Search results for: imbalanced datasets

Authors: Maria Ezziani, Rabie Zine, Amine Amar, Ilhame Kissani

Abstract:

This paper introduces a predictive model for urban transportation engineering, which is vital for efficient traffic management. Utilizing comprehensive datasets and advanced statistical techniques, the model accurately forecasts travel times by considering temporal variations and traffic dynamics. Machine learning algorithms, including regression trees and neural networks, are employed to capture sequential dependencies. Results indicate significant improvements in predictive accuracy, particularly during peak hours and holidays, with the incorporation of traffic flow and speed variables. Future enhancements may integrate weather conditions and traffic incidents. The model's applications range from adaptive traffic management systems to route optimization algorithms, facilitating congestion reduction and enhancing journey reliability. Overall, this research extends beyond travel time estimation, offering insights into broader transportation planning and policy-making realms, empowering stakeholders to optimize infrastructure utilization and improve network efficiency.

Keywords: predictive analytics, traffic flow, travel time estimation, urban transportation, machine learning, traffic management

Procedia PDF Downloads 85

Authors: Arthur D. Wiedemann, Christopher Fuller, Kyle A. Pascioni

Abstract:

With the introduction of electric motors, small unmanned aerial vehicle designers have to consider trade-offs between acoustic noise and thrust generated. Currently, there are few low-computational tools available for predicting acoustic noise emitted by a propeller into the far-field. Artificial neural networks offer a highly non-linear and adaptive model for predicting isolated and interactive tonal noise. But neural networks require large data sets, exceeding practical considerations in modeling experimental results. A methodology known as physics guided machine learning has been applied in this study to reduce the required data set to train the network. After building and evaluating several neural networks, the best model is investigated to determine how the network successfully predicts the acoustic waveform. Lastly, a post-network transfer function is developed to remove discontinuity from the predicted waveform. Overall, methodologies from physics guided machine learning show a notable improvement in prediction performance, but additional loss functions are necessary for constructing predictive networks on small datasets.

Keywords: aeroacoustics, machine learning, propeller, rotor, neural network, physics guided machine learning

Procedia PDF Downloads 230

Authors: Andrey Khalov

Abstract:

The rapid growth of unstructured data necessitates advanced methods for transforming raw information into structured knowledge, particularly in domain-specific contexts such as IT service management and outsourcing. This paper presents a methodology for automatically constructing domain ontologies using the DOLCE framework as the base ontology. The research focuses on expanding ITIL-based ontologies by integrating concepts from ITSMO, followed by the extraction of entities and relationships from domain-specific texts through transformers and statistical methods like formal concept analysis (FCA). In particular, this work introduces an R-GNN-based approach for ontology mapping, enabling more efficient entity extraction and ontology alignment with existing knowledge bases. Additionally, the research explores transfer learning techniques using pre-trained transformer models (e.g., DeBERTa-v3-large) fine-tuned on synthetic datasets generated via large language models such as LLaMA. The resulting ontology, termed IT Ontology (ITO), is evaluated against existing methodologies, highlighting significant improvements in precision and recall. This study advances the field of ontology engineering by automating the extraction, expansion, and refinement of ontologies tailored to the IT domain, thus bridging the gap between unstructured data and actionable knowledge.

Keywords: ontology mapping, knowledge graphs, R-GNN, ITIL, NER

Procedia PDF Downloads 20

Authors: Akshay Paranjape, Nils Plettenberg, Robert Schmitt

Abstract:

Traditional optimization methods like evolutionary algorithms are widely used in production processes to find an optimal or near-optimal solution of control parameters based on the simulated environment space of a process. These algorithms are computationally intensive and therefore do not provide the opportunity for real-time optimization. This paper utilizes the Deep Reinforcement Learning (DRL) framework to find an optimal or near-optimal solution for control parameters. A model based on maximum a posteriori policy optimization (Hybrid-MPO) that can handle both numerical and categorical parameters is used as a benchmark for comparison. A comparative study shows that DRL can find optimal solutions of similar quality as compared to evolutionary algorithms while requiring significantly less time making them preferable for real-time optimization. The results are confirmed in a large-scale validation study on datasets from production and other fields. A trained XGBoost model is used as a surrogate for process simulation. Finally, multiple ways to improve the model are discussed.

Keywords: reinforcement learning, evolutionary algorithms, production process optimization, real-time optimization, hybrid-MPO

Procedia PDF Downloads 112

Authors: Ghada Badr, Arwa Alturki

Abstract:

The biological function of an RNA molecule depends on its structure. The objective of the alignment is finding the homology between two or more RNA secondary structures. Knowing the common functionalities between two RNA structures allows a better understanding and a discovery of other relationships between them. Besides, identifying non-coding RNAs -that is not translated into a protein- is a popular application in which RNA structural alignment is the first step A few methods for RNA structure-to-structure alignment have been developed. Most of these methods are partial structure-to-structure, sequence-to-structure, or structure-to-sequence alignment. Less attention is given in the literature to the use of efficient RNA structure representation and the structure-to-structure alignment methods are lacking. In this paper, we introduce an O(N2) Component-based Pairwise RNA Structure Alignment (CompPSA) algorithm, where structures are given as a component-based representation and where N is the maximum number of components in the two structures. The proposed algorithm compares the two RNA secondary structures based on their weighted component features rather than on their base-pair details. Extensive experiments are conducted illustrating the efficiency of the CompPSA algorithm when compared to other approaches and on different real and simulated datasets. The CompPSA algorithm shows an accurate similarity measure between components. The algorithm gives the flexibility for the user to align the two RNA structures based on their weighted features (position, full length, and/or stem length). Moreover, the algorithm proves scalability and efficiency in time and memory performance.

Keywords: alignment, RNA secondary structure, pairwise, component-based, data mining

Procedia PDF Downloads 459

Authors: Cheng Fang, Lingwei Quan, Cunyue Lu

Abstract:

Human pose estimation and tracking are to accurately identify and locate the positions of human joints in the video. It is a computer vision task which is of great significance for human motion recognition, behavior understanding and scene analysis. There has been remarkable progress on human pose estimation in recent years. However, more researches are needed for human pose tracking especially for online tracking. In this paper, a framework, called PoseSRPN, is proposed for online single-person pose estimation and tracking. We use Siamese network attaching a pose estimation branch to incorporate Single-person Pose Tracking (SPT) and Visual Object Tracking (VOT) into one framework. The pose estimation branch has a simple network structure that replaces the complex upsampling and convolution network structure with deconvolution. By augmenting the loss of fully convolutional Siamese network with the pose estimation task, pose estimation and tracking can be trained in one stage. Once trained, PoseSRPN only relies on a single bounding box initialization and producing human joints location. The experimental results show that while maintaining the good accuracy of pose estimation on COCO and PoseTrack datasets, the proposed method achieves a speed of 59 frame/s, which is superior to other pose tracking frameworks.

Keywords: computer vision, pose estimation, pose tracking, Siamese network

Procedia PDF Downloads 154

Authors: Junjie Zhong, Kai Hong, Lei Wang

Abstract:

Passage Retrieval (PR) plays an important role in many Natural Language Processing (NLP) tasks. Traditional efficient retrieval models relying on exact term-matching, such as TF-IDF or BM25, have nowadays been exceeded by pre-trained language models which match by semantics. Though they gain effectiveness, deep language models often require large memory as well as time cost. To tackle the trade-off between efficiency and effectiveness in PR, this paper proposes Graph Passage Retriever (GraphPR), a graph-based model inspired by the development of graph learning techniques. Different from existing works, GraphPR is end-to-end and integrates both term-matching information and semantics. GraphPR constructs a passage-level graph from BM25 retrieval results and trains a GCN-like model on the graph with graph-based objectives. Passages were regarded as nodes in the constructed graph and were embedded in dense vectors. PR can then be implemented using embeddings and a fast vector-similarity search. Experiments on a variety of real-world retrieval datasets show that the proposed model outperforms related models in several evaluation metrics (e.g., mean reciprocal rank, accuracy, F1-scores) while maintaining a relatively low query latency and memory usage.

Keywords: efficiency, effectiveness, graph learning, language model, passage retrieval, term-matching model

Procedia PDF Downloads 155

Authors: Aakansha Gupta, Neha Vadnere, Tapasvi Soni, M. Anbarsi

Abstract:

Protein secondary structure prediction is one of the most important goals pursued by bioinformatics and theoretical chemistry; it is highly important in medicine and biotechnology. Some aspects of protein functions and genome analysis can be predicted by secondary structure prediction. This is used to help annotate sequences, classify proteins, identify domains, and recognize functional motifs. In this paper, we represent protein secondary structure as a mathematical model. To extract and predict the protein secondary structure from the primary structure, we require a set of parameters. Any constants appearing in the model are specified by these parameters, which also provide a mechanism for efficient and accurate use of data. To estimate these model parameters there are many algorithms out of which the most popular one is the EM algorithm or called the Expectation Maximization Algorithm. These model parameters are estimated with the use of protein datasets like RS126 by using the Bayesian Probabilistic method (data set being categorical). This paper can then be extended into comparing the efficiency of EM algorithm to the other algorithms for estimating the model parameters, which will in turn lead to an efficient component for the Protein Secondary Structure Prediction. Further this paper provides a scope to use these parameters for predicting secondary structure of proteins using machine learning techniques like neural networks and fuzzy logic. The ultimate objective will be to obtain greater accuracy better than the previously achieved.

Keywords: model parameters, expectation maximization algorithm, protein secondary structure prediction, bioinformatics

Procedia PDF Downloads 482

Authors: Archana Singh, Jyoti Agarwal, Ajay Rana

Abstract:

Data Mining is an efficient technology to discover patterns in large databases. Association Rule Mining techniques are used to find the correlation between the various item sets in a database, and this co-relation between various item sets are used in decision making and pattern analysis. In recent years, the problem of finding association rules from large datasets has been proposed by many researchers. Various research papers on association rule mining (ARM) are studied and analyzed first to understand the existing algorithms. Apriori algorithm is the basic ARM algorithm, but it requires so many database scans. In DIC algorithm, less amount of database scan is needed but complex data structure lattice is used. The main focus of this paper is to propose a new optimized algorithm (Friendly Algorithm) and compare its performance with the existing algorithms A data set is used to find out frequent itemsets and association rules with the help of existing and proposed (Friendly Algorithm) and it has been observed that the proposed algorithm also finds all the frequent itemsets and essential association rules from databases as compared to existing algorithms in less amount of database scan. In the proposed algorithm, an optimized data structure is used i.e. Graph and Adjacency Matrix.

Keywords: association rules, data mining, dynamic item set counting, FP-growth, friendly algorithm, graph

Procedia PDF Downloads 422

Authors: Purva Mishra, Rajesh Potlia, Kuljeet Singh Sandhu

Abstract:

The role of glycyl residues in the protein structure has lingered within the research community for the last several decades. Glycyl residue is the only amino acid that is achiral due to the lack of a side chain and can, therefore, exhibit Ramachandran conformations that are disallowed for L-amino acids. The structural and functional significance of glycyl residues with L-disallowed conformation, however, remains obscure. Through statistical analysis of various datasets, we found that the glycyls with L-disallowed conformations are over-represented at disease-associated sites and tend to be evolutionarily conserved. The mutations of L-disallowed glycyls tend to destabilize the native conformation, reduce protein solubility, and promote inter-molecular aggregations. We uncovered a structural motif referred to as “β-crescent” formed around the L-disallowed glycyl, which prevents β-sheet aggregation by disrupting the alternating pattern of β-pleats. The L-disallowed conformation of glycyls also holds predictive power to infer the pathogenic missense variants. Altogether, our observations highlight that the L-disallowed conformation of glycyls is selected to facilitate native folding and prevent inter-molecular aggregations. The findings may also have implications for designing more stable proteins and prioritizing the genetic lesions implicated in diseases.

Keywords: Ramachandran plot, β-sheet, protein stability, protein aggregation

Procedia PDF Downloads 72

Authors: Xinyuan Chen, Mohd Nizam Husen, Zhongmei Zhou, Gongde Guo, Wei Gao

Abstract:

Temporal Knowledge Graph Completion (TKGC) has attracted increasing attention for its enormous value; however, existing models lack capabilities to capture both local interactions and global dependencies simultaneously with evolutionary dynamics, while the latest achievements in convolutions and Transformers haven't been employed in this area. What’s more, periodic patterns in TKGs haven’t been fully explored either. To this end, a multi-stage hybrid architecture with convolution-backed Transformers is introduced in TKGC tasks for the first time combining the Hawkes process to model evolving event sequences in a continuous-time domain. In addition, the seasonal-trend decomposition is adopted to identify periodic patterns. Experiments on six public datasets are conducted to verify model effectiveness against state-of-the-art (SOTA) methods. An extensive ablation study is carried out accordingly to evaluate architecture variants as well as the contributions of independent components in addition, paving the way for further potential exploitation. Besides complexity analysis, input sensitivity and safety challenges are also thoroughly discussed for comprehensiveness with novel methods.

Keywords: temporal knowledge graph completion, convolution, transformer, Hawkes process, periodicity

Procedia PDF Downloads 79

Authors: Adrian-Gabriel Chifu, Sebastien Fournier

Abstract:

One of the challenges of natural language understanding is to deal with the subjectivity of sentences, which may express opinions and emotions that add layers of complexity and nuance. Sentiment analysis is a field that aims to extract and analyze these subjective elements from text, and it can be applied at different levels of granularity, such as document, paragraph, sentence, or aspect. Aspect-based sentiment analysis is a well-studied topic with many available data sets and models. However, there is no clear definition of what makes a sentence difficult for aspect-based sentiment analysis. In this paper, we explore this question by conducting an experiment with three data sets: ”Laptops”, ”Restaurants”, and ”MTSC” (Multi-Target-dependent Sentiment Classification), and a merged version of these three datasets. We study the impact of domain diversity and syntactic diversity on difficulty. We use a combination of classifiers to identify the most difficult sentences and analyze their characteristics. We employ two ways of defining sentence difficulty. The first one is binary and labels a sentence as difficult if the classifiers fail to correctly predict the sentiment polarity. The second one is a six-level scale based on how many of the top five best-performing classifiers can correctly predict the sentiment polarity. We also define 9 linguistic features that, combined, aim at estimating the difficulty at sentence level.

Keywords: sentiment analysis, difficulty, classification, machine learning

Procedia PDF Downloads 92

Authors: Tian Zhang

Abstract:

Unsupervised object segmentation aims at segmenting objects in sequential images and obtaining the mask of each object without any manual intervention. Unsupervised segmentation remains a challenging task due to the lack of prior knowledge about these objects. Previous methods often require manually specifying the action of each object, which is often difficult to obtain. Instead, this paper does not need action information of objects and automatically learns the actions and relations among objects from the structured environment. To obtain the object segmentation of sequential images, the relationships between objects and images are extracted to infer the action and interaction of objects based on the multi-head attention mechanism. Three types of objects’ relationships in the object segmentation task are proposed: the relationship between objects in the same frame, the relationship between objects in two frames, and the relationship between objects and historical information. Based on these relationships, the proposed model (1) is effective in multiple objects segmentation tasks, (2) just needs images as input, and (3) produces better segmentation results as more relationships are considered. The experimental results on multiple datasets show that this paper’s method achieves state-of-art performance. The quantitative and qualitative analyses of the result are conducted. The proposed method could be easily extended to other similar applications.

Keywords: unsupervised object segmentation, attention mechanism, contrastive learning, structured environment

Procedia PDF Downloads 111

Authors: Taylor Kolody, Farkhund Iqbal, Rabia Batool, Benjamin Fung, Mohammed Hussaeni, Saiqa Aleem

Abstract:

The rising threat of climate change has led to an increase in public awareness and care about our collective and individual environmental impact. A key component of this impact is our use of cars and other polluting forms of transportation, but it is often difficult for an individual to know how severe this impact is. While there are applications that offer this feedback, they require manual entry of what transportation mode was used for a given trip, which can be burdensome. In order to alleviate this shortcoming, a data from the 2016 TRIPlab datasets has been used to train a variety of machine learning models to automatically recognize the mode of transportation. The accuracy of 89.6% is achieved using single deep neural network model with Gated Recurrent Unit (GRU) architecture applied directly to trip data points over 4 primary classes, namely walking, public transit, car, and bike. These results are comparable in accuracy to results achieved by others using ensemble methods and require far less computation when classifying new trips. The lack of trip context data, e.g., bus routes, bike paths, etc., and the need for only a single set of weights make this an appropriate methodology for applications hoping to reach a broad demographic and have responsive feedback.

Keywords: classification, gated recurrent unit, recurrent neural network, transportation

Procedia PDF Downloads 138

Authors: Arturo Maldonado

Abstract:

In the iron ore mining industry, the intact strength of rock units is defined using the uniaxial compressive strength (UCS). This parameter is very important for the classification of shale materials, allowing the split between rock and cohesive soils based on the magnitude of UCS. For this research, it is assumed that UCS less than or equal to 1 MPa is representative of soils. Several researchers have anticipated that the magnitude of UCS reduces with weathering progression, also since UCS is a directional property, its magnitude depends upon the rock fabric orientation. Thus, the paper presents how the UCS of shales is affected by both weathering grade and bedding orientation. The mineralogy of shales has been defined using Hyper-spectral and chemical assays to define the mineral constituents of shale and other non-shale materials. Geological classification tools have been used to define distinct lithological types, and in this manner, the author uses mineralogical datasets to recognize and isolate shales from other rock types and develop tertiary plots for fresh and weathered shales. The mineralogical classification of shales has reduced the contamination of lithology types and facilitated the study of the physical factors affecting the intact strength of shales, like anisotropic strength due to bedding orientation. The analysis of mineralogical characteristics of shales is perhaps the most important contribution of this paper to other researchers who may wish to explore similar methods.

Keywords: rock mechanics, mineralogy, shales, weathering, anisotropy

Procedia PDF Downloads 65

Authors: Fang Gong

Abstract:

Distance measures are an indispensable part of many instance-based learning (IBL) and machine learning (ML) algorithms. The value difference metrics (VDM) and inverted specific-class distance measure (ISCDM) are among the top-performing distance measures that address nominal attributes. VDM performs well in some domains owing to its simplicity and poorly in others that exist missing value and non-class attribute noise. ISCDM, however, typically works better than VDM on such domains. To maximize their advantages and avoid disadvantages, in this paper, a selection approach: a discriminative model for nominal attributes-based distance measures is proposed. More concretely, VDM and ISCDM are built independently on a training dataset at the training stage, and the most credible one is recorded for each training instance. At the test stage, its nearest neighbor for each test instance is primarily found by any of VDM and ISCDM and then chooses the most reliable model of its nearest neighbor to predict its class label. It is simply denoted as a discriminative distance measure (DDM). Experiments are conducted on the 34 University of California at Irvine (UCI) machine learning repository datasets, and it shows DDM retains the interpretability and simplicity of VDM and ISCDM but significantly outperforms the original VDM and ISCDM and other state-of-the-art competitors in terms of accuracy.

Keywords: distance measure, discriminative model, nominal attributes, nearest neighbor

Procedia PDF Downloads 115

Authors: Aghamusa Azizov

Abstract:

This study investigates the correlation between solar activity and sentiment in news media coverage, using a large-scale dataset of solar activity since 1750 and over 15 million articles from "The New York Times" dating from 1851 onwards. Employing Pearson's correlation coefficient and multiple Natural Language Processing (NLP) tools—TextBlob, Vader, and DistillBERT—the research examines the extent to which fluctuations in solar phenomena are reflected in the sentiment of historical news narratives. The findings reveal that the correlation between solar activity and media sentiment is generally negligible, suggesting a weak influence of solar patterns on the portrayal of events in news media. Notably, a moderate positive correlation was observed between the sentiments derived from TextBlob and Vader, indicating consistency across NLP tools. The analysis provides insights into the historical impact of solar activity on human affairs and highlights the importance of using multiple analytical methods to understand complex relationships in large datasets. The study contributes to the broader understanding of how extraterrestrial factors may intersect with media-reported events and underlines the intricate nature of interdisciplinary research in the data science and historical domains.

Keywords: solar activity correlation, media sentiment analysis, natural language processing, historical event patterns

Procedia PDF Downloads 79

Authors: Bruno Pinto Vieira, Marco Antonio Calijorne Soares, Armando Sérgio de Aguiar Filho

Abstract:

Nowadays, a massive amount of information has been produced by different data sources, including mobile devices and transactional systems. In this scenario, concerns arise on how to maintain or establish data quality, which is now treated as a product to be defined, measured, analyzed, and improved to meet consumers' needs, which is the one who uses these data in decision making and companies strategies. Information that reaches low levels of quality can lead to issues that can consume time and money, such as missed business opportunities, inadequate decisions, and bad risk management actions. The step of selecting, identifying, evaluating, and selecting data sources with significant quality according to the need has become a costly task for users since the sources do not provide information about their quality. Traditional data quality control methods are based on user experience or business rules limiting performance and slowing down the process with less than desirable accuracy. Using advanced machine learning algorithms, it is possible to take advantage of computational resources to overcome challenges and add value to companies and users. In this study, machine learning is applied to data quality analysis on different datasets, seeking to compare the performance of the techniques according to the dimensions of quality assessment. As a result, we could create a ranking of approaches used, besides a system that is able to carry out automatically, data quality assessment.

Keywords: machine learning, data quality, quality dimension, quality assessment

Procedia PDF Downloads 150

Authors: Htay Htay Yi, Zin May Aye

Abstract:

Network security is role of the ICT environment because malicious users are continually growing that realm of education, business, and then related with ICT. The network security contravention is typically described and examined centrally based on a security event management system. The firewalls, Intrusion Detection System (IDS), and Intrusion Prevention System are becoming essential to monitor or prevent of potential violations, incidents attack, and imminent threats. In this system, the firewall rules are set only for where the system policies are needed. Dataset deployed in this system are derived from the testbed environment. The traffic as in DoS and PortScan traffics are applied in the testbed with firewall and IDS implementation. The network traffics are classified as normal or attacks in the existing testbed environment based on six machine learning classification methods applied in the system. It is required to be tested to get datasets and applied for DoS and PortScan. The dataset is based on CICIDS2017 and some features have been added. This system tested 26 features from the applied dataset. The system is to reduce false positive rates and to improve accuracy in the implemented testbed design. The system also proves good performance by selecting important features and comparing existing a dataset by machine learning classifiers.

Keywords: false negative rate, intrusion detection system, machine learning methods, performance

Procedia PDF Downloads 118

Authors: Huang Bai-Cheng

Abstract:

When deploying the Oriented Fast and Rotated Binary Robust Independent Elementary Feature (BRIEF) (ORB) extraction algorithm on field-programmable gate arrays (FPGA), the access of global storage for 31×31 pixel patches of the features has become the bottleneck of the system efficiency. Therefore, a feature zone strategy has been proposed. Zones are searched as features are detected. Pixels around the feature zones are extracted from global memory and distributed into patches corresponding to feature coordinates. The proposed FPGA structure is targeted on a Xilinx FPGA development board of Zynq UltraScale+ series, and multiple datasets are tested. Compared with the streaming pixel patch extraction method, the proposed architecture obtains at least two times acceleration consuming extra 3.82% Flip-Flops (FFs) and 7.78% Look-Up Tables (LUTs). Compared with the non-streaming one, the proposed architecture saves 22.3% LUT and 1.82% FF, causing a latency of only 0.2ms and a drop in frame rate for 1. Compared with the related works, the proposed strategy and hardware architecture have the superiority of keeping a balance between FPGA resources and performance.

Keywords: feature extraction, real-time, ORB, FPGA implementation

Procedia PDF Downloads 122

Authors: Andrey Khalov

Abstract:

The rapid proliferation of unstructured data in IT infrastructure management demands innovative approaches for extracting actionable knowledge. This paper presents a framework for ontology-based knowledge extraction that combines relational graph neural networks (R-GNN) with large language models (LLMs). The proposed method leverages the DOLCE framework as the foundational ontology, extending it with concepts from ITSMO for domain-specific applications in IT service management and outsourcing. A key component of this research is the use of transformer-based models, such as DeBERTa-v3-large, for automatic entity and relationship extraction from unstructured texts. Furthermore, the paper explores how transfer learning techniques can be applied to fine-tune large language models (LLaMA) for using to generate synthetic datasets to improve precision in BERT-based entity recognition and ontology alignment. The resulting IT Ontology (ITO) serves as a comprehensive knowledge base that integrates domain-specific insights from ITIL processes, enabling more efficient decision-making. Experimental results demonstrate significant improvements in knowledge extraction and relationship mapping, offering a cutting-edge solution for enhancing cognitive computing in IT service environments.

Keywords: ontology mapping, R-GNN, knowledge extraction, large language models, NER, knowlege graph

Procedia PDF Downloads 19

Authors: Dhurgham Al-Karawi, Nisreen Polus, Shakir Al-Zaidi, Sabah Jassim

Abstract:

The management of COVID-19 patients based on chest imaging is emerging as an essential tool for evaluating the spread of the pandemic which has gripped the global community. It has already been used to monitor the situation of COVID-19 patients who have issues in respiratory status. There has been increase to use chest imaging for medical triage of patients who are showing moderate-severe clinical COVID-19 features, this is due to the fast dispersal of the pandemic to all continents and communities. This article demonstrates the development of machine learning techniques for the test of COVID-19 patients using Chest X-Ray (CXR) images in nearly real-time, to distinguish the COVID-19 infection with a significantly high level of accuracy. The testing performance has covered a combination of different datasets of CXR images of positive COVID-19 patients, patients with viral and bacterial infections, also, people with a clear chest. The proposed AI scheme successfully distinguishes CXR scans of COVID-19 infected patients from CXR scans of viral and bacterial based pneumonia as well as normal cases with an average accuracy of 94.43%, sensitivity 95%, and specificity 93.86%. Predicted decisions would be supported by visual evidence to help clinicians speed up the initial assessment process of new suspected cases, especially in a resource-constrained environment.

Keywords: COVID-19, chest x-ray scan, artificial intelligence, texture analysis, local binary pattern transform, Gabor filter

Procedia PDF Downloads 147

Authors: Sandy Black, Calum McLaughlin, Alessandro Pranzitelli, Marc Laing

Abstract:

Multiphase flow structures of two-phase multicomponent fluids were experimentally investigated in a large diameter high-pressure pipeline up to 130 bar at TÜV SÜD’s National Engineering Laboratory Advanced Multiphase Facility. One of the main objectives of the experimental test campaign was to evaluate the impact of pressure on multiphase flow patterns as much of the existing information is based on low-pressure measurements. The experiments were performed in a horizontal and vertical orientation in both 4-inch and 6-inch pipework using nitrogen, ExxsolTM D140 oil, and a 6% aqueous solution of NaCl at incremental pressures from 10 bar to 130 bar. To visualise the detailed structure of the flow of the entire cross-section of the pipe, a fast response X-ray tomography system was used. A wide range of superficial velocities from 0.6 m/s to 24.0 m/s for gas and 0.04 m/s and 6.48 m/s for liquid was examined to evaluate different flow regimes. The results illustrated the suppression of instabilities between the gas and the liquid at the measurement location and that intermittent or slug flow was observed less frequently as the pressure was increased. CFD modellings of low and high-pressure simulations were able to successfully predict the likelihood of intermittent flow; however, further tuning is necessary to predict the slugging frequency. The dataset generated is unique as limited datasets exist above 100 bar and is of considerable value to multiphase flow specialists and numerical modellers.

Keywords: computational fluid dynamics, high pressure, multiphase, X-ray tomography

Procedia PDF Downloads 145

Authors: Amit Kumar, Amanpreet Chander, Ashish Sahani

Abstract:

This paper presents our recent work on real-time human activity detection based on the media pipe pipeline and machine learning algorithms. The proposed system can detect human activities, including running, jumping, squatting, bending to the left or right, and standing still. This is a robust solution for developing a yoga, dance, metaverse, and fitness application that checks for the correction of the pose without having any additional monitor like a personal trainer. MediaPipe solution offers an open-source cross-platform which utilizes a two-step detector-tracker ML pipeline for live detection of key landmarks on our body which can be used for motion data collection. The prediction of real-time poses uses a variety of machine learning techniques and different types of analysis. Without primarily relying on powerful desktop environments for inference, our method achieves real-time performance on the majority of contemporary mobile phones, desktops/laptops, Python, or even the web. Experimental results show that our method outperforms the existing method in terms of accuracy and real-time capability, achieving an accuracy of 99.92% on testing datasets.

Keywords: human activity detection, media pipe, machine learning, metaverse applications

Procedia PDF Downloads 180

Authors: Georgina Cherry, Taranpreet Rai, Luke Boyden, Sitira Williams, Andrea Wright, Richard Brown, Viva Chu, Alasdair Cook, Kevin Wells

Abstract:

Estimating population-level burden, abilities of pet-parents to identify disease and demand for veterinary services worldwide is challenging. The purpose of this study is to compare a feline pruritus survey with social media listening (SML) data discussing this condition. Surveys are expensive and labour intensive to analyse, but SML data is freeform and requires careful filtering for relevancy. This study considers data from a survey of owner-observed symptoms of 156 pruritic cats conducted using Pet Parade® and SML posts collected through web-scraping to gain insights into the characterisation and management of feline pruritus. SML posts meeting a feline body area, behaviour and symptom were captured and reviewed for relevance representing 1299 public posts collected from 2021 to 2023. The survey involved 1067 pet-parents who reported on pruritic symptoms in their cats. Among the observed cats, approximately 18.37% (n=196) exhibited at least one symptom. The most frequently reported symptoms were hair loss (9.2%), bald spots (7.3%) and infection, crusting, scaling, redness, scabbing, scaling, or bumpy skin (8.2%). Notably, bald spots were the primary symptom reported for short-haired cats, while other symptoms were more prevalent in medium and long-haired cats. Affected body areas, according to pet-parents, were primarily the head, face, chin, neck (27%), and the top of the body, along the spine (22%). 35% of all cats displayed excessive behaviours consistent with pruritic skin disease. Interestingly, 27% of these cats were perceived as non-symptomatic by their owners, suggesting an under-identification of itch-related signs. Furthermore, a significant proportion of symptomatic cats did not receive any skin disease medication, whether prescribed or over the counter (n=41). These findings indicate a higher incidence of pruritic skin disease in cats than recognized by pet owners, potentially leading to a lack of medical intervention for clinically symptomatic cases. The comparison between the survey and social media listening data revealed bald spots were reported in similar proportions in both datasets (25% in the survey and 28% in SML). Infection, crusting, scaling, redness, scabbing, scaling, or bumpy skin accounted for 31% of symptoms in the survey, whereas it represented 53% of relevant SML posts (excluding bumpy skin). Abnormal licking or chewing behaviours were mentioned by pet-parents in 40% of SML posts compared to 38% in the survey. The consistency in the findings of these two disparate data sources, including a complete overlap in affected body areas for the top 80% of social media listening posts, indicates minimal biases in each method, as significant biases would likely yield divergent results. Therefore, the strong agreement across pruritic symptoms, affected body areas, and reported behaviours enhances our confidence in the reliability of the findings. Moreover, the small differences identified between the datasets underscore the valuable insights that arise from utilising multiple data sources. These variations provide additional depth in characterising and managing feline pruritus, allowing for more comprehensive understanding of the condition. By combining survey data and social media listening, researchers can obtain a nuanced perspective and capture a wider range of experiences and perspectives, supporting informed decision-making in veterinary practice.

Keywords: social media listening, feline pruritus, surveys, felines, cats, pet owners

Procedia PDF Downloads 134

Authors: John M. Lyne, K. Andrea Scott

Abstract:

In fluid dynamics, direct numerical simulations (DNS) of turbulent flows require large amounts of nodes to appropriately resolve all scales of energy transfer. Due to the size of these databases, sharing these datasets amongst the academic community is a challenge. Recent work has been done to investigate the use of super-resolution to enable database sharing, where a low-resolution flow field is super-resolved to high resolutions using a neural network. Recently, Generative Adversarial Networks (GAN) have grown in popularity with impressive results in the generation of faces, landscapes, and more. This work investigates the generation of unique high-resolution channel flow velocity fields from a low-dimensional latent space using a GAN. The training objective of the GAN is to generate samples in which the distribution of the generated samplesis ideally indistinguishable from the distribution of the training data. In this study, the network is trained using samples drawn from a statistically stationary channel flow at a Reynolds number of 560. Results show that the turbulent statistics and energy spectra of the generated flow fields are within reasonable agreement with those of the DNS data, demonstrating that GANscan produce the intricate multi-scale phenomena of turbulence.

Keywords: computational fluid dynamics, channel flow, turbulence, generative adversarial network

Procedia PDF Downloads 207

Authors: Tanmay Bisen, Aastha Shayla

Abstract:

This study introduces an advanced approach to tree canopy detection in urban environments and a regional analysis of Northern and Central Kolkata that delves into the intricate relationship between urban development and environmental conservation. Leveraging high-resolution drone imagery from diverse urban green spaces in Kolkata, we fine-tuned the deep forest model to enhance its precision and accuracy. Our results, characterized by an impressive Intersection over Union (IoU) score of 0.90 and a mean average precision (mAP) of 0.87, underscore the model's robustness in detecting and classifying tree crowns amidst the complexities of aerial imagery. This research not only emphasizes the importance of model customization for specific datasets but also highlights the potential of drone-based remote sensing in urban forestry studies. The study investigates the spatial distribution, density, and environmental impact of trees in Northern and Central Kolkata. The findings underscore the significance of urban green spaces in met-ropolitan cities, emphasizing the need for sustainable urban planning that integrates green infrastructure for ecological balance and human well-being.

Keywords: urban greenery, advanced spatial distribution analysis, drone imagery, deep learning, tree detection

Procedia PDF Downloads 57

Authors: Hussin K. Ragb, Vijayan K. Asari

Abstract:

In this paper, we present a pedestrian detection descriptor called Fused Structure and Texture (FST) features based on the combination of the local phase information with the texture features. Since the phase of the signal conveys more structural information than the magnitude, the phase congruency concept is used to capture the structural features. On the other hand, the Center-Symmetric Local Binary Pattern (CSLBP) approach is used to capture the texture information of the image. The dimension less quantity of the phase congruency and the robustness of the CSLBP operator on the flat images, as well as the blur and illumination changes, lead the proposed descriptor to be more robust and less sensitive to the light variations. The proposed descriptor can be formed by extracting the phase congruency and the CSLBP values of each pixel of the image with respect to its neighborhood. The histogram of the oriented phase and the histogram of the CSLBP values for the local regions in the image are computed and concatenated to construct the FST descriptor. Several experiments were conducted on INRIA and the low resolution DaimlerChrysler datasets to evaluate the detection performance of the pedestrian detection system that is based on the FST descriptor. A linear Support Vector Machine (SVM) is used to train the pedestrian classifier. These experiments showed that the proposed FST descriptor has better detection performance over a set of state of the art feature extraction methodologies.

Keywords: pedestrian detection, phase congruency, local phase, LBP features, CSLBP features, FST descriptor

Procedia PDF Downloads 490

Authors: Yao-Hong Tsai

Abstract:

Due to the sensor technology, video surveillance has become the main way for security control in every big city in the world. Surveillance is usually used by governments for intelligence gathering, the prevention of crime, the protection of a process, person, group or object, or the investigation of crime. Many surveillance systems based on computer vision technology have been developed in recent years. Moving target tracking is the most common task for Unmanned Aerial Vehicle (UAV) to find and track objects of interest in mobile aerial surveillance for civilian applications. The paper is focused on vision-based collision avoidance for UAVs by recurrent neural networks. First, images from cameras on UAV were fused based on deep convolutional neural network. Then, a recurrent neural network was constructed to obtain high-level image features for object tracking and extracting low-level image features for noise reducing. The system distributed the calculation of the whole system to local and cloud platform to efficiently perform object detection, tracking and collision avoidance based on multiple UAVs. The experiments on several challenging datasets showed that the proposed algorithm outperforms the state-of-the-art methods.

Keywords: unmanned aerial vehicle, object tracking, deep learning, collision avoidance

Procedia PDF Downloads 161

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

Abstract:

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

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

Procedia PDF Downloads 67