accuracies were computed based on sixty nine (69) horizontal and twenty (20) vertical surveyed points. As for the qualitative assessment, it is shown that the appearance of the 3D building model is adequate to support the requirements of LOD3 presentation based on the OGC (Open Geospatial Consortium) standard CityGML.

Keywords: LiDAR datasets, DSM, DTM, 3D building models

Procedia PDF Downloads 299
662 Facial Pose Classification Using Hilbert Space Filling Curve and Multidimensional Scaling

Authors: Mekamı Hayet, Bounoua Nacer, Benabderrahmane Sidahmed, Taleb Ahmed

Abstract:

Pose estimation is an important task in computer vision. Though the majority of the existing solutions provide good accuracy results, they are often overly complex and computationally expensive. In this perspective, we propose the use of dimensionality reduction techniques to address the problem of facial pose estimation. Firstly, a face image is converted into one-dimensional time series using Hilbert space filling curve, then the approach converts these time series data to a symbolic representation. Furthermore, a distance matrix is calculated between symbolic series of an input learning dataset of images, to generate classifiers of frontal vs. profile face pose. The proposed method is evaluated with three public datasets. Experimental results have shown that our approach is able to achieve a correct classification rate exceeding 97% with K-NN algorithm.

Keywords: machine learning, pattern recognition, facial pose classification, time series

Procedia PDF Downloads 327
661 Satellite Image Classification Using Firefly Algorithm

Authors: Paramjit Kaur, Harish Kundra

Abstract:

In the recent years, swarm intelligence based firefly algorithm has become a great focus for the researchers to solve the real time optimization problems. Here, firefly algorithm is used for the application of satellite image classification. For experimentation, Alwar area is considered to multiple land features like vegetation, barren, hilly, residential and water surface. Alwar dataset is considered with seven band satellite images. Firefly Algorithm is based on the attraction of less bright fireflies towards more brightener one. For the evaluation of proposed concept accuracy assessment parameters are calculated using error matrix. With the help of Error matrix, parameters of Kappa Coefficient, Overall Accuracy and feature wise accuracy parameters of user’s accuracy & producer’s accuracy can be calculated. Overall results are compared with BBO, PSO, Hybrid FPAB/BBO, Hybrid ACO/SOFM and Hybrid ACO/BBO based on the kappa coefficient and overall accuracy parameters.

Keywords: image classification, firefly algorithm, satellite image classification, terrain classification

Procedia PDF Downloads 375
660 A Meta Regression Analysis to Detect Price Premium Threshold for Eco-Labeled Seafood

Authors: Cristina Giosuè, Federica Biondo, Sergio Vitale

Abstract:

In the last years, the consumers' awareness for environmental concerns has been increasing, and seafood eco-labels are considered as a possible instrument to improve both seafood markets and sustainable fishing management. In this direction, the aim of this study was to carry out a meta-analysis on consumers’ willingness to pay (WTP) for eco-labeled wild seafood, by a meta-regression. Therefore, only papers published on ISI journals were searched on “Web of Knowledge” and “SciVerse Scopus” platforms, using the combinations of the following key words: seafood, ecolabel, eco-label, willingness, WTP and premium. The dataset was built considering: paper’s and survey’s codes, year of publication, first author’s nationality, species’ taxa and family, sample size, survey’s continent and country, data collection (where and how), gender and age of consumers, brand and ΔWTP. From analysis the interest on eco labeled seafood emerged clearly, in particular in developed countries. In general, consumers declared greater willingness to pay than that actually applied for eco-label products, with difference related to taxa and brand.

Keywords: eco label, meta regression, seafood, willingness to pay

Procedia PDF Downloads 100
659 Enhancing Financial Security: Real-Time Anomaly Detection in Financial Transactions Using Machine Learning

Authors: Ali Kazemi

Abstract:

The digital evolution of financial services, while offering unprecedented convenience and accessibility, has also escalated the vulnerabilities to fraudulent activities. In this study, we introduce a distinct approach to real-time anomaly detection in financial transactions, aiming to fortify the defenses of banking and financial institutions against such threats. Utilizing unsupervised machine learning algorithms, specifically autoencoders and isolation forests, our research focuses on identifying irregular patterns indicative of fraud within transactional data, thus enabling immediate action to prevent financial loss. The data we used in this study included the monetary value of each transaction. This is a crucial feature as fraudulent transactions may have distributions of different amounts than legitimate ones, such as timestamps indicating when transactions occurred. Analyzing transactions' temporal patterns can reveal anomalies (e.g., unusual activity in the middle of the night). Also, the sector or category of the merchant where the transaction occurred, such as retail, groceries, online services, etc. Specific categories may be more prone to fraud. Moreover, the type of payment used (e.g., credit, debit, online payment systems). Different payment methods have varying risk levels associated with fraud. This dataset, anonymized to ensure privacy, reflects a wide array of transactions typical of a global banking institution, ranging from small-scale retail purchases to large wire transfers, embodying the diverse nature of potentially fraudulent activities. By engineering features that capture the essence of transactions, including normalized amounts and encoded categorical variables, we tailor our data to enhance model sensitivity to anomalies. The autoencoder model leverages its reconstruction error mechanism to flag transactions that deviate significantly from the learned normal pattern, while the isolation forest identifies anomalies based on their susceptibility to isolation from the dataset's majority. Our experimental results, validated through techniques such as k-fold cross-validation, are evaluated using precision, recall, and the F1 score alongside the area under the receiver operating characteristic (ROC) curve. Our models achieved an F1 score of 0.85 and a ROC AUC of 0.93, indicating high accuracy in detecting fraudulent transactions without excessive false positives. This study contributes to the academic discourse on financial fraud detection and provides a practical framework for banking institutions seeking to implement real-time anomaly detection systems. By demonstrating the effectiveness of unsupervised learning techniques in a real-world context, our research offers a pathway to significantly reduce the incidence of financial fraud, thereby enhancing the security and trustworthiness of digital financial services.

Keywords: anomaly detection, financial fraud, machine learning, autoencoders, isolation forest, transactional data analysis

Procedia PDF Downloads 32
658 Hybrid Fuzzy Weighted K-Nearest Neighbor to Predict Hospital Readmission for Diabetic Patients

Authors: Soha A. Bahanshal, Byung G. Kim

Abstract:

Identification of patients at high risk for hospital readmission is of crucial importance for quality health care and cost reduction. Predicting hospital readmissions among diabetic patients has been of great interest to many researchers and health decision makers. We build a prediction model to predict hospital readmission for diabetic patients within 30 days of discharge. The core of the prediction model is a modified k Nearest Neighbor called Hybrid Fuzzy Weighted k Nearest Neighbor algorithm. The prediction is performed on a patient dataset which consists of more than 70,000 patients with 50 attributes. We applied data preprocessing using different techniques in order to handle data imbalance and to fuzzify the data to suit the prediction algorithm. The model so far achieved classification accuracy of 80% compared to other models that only use k Nearest Neighbor.

Keywords: machine learning, prediction, classification, hybrid fuzzy weighted k-nearest neighbor, diabetic hospital readmission

Procedia PDF Downloads 165
657 Multi-Sensor Target Tracking Using Ensemble Learning

Authors: Bhekisipho Twala, Mantepu Masetshaba, Ramapulana Nkoana

Abstract:

Multiple classifier systems combine several individual classifiers to deliver a final classification decision. However, an increasingly controversial question is whether such systems can outperform the single best classifier, and if so, what form of multiple classifiers system yields the most significant benefit. Also, multi-target tracking detection using multiple sensors is an important research field in mobile techniques and military applications. In this paper, several multiple classifiers systems are evaluated in terms of their ability to predict a system’s failure or success for multi-sensor target tracking tasks. The Bristol Eden project dataset is utilised for this task. Experimental and simulation results show that the human activity identification system can fulfill requirements of target tracking due to improved sensors classification performances with multiple classifier systems constructed using boosting achieving higher accuracy rates.

Keywords: single classifier, ensemble learning, multi-target tracking, multiple classifiers

Procedia PDF Downloads 238
656 A Comparative Study of Malware Detection Techniques Using Machine Learning Methods

Authors: Cristina Vatamanu, Doina Cosovan, Dragos Gavrilut, Henri Luchian

Abstract:

In the past few years, the amount of malicious software increased exponentially and, therefore, machine learning algorithms became instrumental in identifying clean and malware files through semi-automated classification. When working with very large datasets, the major challenge is to reach both a very high malware detection rate and a very low false positive rate. Another challenge is to minimize the time needed for the machine learning algorithm to do so. This paper presents a comparative study between different machine learning techniques such as linear classifiers, ensembles, decision trees or various hybrids thereof. The training dataset consists of approximately 2 million clean files and 200.000 infected files, which is a realistic quantitative mixture. The paper investigates the above mentioned methods with respect to both their performance (detection rate and false positive rate) and their practicability.

Keywords: ensembles, false positives, feature selection, one side class algorithm

Procedia PDF Downloads 264
655 A Neural Network Classifier for Identifying Duplicate Image Entries in Real-Estate Databases

Authors: Sergey Ermolin, Olga Ermolin

Abstract:

A Deep Convolution Neural Network with Triplet Loss is used to identify duplicate images in real-estate advertisements in the presence of image artifacts such as watermarking, cropping, hue/brightness adjustment, and others. The effects of batch normalization, spatial dropout, and various convergence methodologies on the resulting detection accuracy are discussed. For comparative Return-on-Investment study (per industry request), end-2-end performance is benchmarked on both Nvidia Titan GPUs and Intel’s Xeon CPUs. A new real-estate dataset from San Francisco Bay Area is used for this work. Sufficient duplicate detection accuracy is achieved to supplement other database-grounded methods of duplicate removal. The implemented method is used in a Proof-of-Concept project in the real-estate industry.

Keywords: visual recognition, convolutional neural networks, triplet loss, spatial batch normalization with dropout, duplicate removal, advertisement technologies, performance benchmarking

Procedia PDF Downloads 316
654 A Comparison between Artificial Neural Network Prediction Models for Coronal Hole Related High Speed Streams

Authors: Rehab Abdulmajed, Amr Hamada, Ahmed Elsaid, Hisashi Hayakawa, Ayman Mahrous

Abstract:

Solar emissions have a high impact on the Earth’s magnetic field, and the prediction of solar events is of high interest. Various techniques have been used in the prediction of solar wind using mathematical models, MHD models, and neural network (NN) models. This study investigates the coronal hole (CH) derived high-speed streams (HSSs) and their correlation to the CH area and create a neural network model to predict the HSSs. Two different algorithms were used to compare different models to find a model that best simulates the HSSs. A dataset of CH synoptic maps through Carrington rotations 1601 to 2185 along with Omni-data set solar wind speed averaged over the Carrington rotations is used, which covers Solar cycles (sc) 21, 22, 23, and most of 24.

Keywords: artificial neural network, coronal hole area, feed-forward neural network models, solar high speed streams

Procedia PDF Downloads 63
653 BigCrypt: A Probable Approach of Big Data Encryption to Protect Personal and Business Privacy

Authors: Abdullah Al Mamun, Talal Alkharobi

Abstract:

As data size is growing up, people are became more familiar to store big amount of secret information into cloud storage. Companies are always required to need transfer massive business files from one end to another. We are going to lose privacy if we transmit it as it is and continuing same scenario repeatedly without securing the communication mechanism means proper encryption. Although asymmetric key encryption solves the main problem of symmetric key encryption but it can only encrypt limited size of data which is inapplicable for large data encryption. In this paper we propose a probable approach of pretty good privacy for encrypt big data using both symmetric and asymmetric keys. Our goal is to achieve encrypt huge collection information and transmit it through a secure communication channel for committing the business and personal privacy. To justify our method an experimental dataset from three different platform is provided. We would like to show that our approach is working for massive size of various data efficiently and reliably.

Keywords: big data, cloud computing, cryptography, hadoop, public key

Procedia PDF Downloads 300
652 Long Short-Time Memory Neural Networks for Human Driving Behavior Modelling

Authors: Lu Zhao, Nadir Farhi, Yeltsin Valero, Zoi Christoforou, Nadia Haddadou

Abstract:

In this paper, a long short-term memory (LSTM) neural network model is proposed to replicate simultaneously car-following and lane-changing behaviors in road networks. By combining two kinds of LSTM layers and three input designs of the neural network, six variants of the LSTM model have been created. These models were trained and tested on the NGSIM 101 dataset, and the results were evaluated in terms of longitudinal speed and lateral position, respectively. Then, we compared the LSTM model with a classical car-following model (the intelligent driving model (IDM)) in the part of speed decision. In addition, the LSTM model is compared with a model using classical neural networks. After the comparison, the LSTM model demonstrates higher accuracy than the physical model IDM in terms of car-following behavior and displays better performance with regard to both car-following and lane-changing behavior compared to the classical neural network model.

Keywords: traffic modeling, neural networks, LSTM, car-following, lane-change

Procedia PDF Downloads 224
651 How Western Donors Allocate Official Development Assistance: New Evidence From a Natural Language Processing Approach

Authors: Daniel Benson, Yundan Gong, Hannah Kirk

Abstract:

Advancement in national language processing techniques has led to increased data processing speeds, and reduced the need for cumbersome, manual data processing that is often required when processing data from multilateral organizations for specific purposes. As such, using named entity recognition (NER) modeling and the Organisation of Economically Developed Countries (OECD) Creditor Reporting System database, we present the first geotagged dataset of OECD donor Official Development Assistance (ODA) projects on a global, subnational basis. Our resulting data contains 52,086 ODA projects geocoded to subnational locations across 115 countries, worth a combined $87.9bn. This represents the first global, OECD donor ODA project database with geocoded projects. We use this new data to revisit old questions of how ‘well’ donors allocate ODA to the developing world. This understanding is imperative for policymakers seeking to improve ODA effectiveness.

Keywords: international aid, geocoding, subnational data, natural language processing, machine learning

Procedia PDF Downloads 45
650 An Accurate Computer-Aided Diagnosis: CAD System for Diagnosis of Aortic Enlargement by Using Convolutional Neural Networks

Authors: Mahdi Bazarganigilani

Abstract:

Aortic enlargement, also known as an aortic aneurysm, can occur when the walls of the aorta become weak. This disease can become deadly if overlooked and undiagnosed. In this paper, a computer-aided diagnosis (CAD) system was introduced to accurately diagnose aortic enlargement from chest x-ray images. An enhanced convolutional neural network (CNN) was employed and then trained by transfer learning by using three different main areas from the original images. The areas included the left lung, heart, and right lung. The accuracy of the system was then evaluated on 1001 samples by using 4-fold cross-validation. A promising accuracy of 90% was achieved in terms of the F-measure indicator. The results showed using different areas from the original image in the training phase of CNN could increase the accuracy of predictions. This encouraged the author to evaluate this method on a larger dataset and even on different CAD systems for further enhancement of this methodology.

Keywords: computer-aided diagnosis systems, aortic enlargement, chest X-ray, image processing, convolutional neural networks

Procedia PDF Downloads 134
649 Assessment of Soil Quality Indicators in Rice Soil of Tamil Nadu

Authors: Kaleeswari R. K., Seevagan L .

Abstract:

Soil quality in an agroecosystem is influenced by the cropping system, water and soil fertility management. A valid soil quality index would help to assess the soil and crop management practices for desired productivity and soil health. The soil quality indices also provide an early indication of soil degradation and needy remedial and rehabilitation measures. Imbalanced fertilization and inadequate organic carbon dynamics deteriorate soil quality in an intensive cropping system. The rice soil ecosystem is different from other arable systems since rice is grown under submergence, which requires a different set of key soil attributes for enhancing soil quality and productivity. Assessment of the soil quality index involves indicator selection, indicator scoring and comprehensive score into one index. The most appropriate indicator to evaluate soil quality can be selected by establishing the minimum data set, which can be screened by linear and multiple regression factor analysis and score function. This investigation was carried out in intensive rice cultivating regions (having >1.0 lakh hectares) of Tamil Nadu viz., Thanjavur, Thiruvarur, Nagapattinam, Villupuram, Thiruvannamalai, Cuddalore and Ramanathapuram districts. In each district, intensive rice growing block was identified. In each block, two sampling grids (10 x 10 sq.km) were used with a sampling depth of 10 – 15 cm. Using GIS coordinates, and soil sampling was carried out at various locations in the study area. The number of soil sampling points were 41, 28, 28, 32, 37, 29 and 29 in Thanjavur, Thiruvarur, Nagapattinam, Cuddalore, Villupuram, Thiruvannamalai and Ramanathapuram districts, respectively. Principal Component Analysis is a data reduction tool to select some of the potential indicators. Principal Component is a linear combination of different variables that represents the maximum variance of the dataset. Principal Component that has eigenvalues equal or higher than 1.0 was taken as the minimum data set. Principal Component Analysis was used to select the representative soil quality indicators in rice soils based on factor loading values and contribution percent values. Variables having significant differences within the production system were used for the preparation of the minimum data set. Each Principal Component explained a certain amount of variation (%) in the total dataset. This percentage provided the weight for variables. The final Principal Component Analysis based soil quality equation is SQI = ∑ i=1 (W ᵢ x S ᵢ); where S- score for the subscripted variable; W-weighing factor derived from PCA. Higher index scores meant better soil quality. Soil respiration, Soil available Nitrogen and Potentially Mineralizable Nitrogen were assessed as soil quality indicators in rice soil of the Cauvery Delta zone covering Thanjavur, Thiruvavur and Nagapattinam districts. Soil available phosphorus could be used as a soil quality indicator of rice soils in the Cuddalore district. In rain-fed rice ecosystems of coastal sandy soil, DTPA – Zn could be used as an effective soil quality indicator. Among the soil parameters selected from Principal Component Analysis, Microbial Biomass Nitrogen could be used quality indicator for rice soils of the Villupuram district. Cauvery Delta zone has better SQI as compared with other intensive rice growing zone of Tamil Nadu.

Keywords: soil quality index, soil attributes, soil mapping, and rice soil

Procedia PDF Downloads 58
648 Addressing the Exorbitant Cost of Labeling Medical Images with Active Learning

Authors: Saba Rahimi, Ozan Oktay, Javier Alvarez-Valle, Sujeeth Bharadwaj

Abstract:

Successful application of deep learning in medical image analysis necessitates unprecedented amounts of labeled training data. Unlike conventional 2D applications, radiological images can be three-dimensional (e.g., CT, MRI), consisting of many instances within each image. The problem is exacerbated when expert annotations are required for effective pixel-wise labeling, which incurs exorbitant labeling effort and cost. Active learning is an established research domain that aims to reduce labeling workload by prioritizing a subset of informative unlabeled examples to annotate. Our contribution is a cost-effective approach for U-Net 3D models that uses Monte Carlo sampling to analyze pixel-wise uncertainty. Experiments on the AAPM 2017 lung CT segmentation challenge dataset show that our proposed framework can achieve promising segmentation results by using only 42% of the training data.

Keywords: image segmentation, active learning, convolutional neural network, 3D U-Net

Procedia PDF Downloads 126
647 One Species into Five: Nucleo-Mito Barcoding Reveals Cryptic Species in 'Frankliniella Schultzei Complex': Vector for Tospoviruses

Authors: Vikas Kumar, Kailash Chandra, Kaomud Tyagi

Abstract:

The insect order Thysanoptera includes small insects commonly called thrips. As insect vectors, only thrips are capable of Tospoviruses transmission (genus Tospovirus, family Bunyaviridae) affecting various crops. Currently, fifteen species of subfamily Thripinae (Thripidae) have been reported as vectors for tospoviruses. Frankliniella schultzei, which is reported as act as a vector for at least five tospovirses, have been suspected to be a species complex with more than one species. It is one of the historical unresolved issues where, two species namely, F. schultzei Trybom and F. sulphurea Schmutz were erected from South Africa and Srilanaka respectively. These two species were considered to be valid until 1968 when sulphurea was treated as colour morph (pale form) and synonymised under schultzei (dark form) However, these two have been considered as valid species by some of the thrips workers. Parallel studies have indicated that brown form of schultzei is a vector for tospoviruses while yellow form is a non-vector. However, recent studies have shown that yellow populations have also been documented as vectors. In view of all these facts, it is highly important to have a clear understanding whether these colour forms represent true species or merely different populations with different vector carrying capacities and whether there is some hidden diversity in 'Frankliniella schultzei species complex'. In this study, we aim to study the 'Frankliniella schultzei species complex' with molecular spectacles with DNA data from India and Australia and Africa. A total of fifty-five specimens was collected from diverse locations in India and Australia. We generated molecular data using partial fragments of mitochondrial cytochrome c oxidase I gene (mtCOI) and 28S rRNA gene. For COI dataset, there were seventy-four sequences, out of which data on fifty-five was generated in the current study and others were retrieved from NCBI. All the four different tree construction methods: neighbor-joining, maximum parsimony, maximum likelihood and Bayesian analysis, yielded the same tree topology and produced five cryptic species with high genetic divergence. For, rDNA, there were forty-five sequences, out of which data on thirty-nine was generated in the current study and others were retrieved from NCBI. The four tree building methods yielded four cryptic species with high bootstrap support value/posterior probability. Here we could not retrieve one cryptic species from South Africa as we could not generate data on rDNA from South Africa and sequence for rDNA from African region were not available in the database. The results of multiple species delimitation methods (barcode index numbers, automatic barcode gap discovery, general mixed Yule-coalescent, and Poisson-tree-processes) also supported the phylogenetic data and produced 5 and 4 Molecular Operational Taxonomic Units (MOTUs) for mtCOI and 28S dataset respectively. These results of our study indicate the likelihood that F. sulphurea may be a valid species, however, more morphological and molecular data is required on specimens from type localities of these two species and comparison with type specimens.

Keywords: DNA barcoding, species complex, thrips, species delimitation

Procedia PDF Downloads 111
646 Automated Detection of Women Dehumanization in English Text

Authors: Maha Wiss, Wael Khreich

Abstract:

Animals, objects, foods, plants, and other non-human terms are commonly used as a source of metaphors to describe females in formal and slang language. Comparing women to non-human items not only reflects cultural views that might conceptualize women as subordinates or in a lower position than humans, yet it conveys this degradation to the listeners. Moreover, the dehumanizing representation of females in the language normalizes the derogation and even encourages sexism and aggressiveness against women. Although dehumanization has been a popular research topic for decades, according to our knowledge, no studies have linked women's dehumanizing language to the machine learning field. Therefore, we introduce our research work as one of the first attempts to create a tool for the automated detection of the dehumanizing depiction of females in English texts. We also present the first labeled dataset on the charted topic, which is used for training supervised machine learning algorithms to build an accurate classification model. The importance of this work is that it accomplishes the first step toward mitigating dehumanizing language against females.

Keywords: gender bias, machine learning, NLP, women dehumanization

Procedia PDF Downloads 60