Search results for: imbalanced datasets

Authors: Prathiksha Kamath, Pratibha Jamkhandi, Prateek Ghanti, Priyanshu Gupta, M. Lakshmi Neelima

Abstract:

Combining Image Captioning and Visual Question Answering (VQA) tasks have emerged as a new and exciting research area. The image captioning task involves generating a textual description that summarizes the content of the image. VQA aims to answer a natural language question about the image. Both these tasks include computer vision and natural language processing (NLP) and require a deep understanding of the content of the image and semantic relationship within the image and the ability to generate a response in natural language. There has been remarkable growth in both these tasks with rapid advancement in deep learning. In this paper, we present a comprehensive review of recent progress in combining image captioning and visual question-answering (VQA) tasks. We first discuss both image captioning and VQA tasks individually and then the various ways in which both these tasks can be integrated. We also analyze the challenges associated with these tasks and ways to overcome them. We finally discuss the various datasets and evaluation metrics used in these tasks. This paper concludes with the need for generating captions based on the context and captions that are able to answer the most likely asked questions about the image so as to aid the VQA task. Overall, this review highlights the significant progress made in combining image captioning and VQA, as well as the ongoing challenges and opportunities for further research in this exciting and rapidly evolving field, which has the potential to improve the performance of real-world applications such as autonomous vehicles, robotics, and image search.

Keywords: image captioning, visual question answering, deep learning, natural language processing

Procedia PDF Downloads 51

Authors: Edward K. Boahen, Brunel E. Bouya-Moko, Changda Wang

Abstract:

The major threat to online social network (OSN) users is account compromisation. Spammers now spread malicious messages by exploiting the trust relationship established between account owners and their friends. The challenge in detecting a compromised account by service providers is validating the trusted relationship established between the account owners, their friends, and the spammers. Another challenge is the increase in required human interaction with the feature selection. Research available on supervised learning (machine learning) has limitations with the feature selection and accounts that cannot be profiled, like application programming interface (API). Therefore, this paper discusses the various behaviours of the OSN users and the current approaches in detecting a compromised OSN account, emphasizing its limitations and challenges. We propose a deep learning approach that addresses and resolve the constraints faced by the previous schemes. We detailed our proposed optimized nonsymmetric deep auto-encoder (OPT_NDAE) for unsupervised feature learning, which reduces the required human interaction levels in the selection and extraction of features. We evaluated our proposed classifier using the NSL-KDD and KDDCUP'99 datasets in a graphical user interface enabled Weka application. The results obtained indicate that our proposed approach outperformed most of the traditional schemes in OSN compromised account detection with an accuracy rate of 99.86%.

Keywords: computer security, network security, online social network, account compromisation

Procedia PDF Downloads 91

Authors: Omar Elharrouss, Youssef Hmamouche, Assia Kamal Idrissi, Btissam El Khamlichi, Amal El Fallah-Seghrouchni

Abstract:

Edge detection is represented as one of the most challenging tasks in computer vision, due to the complexity of detecting the edges or boundaries in real-world images that contains objects of different types and scales like trees, building as well as various backgrounds. Edge detection is represented also as a key task for many computer vision applications. Using a set of backbones as well as attention modules, deep-learning-based methods improved the detection of edges compared with the traditional methods like Sobel and Canny. However, images of complex scenes still represent a challenge for these methods. Also, the detected edges using the existing approaches suffer from non-refined results while the image output contains many erroneous edges. To overcome this, n this paper, by using the mechanism of residual learning, a refined edge detection network is proposed (RED-Net). By maintaining the high resolution of edges during the training process, and conserving the resolution of the edge image during the network stage, we make the pooling outputs at each stage connected with the output of the previous layer. Also, after each layer, we use an affined batch normalization layer as an erosion operation for the homogeneous region in the image. The proposed methods are evaluated using the most challenging datasets including BSDS500, NYUD, and Multicue. The obtained results outperform the designed edge detection networks in terms of performance metrics and quality of output images.

Keywords: edge detection, convolutional neural networks, deep learning, scale-representation, backbone

Procedia PDF Downloads 69

Authors: Geetika Barman, B. S. Daya Sagar

Abstract:

In this article, we proposed a pixel-wise classification of hyperspectral images using a mathematical morphology operator-based distance metric called “dilation distance” and “erosion distance”. This method involves measuring the spatial distance between the spectral features of a hyperspectral image across the bands. The key concept of the proposed approach is that the “dilation distance” is the maximum distance a pixel can be moved without changing its classification, whereas the “erosion distance” is the maximum distance that a pixel can be moved before changing its classification. The spectral signature of the hyperspectral image carries unique class information and shape for each class. This article demonstrates how easily the dilation and erosion distance can measure spatial distance compared to other approaches. This property is used to calculate the spatial distance between hyperspectral image feature vectors across the bands. The dissimilarity matrix is then constructed using both measures extracted from the feature spaces. The measured distance metric is used to distinguish between the spectral features of various classes and precisely distinguish between each class. This is illustrated using both toy data and real datasets. Furthermore, we investigated the role of flat vs. non-flat structuring elements in capturing the spatial features of each class in the hyperspectral image. In order to validate, we compared the proposed approach to other existing methods and demonstrated empirically that mathematical operator-based distance metric classification provided competitive results and outperformed some of them.

Keywords: dilation distance, erosion distance, hyperspectral image classification, mathematical morphology

Procedia PDF Downloads 55

Authors: Subhan Ahmad, Pradeep Bhargava, Ajay Chourasia

Abstract:

Shear failure in reinforced concrete beams without shear reinforcement leads to loss of property and life since a very little or no warning occurs before failure as in case of flexural failure. Shear strength of reinforced concrete beams decreases as its depth increases. This phenomenon is generally called as the size effect. In this paper, a comparative analysis is performed to estimate the performance of shear strength models in capturing the size effect of reinforced concrete beams made with conventional concrete, self-compacting concrete, and recycled aggregate concrete. Four shear strength models that account for the size effect in shear are selected from the literature and applied on the datasets of slender reinforced concrete beams. Beams prepared with conventional concrete, self-compacting concrete, and recycled aggregate concrete are considered for the analysis. Results showed that all the four models captured the size effect in shear effectively and produced conservative estimates of the shear strength for beams made with normal strength conventional concrete. These models yielded unconservative estimates for high strength conventional concrete beams with larger effective depths ( > 450 mm). Model of Bazant and Kim (1984) captured the size effect precisely and produced conservative estimates of shear strength of self-compacting concrete beams at all the effective depths. Also, shear strength models considered in this study produced unconservative estimates of shear strength for recycled aggregate concrete beams at all effective depths.

Keywords: reinforced concrete beams; shear strength; prediction models; size effect

Procedia PDF Downloads 129

Authors: Delphine Vendryes, Dushyanth Sekhar, Baojia Tong, Matthew Theisen, Chester Curme

Abstract:

Much of the data that inform the decisions of governments, corporations and individuals are harvested from unstructured documents. Data extraction is defined here as a process that turns non-machine-readable information into a machine-readable format that can be stored, for instance, in a database. In financial services, introducing more automation in data extraction pipelines is a major challenge. Information sought by financial data consumers is often buried within vast bodies of unstructured documents, which have historically required thorough manual extraction. Automated solutions provide faster access to non-machine-readable datasets, in a context where untimely information quickly becomes irrelevant. Data quality standards cannot be compromised, so automation requires high data integrity. This multifaceted task is broken down into smaller steps: ingestion, table parsing (detection and structure recognition), text analysis (entity detection and disambiguation), schema-based record extraction, user feedback incorporation. Selected intermediary steps are phrased as machine learning problems. Solutions leveraging cutting-edge approaches from the fields of computer vision (e.g. table detection) and natural language processing (e.g. entity detection and disambiguation) are proposed.

Keywords: computer vision, entity recognition, finance, information retrieval, machine learning, natural language processing

Procedia PDF Downloads 87

Authors: Chong Zhang, Guoming Tang, Bin Ge, Jiuyang Tang

Abstract:

With the fast growing road traffic and increasingly severe traffic congestion, more and more citizens choose to use the public transportation for daily travelling. Meanwhile, the shared bike provides a convenient option for the first and last mile to the public transit. As of 2016, over one thousand cities around the world have deployed the bike-sharing system. The combination of these two transportations have stimulated the development of each other and made significant contribution to the reduction of carbon footprint. A lot of work has been done on mining the riding behaviors in various bike-sharing systems. Most of them, however, treated the bike-sharing system as an isolated system and thus their results provide little reference for the public transit construction and optimization. In this work, we treat the bike-sharing and public transit as a whole and investigate the customers’ bike-and-ride behaviors. Specifically, we develop a spatio-temporal traffic delivery model to study the riding patterns between the two transportation systems and explore the traffic characteristics (e.g., distributions of customer arrival/departure and traffic peak hours) from the time and space dimensions. During the model construction and evaluation, we make use of large open datasets from real-world bike-sharing systems (the CitiBike in New York, GoBike in San Francisco and BIXI in Montreal) along with corresponding public transit information. The developed two-dimension traffic model, as well as the mined bike-and-ride behaviors, can provide great help to the deployment of next-generation intelligent transportation systems.

Keywords: riding pattern mining, bike-sharing system, public transportation, bike-and-ride behavior

Procedia PDF Downloads 745

Authors: Todd Zhou, Mikhail Yurochkin

Abstract:

Out-of-distribution (OOD) detection is receiving increasing amounts of attention in the machine learning research community, boosted by recent technologies, such as autonomous driving and image processing. This newly-burgeoning field has called for the need for more effective and efficient methods for out-of-distribution generalization methods. Without accessing the label information, deploying machine learning models to out-of-distribution domains becomes extremely challenging since it is impossible to evaluate model performance on unseen domains. To tackle this out-of-distribution detection difficulty, we designed a model selection pipeline algorithm and developed a model selection framework with different impurity and randomness measurements to evaluate and choose the best-performing models for out-of-distribution data. By exploring different randomness scores based on predicted probabilities, we adopted the out-of-distribution entropy and developed a custom-designed score, ”CombinedScore,” as the evaluation criterion. This proposed score was created by adding labeled source information into the judging space of the uncertainty entropy score using harmonic mean. Furthermore, the prediction bias was explored through the equality of opportunity violation measurement. We also improved machine learning model performance through model calibration. The effectiveness of the framework with the proposed evaluation criteria was validated on the Folktables American Community Survey (ACS) datasets.

Keywords: model selection, domain generalization, model fairness, randomness measurements, bias index

Procedia PDF Downloads 101

Authors: Yun Wang, Ruoyu Wang, Tuo Chen, Guangxiu Liu, Guodong Chen, Wei Zhang

Abstract:

Tuberculosis (TB) is a major public health problem in China. China has the world's second largest tuberculosis epidemic (after India). Xinjiang almost has the highest annual attendance rate of TB in China, and the province is also famous because of its severe dust storms. The epidemic timing starts in February and ends in July, and the dust storm mainly distribute throughout the spring and early summer, which strongly indicate a close linkage between causative agent of TB and dust storm events. However, mechanisms responsible for the observed patterns are still not clearly indentified. By comparing the information on cases of TB from Centers for Disease Control of China annual reports with dust storm atmosphere datasets, we constructed the relationship between the large scale annual occurrence of TB in Xinjiang, a Northwest province of China, and dust storm occurrence. Regional atmospheric indexes of dust storm based on surface wind speed show a clear link between population dynamics of the disease and the climate disaster: the onset of epidemics and the dust storm defined by the atmospheric index share the same mean year. This study is the first that provides a clear demonstration of connections that exist between TB epidemics and dust storm events in China. The development of this study will undoubtedly help early warning for tuberculosis epidemic onset in China and help nationwide and international public health institutions and policy makers to better control TB disease in Norwest China.

Keywords: dust storm, tuberculosis, Xinjiang province, epidemic

Procedia PDF Downloads 417

Authors: Hala Alshamlan, Ghada Badr, Yousef Alohali

Abstract:

Cancer is one of the dreadful diseases, which causes considerable death rate in humans. DNA microarray-based gene expression profiling has been emerged as an efficient technique for cancer classification, as well as for diagnosis, prognosis, and treatment purposes. In recent years, a DNA microarray technique has gained more attraction in both scientific and in industrial fields. It is important to determine the informative genes that cause cancer to improve early cancer diagnosis and to give effective chemotherapy treatment. In order to gain deep insight into the cancer classification problem, it is necessary to take a closer look at the proposed gene selection methods. We believe that they should be an integral preprocessing step for cancer classification. Furthermore, finding an accurate gene selection method is a very significant issue in a cancer classification area because it reduces the dimensionality of microarray dataset and selects informative genes. In this paper, we classify and review the state-of-art gene selection methods. We proceed by evaluating the performance of each gene selection approach based on their classification accuracy and number of informative genes. In our evaluation, we will use four benchmark microarray datasets for the cancer diagnosis (leukemia, colon, lung, and prostate). In addition, we compare the performance of gene selection method to investigate the effective gene selection method that has the ability to identify a small set of marker genes, and ensure high cancer classification accuracy. To the best of our knowledge, this is the first attempt to compare gene selection approaches for cancer classification using microarray gene expression profile.

Keywords: gene selection, feature selection, cancer classification, microarray, gene expression profile

Procedia PDF Downloads 423

Authors: Florian Monetti, Brett Beamsley, Peter McComb, Simon Weppe

Abstract:

Land reclamation from the ocean has considerably increased over past decades to support worldwide rapid urban growth. Reshaping the coastline, however, inevitably affects coastal systems. One of the main challenges for coastal oceanographers is to predict the physical and morphological responses for nearshore systems to man-made changes over multiple time-scales. Fully-coupled numerical models are powerful tools for simulating the wide range of interactions between flow field and bedform morphology. Restricted and inconsistent measurements, combined with limited computational resources, typically make this exercise complex and uncertain. In the present study, we investigate the impact of proposed land reclamation within a wave-dominated bay in New Zealand. For this purpose, we first calibrated our morphological model based on the long-term evolution of the bay resulting from land reclamation carried out in the 1950s. This included the application of sedimentological spin-up and reduction techniques based on historical bathymetry datasets. The updated bathymetry, including the proposed modifications of the bay, was then used to predict the effect of the proposed land reclamation on the wave climate and morphology of the bay after one decade. We show that reshaping the bay induces a distinct symmetrical response of the shoreline which likely will modify the nearshore wave patterns and consequently recreational activities in the area.

Keywords: coastal waves, impact of land reclamation, long-term coastal evolution, morphodynamic modeling

Procedia PDF Downloads 145

Authors: Pitigalage Chamath Chandira Peiris

Abstract:

A widely researched domain in the field of image processing in recent times has been single image super-resolution, which tries to restore a high-resolution image from a single low-resolution image. Many more single image super-resolution efforts have been completed utilizing equally traditional and deep learning methodologies, as well as a variety of other methodologies. Deep learning-based super-resolution methods, in particular, have received significant interest. As of now, the most advanced image restoration approaches are based on convolutional neural networks; nevertheless, only a few efforts have been performed using Transformers, which have demonstrated excellent performance on high-level vision tasks. The effectiveness of CNN-based algorithms in image super-resolution has been impressive. However, these methods cannot completely capture the non-local features of the data. Enhancer is a simple yet powerful Transformer-based approach for enhancing the resolution of images. A method for single image super-resolution was developed in this study, which utilized an efficient and effective transformer design. This proposed architecture makes use of a locally enhanced window transformer block to alleviate the enormous computational load associated with non-overlapping window-based self-attention. Additionally, it incorporates depth-wise convolution in the feed-forward network to enhance its ability to capture local context. This study is assessed by comparing the results obtained for popular datasets to those obtained by other techniques in the domain.

Keywords: single image super resolution, computer vision, vision transformers, image restoration

Procedia PDF Downloads 78

Authors: Po-Fang Hsu, Chiching Wei

Abstract:

In this paper, we present a novel neural graph matching approach applied to document comparison. Document comparison is a common task in the legal and financial industries. In some cases, the most important differences may be the addition or omission of words, sentences, clauses, or paragraphs. However, it is a challenging task without recording or tracing the whole edited process. Under many temporal uncertainties, we explore the potentiality of our approach to proximate the accurate comparison to make sure which element blocks have a relation of edition with others. In the beginning, we apply a document layout analysis that combines traditional and modern technics to segment layouts in blocks of various types appropriately. Then we transform this issue into a problem of layout graph matching with textual awareness. Regarding graph matching, it is a long-studied problem with a broad range of applications. However, different from previous works focusing on visual images or structural layout, we also bring textual features into our model for adapting this domain. Specifically, based on the electronic document, we introduce an encoder to deal with the visual presentation decoding from PDF. Additionally, because the modifications can cause the inconsistency of document layout analysis between modified documents and the blocks can be merged and split, Sinkhorn divergence is adopted in our neural graph approach, which tries to overcome both these issues with many-to-many block matching. We demonstrate this on two categories of layouts, as follows., legal agreement and scientific articles, collected from our real-case datasets.

Keywords: document comparison, graph matching, graph neural network, modification similarity, multi-modal

Procedia PDF Downloads 151

Authors: Naoum Abderrahmane, Boumehed Meriem, Alshaqaqi Belal

Abstract:

Background subtraction algorithm is a larger used technique for detecting moving objects in video surveillance to extract the foreground objects from a reference background image. There are many challenges to test a good background subtraction algorithm, like changes in illumination, dynamic background such as swinging leaves, rain, snow, and the changes in the background, for example, moving and stopping of vehicles. In this paper, we propose an efficient and accurate background subtraction method for moving object detection in video surveillance. The main idea is to use a developed fast-independent component analysis (ICA) algorithm to separate background, noise, and foreground masks from an image sequence in practical environments. The fast-ICA algorithm is adapted and adjusted with a matrix calculation and searching for an optimum non-quadratic function to be faster and more robust. Moreover, in order to estimate the de-mixing matrix and the denoising de-mixing matrix parameters, we propose to convert all images to YCrCb color space, where the luma component Y (brightness of the color) gives suitable results. The proposed technique has been verified on the publicly available datasets CD net 2012 and CD net 2014, and experimental results show that our algorithm can detect competently and accurately moving objects in challenging conditions compared to other methods in the literature in terms of quantitative and qualitative evaluations with real-time frame rate.

Keywords: background subtraction, moving object detection, fast-ICA, de-mixing matrix

Procedia PDF Downloads 71

Authors: Chris Schoborg, Morgan C. Wang

Abstract:

In the realm of the National Football League (NFL), substantial dedication of time and effort is invested by both players and coaches in meticulously analyzing the game footage of their opponents. The primary aim is to anticipate the actions of the opposing team. Defensive players and coaches are especially focused on deciphering their adversaries' intentions to effectively counter their strategies. Acquiring insights into the specific play type and its intended direction on the field would confer a significant competitive advantage. This study establishes pre-snap information as the cornerstone for predicting both the play type (e.g., deep pass, short pass, or run) and its spatial trajectory (right, left, or center). The dataset for this research spans the regular NFL season data for all 32 teams from 2013 to 2022. This dataset is acquired using the nflreadr package, which conveniently extracts play-by-play data from NFL games and imports it into the R environment as structured datasets. In this study, we employ a recently developed machine learning algorithm, XGBoost. The final predictive model achieves an impressive lift of 2.61. This signifies that the presented model is 2.61 times more effective than random guessing—a significant improvement. Such a model has the potential to markedly enhance defensive coaches' ability to formulate game plans and adequately prepare their players, thus mitigating the opposing offense's yardage and point gains.

Keywords: lift, NFL, sports analytics, XGBoost

Procedia PDF Downloads 36

Authors: Andrei Nosov

Abstract:

This paper discusses the development directions of dialogue systems within the digitalisation of the railway industry, where technologies based on conversational AI are already potentially applied or will be applied. Conversational AI is one of the popular natural language processing (NLP) tasks, as it has great prospects for real-world applications today. At the same time, it is a challenging task as it involves many areas of NLP based on complex computations and deep insights from linguistics and psychology. In this review, we focus on dialogue systems and their implementation in the railway domain. We comprehensively review the state-of-the-art research results on dialogue systems and analyse them from three perspectives: type of problem to be solved, type of model, and type of system. In particular, from the perspective of the type of tasks to be solved, we discuss characteristics and applications. This will help to understand how to prioritise tasks. In terms of the type of models, we give an overview that will allow researchers to become familiar with how to apply them in dialogue systems. By analysing the types of dialogue systems, we propose an unconventional approach in contrast to colleagues who traditionally contrast goal-oriented dialogue systems with open-domain systems. Our view focuses on considering retrieval and generative approaches. Furthermore, the work comprehensively presents evaluation methods and datasets for dialogue systems in the railway domain to pave the way for future research. Finally, some possible directions for future research are identified based on recent research results.

Keywords: digitalisation, railway, dialogue systems, conversational AI, natural language processing, natural language understanding, natural language generation

Procedia PDF Downloads 34

Authors: Maria Amodeo, Pasquale Arpaia, Marco Buzio, Vincenzo Di Capua, Francesco Donnarumma

Abstract:

Different deep neural network architectures have been compared and tested to predict magnetic hysteresis in the context of pulsed electromagnets for experimental physics applications. Modelling quasi-static or dynamic major and especially minor hysteresis loops is one of the most challenging topics for computational magnetism. Recent attempts at mathematical prediction in this context using Preisach models could not attain better than percent-level accuracy. Hence, this work explores neural network approaches and shows that the architecture that best fits the measured magnetic field behaviour, including the effects of hysteresis and eddy currents, is the nonlinear autoregressive exogenous neural network (NARX) model. This architecture aims to achieve a relative RMSE of the order of a few 100 ppm for complex magnetic field cycling, including arbitrary sequences of pseudo-random high field and low field cycles. The NARX-based architecture is compared with the state-of-the-art, showing better performance than the classical operator-based and differential models, and is tested on a reference quadrupole magnetic lens used for CERN particle beams, chosen as a case study. The training and test datasets are a representative example of real-world magnet operation; this makes the good result obtained very promising for future applications in this context.

Keywords: deep neural network, magnetic modelling, measurement and empirical software engineering, NARX

Procedia PDF Downloads 106

Authors: Mohammed-Hareef Asunramu, Sana Hashemi, Raja Ohri, Luc Worthington, Nadia Zaman, Junkai Zhu

Abstract:

Background: The 2012 Health and Social Care Act committed to a ‘parity of esteem between mental and physical health services. Although this investment, aimed to both increase the quality of services and ensure the retention of mental health staff, questions remained regarding its ability to prevent mental health problems. One possible solution is a focus on the social determinants of health which have been shown to impact mental health. Aim: To examine the relationship between the index of multiple deprivations (IMD) and the prevalence of common mental health disorders (CMD) for CCGs in NHS England between 2019 and 2020. Design and setting: Cross-sectional analysis of 189 CCGs in NHS England. Methods: A multivariate linear regression model was utilized with CMD as outcome variable and IMD, age and ethnicity as explanatory variables. Datasets were obtained from Public Health England and the latest UK Census. Results: CCG IMD was found to have a significantly positive relationship with CMD. For every 1-point increase in IMD, CMD increases by 0.25%. Ethnicity had a significantly positive relationship with CMD. For every 1% increase in the population that identifies as BME, there is a 0.03% increase in CMD. Age had a significantly negative relationship with CMD. For every 1% increase in the population aged 60+, there is a 0.11% decrease in CMD. Conclusion: This study demonstrates that addressing mental health issues may require a multi-pronged approach. Beyond budget increases, it is essential to prioritize health equity, with careful considerations towards ethnic minorities and different age brackets.

Keywords: deprivation, health inequality, mental health, social determinants

Procedia PDF Downloads 107

Authors: Zhifei Hu, Feng Xia

Abstract:

In recent years, Graph neural network has been widely used in knowledge graph recommendation. The existing recommendation methods based on graph neural network extract information from knowledge graph through entity and relation, which may not be efficient in the way of information extraction. In order to better propose useful entity information for the current recommendation task in the knowledge graph, we propose an end-to-end Neural network Model based on multi-stream graph attentional Mechanism (MSGAT), which can effectively integrate the knowledge graph into the recommendation system by evaluating the importance of entities from both users and items. Specifically, we use the attention mechanism from the user's perspective to distil the domain nodes information of the predicted item in the knowledge graph, to enhance the user's information on items, and generate the feature representation of the predicted item. Due to user history, click items can reflect the user's interest distribution, we propose a multi-stream attention mechanism, based on the user's preference for entities and relationships, and the similarity between items to be predicted and entities, aggregate user history click item's neighborhood entity information in the knowledge graph and generate the user's feature representation. We evaluate our model on three real recommendation datasets: Movielens-1M (ML-1M), LFM-1B 2015 (LFM-1B), and Amazon-Book (AZ-book). Experimental results show that compared with the most advanced models, our proposed model can better capture the entity information in the knowledge graph, which proves the validity and accuracy of the model.

Keywords: graph attention network, knowledge graph, recommendation, information propagation

Procedia PDF Downloads 90

Authors: Yuguang Gao, Mingtao Deng

Abstract:

The integration of independently operating hospitals into a unified healthcare service system has become a strategic imperative in the pursuit of hospitals’ high-quality development. Central to the concept of group governance over such transformation, exemplified by a collaborative medical alliance, is the delineation of shared value, vision, and goals. Given the inherent disparity in capabilities among hospitals within the alliance, particularly in the treatment of different diseases characterized by Disease Related Groups (DRG) in terms of effectiveness, efficiency and resource utilization, this study aims to address the centralized decision-making of intra-hospital patient referral within the medical alliance to enhance the overall production and quality of service provided. We first introduce the notion of production utility, where a higher production utility for a hospital implies better performance in treating patients diagnosed with that specific DRG group of diseases. Then, a Discrete-Event Simulation (DES) framework is established for patient referral among hospitals, where patient flow modeling incorporates a queueing system with fixed capacities for each hospital. The simulation study begins with a two-member alliance. The pivotal strategy examined is a "whether-to-refer" decision triggered when the bed usage rate surpasses a predefined threshold for either hospital. Then, the decision encompasses referring patients to the other hospital based on DRG groups’ production utility differentials as well as bed availability. The objective is to maximize the total production utility of the alliance while minimizing patients’ average length of stay and turnover rate. Thus the parameter under scrutiny is the bed usage rate threshold, influencing the efficacy of the referral strategy. Extending the study to a three-member alliance, which could readily be generalized to multi-member alliances, we maintain the core setup while introducing an additional “which-to-refer" decision that involves referring patients with specific DRG groups to the member hospital according to their respective production utility rankings. The overarching goal remains consistent, for which the bed usage rate threshold is once again a focal point for analysis. For the two-member alliance scenario, our simulation results indicate that the optimal bed usage rate threshold hinges on the discrepancy in the number of beds between member hospitals, the distribution of DRG groups among incoming patients, and variations in production utilities across hospitals. Transitioning to the three-member alliance, we observe similar dependencies on these parameters. Additionally, it becomes evident that an imbalanced distribution of DRG diagnoses and further disparity in production utilities among member hospitals may lead to an increase in the turnover rate. In general, it was found that the intra-hospital referral mechanism enhances the overall production utility of the medical alliance compared to individual hospitals without partnership. Patients’ average length of stay is also reduced, showcasing the positive impact of the collaborative approach. However, the turnover rate exhibits variability based on parameter setups, particularly when patients are redirected within the alliance. In conclusion, the re-structuring of diagnostic disease groups within the medical alliance proves instrumental in improving overall healthcare service outcomes, providing a compelling rationale for the government's promotion of patient referrals within collaborative medical alliances.

Keywords: collaborative medical alliance, disease related group, patient referral, simulation

Procedia PDF Downloads 24

Authors: Jianhong Xiang, Rui Sun, Linyu Wang

Abstract:

In recent years, due to the continuous improvement of deep learning theory, Convolutional Neural Network (CNN) has played a great superior performance in the research of Hyperspectral Image (HSI) classification. Since HSI has rich spatial-spectral information, only utilizing a single dimensional or single size convolutional kernel will limit the detailed feature information received by CNN, which limits the classification accuracy of HSI. In this paper, we design a multi-scale CNN with MLP based on residual and attention modules (INRAM-3DCNN) for the HSI classification task. We propose to use multiple 3D convolutional kernels to extract the packet feature information and fully learn the spatial-spectral features of HSI while designing residual 3D convolutional branches to avoid the decline of classification accuracy due to network degradation. Secondly, we also design the 2D Inception module with a joint channel attention mechanism to quickly extract key spatial feature information at different scales of HSI and reduce the complexity of the 3D model. Due to the high parallel processing capability and nonlinear global action of the Multilayer Perceptron (MLP), we use it in combination with the previous CNN structure for the final classification process. The experimental results on two HSI datasets show that the proposed INRAM-3DCNN method has superior classification performance and can perform the classification task excellently.

Keywords: INRAM-3DCNN, residual, channel attention, hyperspectral image classification

Procedia PDF Downloads 42

Authors: Soumya Sajjan

Abstract:

Liver cancer is the most common cancer disease worldwide in men and women, and is one of the few cancers still on the rise. Liver disease is the 4th leading cause of death. According to new NHS (National Health Service) figures, deaths from liver diseases have reached record levels, rising by 25% in less than a decade; heavy drinking, obesity, and hepatitis are believed to be behind the rise. In this study, we focus on Development of Diagnostic Classifier for Ultrasound liver lesion. Ultrasound (US) Sonography is an easy-to-use and widely popular imaging modality because of its ability to visualize many human soft tissues/organs without any harmful effect. This paper will provide an overview of underlying concepts, along with algorithms for processing of liver ultrasound images Naturaly, Ultrasound liver lesion images are having more spackle noise. Developing classifier for ultrasound liver lesion image is a challenging task. We approach fully automatic machine learning system for developing this classifier. First, we segment the liver image by calculating the textural features from co-occurrence matrix and run length method. For classification, Support Vector Machine is used based on the risk bounds of statistical learning theory. The textural features for different features methods are given as input to the SVM individually. Performance analysis train and test datasets carried out separately using SVM Model. Whenever an ultrasonic liver lesion image is given to the SVM classifier system, the features are calculated, classified, as normal and diseased liver lesion. We hope the result will be helpful to the physician to identify the liver cancer in non-invasive method.

Keywords: segmentation, Support Vector Machine, ultrasound liver lesion, co-occurance Matrix

Procedia PDF Downloads 384

Authors: Tonislav Ivanov, Oleksii Nedashkivskyi, Denis Babeshko, Vadim Pinskiy, Matthew Putman

Abstract:

Training a machine learning model for object detection that generalizes well is known to benefit from a training dataset with diverse examples. However, training datasets usually contain many repeats of common examples of a class and lack rarely seen examples. This is due to the process commonly used during human annotation where a person would proceed sequentially through a list of images labeling a sufficiently high total number of examples. Instead, the method presented involves an active process where, after the initial labeling of several images is completed, the next subset of images for labeling is selected by an algorithm. This process of algorithmic image selection and manual labeling continues in an iterative fashion. The algorithm used for the image selection is a deep learning algorithm, based on the U-shaped architecture, which quantifies the presence of unseen data in each image in order to find images that contain the most novel examples. Moreover, the location of the unseen data in each image is highlighted, aiding the labeler in spotting these examples. Experiments performed using semiconductor wafer data show that labeling a subset of the data, curated by this algorithm, resulted in a model with a better performance than a model produced from sequentially labeling the same amount of data. Also, similar performance is achieved compared to a model trained on exhaustive labeling of the whole dataset. Overall, the proposed approach results in a dataset that has a diverse set of examples per class as well as more balanced classes, which proves beneficial when training a deep learning model.

Keywords: computer vision, deep learning, object detection, semiconductor

Procedia PDF Downloads 110

Authors: Adriana Martínez-Arias, John Josephraj Selvaraj, Luis Octavio González-Salcedo

Abstract:

In the Colombian Pacific, Dorado fish (Coryphaena hippurus L.) fisheries is of great commercial interest. However, its habitat and fisheries may be affected by climatic change especially by the actual increase in sea surface temperature. Hence, it is of interest to study the dynamics of these species fishing zones. In this study, we developed Artificial Neural Networks (ANN) models to predict Catch per Unit Effort (CPUE) as an indicator of species abundance. The model was based on four oceanographic variables (Chlorophyll a, Sea Surface Temperature, Sea Level Anomaly and Bathymetry) derived from satellite data. CPUE datasets for model training and cross-validation were obtained from logbooks of commercial fishing vessel. Sea surface Temperature for Colombian Pacific were projected under Representative Concentration Pathway (RCP) scenarios 4.5 and 8.5 using Coupled Model Intercomparison Project Phase 5 (CMIP5) and CPUE maps were created. Our results indicated that an increase in sea surface temperature reduces the potential fishing zones of this species in the Colombian Pacific. We conclude that ANN is a reliable tool for simulation of climate change effects on the potential fishing zones. This research opens a future agenda for other species that have been affected by climate change.

Keywords: climatic change, artificial neural networks, dorado fish, CPUE

Procedia PDF Downloads 219

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

Abstract:

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

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

Procedia PDF Downloads 194

Authors: Sheikh Shams Azam, Manoj Raju, Venkatesh Pagidimarri, Vamsi Kasivajjala

Abstract:

Over the past decade, there has been a steep rise in the data-driven analysis in major areas of medicine, such as clinical decision support system, survival analysis, patient similarity analysis, image analytics etc. Most of the data in the field are well-structured and available in numerical or categorical formats which can be used for experiments directly. But on the opposite end of the spectrum, there exists a wide expanse of data that is intractable for direct analysis owing to its unstructured nature which can be found in the form of discharge summaries, clinical notes, procedural notes which are in human written narrative format and neither have any relational model nor any standard grammatical structure. An important step in the utilization of these texts for such studies is to transform and process the data to retrieve structured information from the haystack of irrelevant data using information retrieval and data mining techniques. To address this problem, the authors present Q-Map in this paper, which is a simple yet robust system that can sift through massive datasets with unregulated formats to retrieve structured information aggressively and efficiently. It is backed by an effective mining technique which is based on a string matching algorithm that is indexed on curated knowledge sources, that is both fast and configurable. The authors also briefly examine its comparative performance with MetaMap, one of the most reputed tools for medical concepts retrieval and present the advantages the former displays over the latter.

Keywords: information retrieval, unified medical language system, syntax based analysis, natural language processing, medical informatics

Procedia PDF Downloads 109

Authors: Arfan Ali Nagra, Tariq Shahzad, Meshal Alharbi, Khalid Masood Khan, Muhammad Mugees Asif, Taher M. Ghazal, Khmaies Ouahada

Abstract:

Gene selection is an essential step for the classification of microarray cancer data. Gene expression cancer data (DNA microarray) facilitates computing the robust and concurrent expression of various genes. Particle swarm optimization (PSO) requires simple operators and less number of parameters for tuning the model in gene selection. The selection of a prognostic gene with small redundancy is a great challenge for the researcher as there are a few complications in PSO based selection method. In this research, a new variant of PSO (Self-inertia weight adaptive PSO) has been proposed. In the proposed algorithm, SIW-APSO-ELM is explored to achieve gene selection prediction accuracies. This new algorithm balances the exploration capabilities of the improved inertia weight adaptive particle swarm optimization and the exploitation. The self-inertia weight adaptive particle swarm optimization (SIW-APSO) is used to search the solution. The SIW-APSO is updated with an evolutionary process in such a way that each particle iteratively improves its velocities and positions. The extreme learning machine (ELM) has been designed for the selection procedure. The proposed method has been to identify a number of genes in the cancer dataset. The classification algorithm contains ELM, K- centroid nearest neighbor (KCNN), and support vector machine (SVM) to attain high forecast accuracy as compared to the start-of-the-art methods on microarray cancer datasets that show the effectiveness of the proposed method.

Keywords: microarray cancer, improved PSO, ELM, SVM, evolutionary algorithms

Procedia PDF Downloads 55

Authors: U. Datta

Abstract:

The main objective of this study is to find a suitable approach to monitor the land infrastructure growth over a period of time using multispectral satellite images. Bi-temporal change detection method is unable to indicate the continuous change occurring over a long period of time. To achieve this objective, the approach used here estimates a statistical model from series of multispectral image data over a long period of time, assuming there is no considerable change during that time period and then compare it with the multispectral image data obtained at a later time. The change is estimated pixel-wise. Statistical composite hypothesis technique is used for estimating pixel based change detection in a defined region. The generalized likelihood ratio test (GLRT) is used to detect the changed pixel from probabilistic estimated model of the corresponding pixel. The changed pixel is detected assuming that the images have been co-registered prior to estimation. To minimize error due to co-registration, 8-neighborhood pixels around the pixel under test are also considered. The multispectral images from Sentinel-2 and Landsat-8 from 2015 to 2018 are used for this purpose. There are different challenges in this method. First and foremost challenge is to get quite a large number of datasets for multivariate distribution modelling. A large number of images are always discarded due to cloud coverage. Due to imperfect modelling there will be high probability of false alarm. Overall conclusion that can be drawn from this work is that the probabilistic method described in this paper has given some promising results, which need to be pursued further.

Keywords: co-registration, GLRT, infrastructure growth, multispectral, multitemporal, pixel-based change detection

Procedia PDF Downloads 108

Authors: Zoe Goldberger, Priscilla S. Briquez, Jeffrey A. Hubbell

Abstract:

Tumor cells have mutations resulting from genetic instability that the immune system can actively recognize. Immune checkpoint immunotherapy (ICI) is commonly used in the clinic to re-activate immune reactions against mutated proteins, called neoantigens, resulting in tumor remission in cancer patients. However, only around 20% of patients show durable response to ICI. While tumor mutational burden (TMB) has been approved by the Food and Drug Administration (FDA) as a criterion for ICI therapy, the relevance of the subcellular localizations of the mutated proteins within the tumor cell has not been investigated. Here, we hypothesized that localization of mutations impacts the effect of immune responsiveness to ICI. We analyzed publicly available tumor mutation sequencing data of ICI treated patients from 3 independent datasets. We extracted the subcellular localization from the UniProtKB/Swiss-Prot database and quantified the proportion of membrane, cytoplasmic, nuclear, or secreted mutations per patient. We analyzed this information in relation to response to ICI treatment and overall survival of patients showing with 1722 ICI-treated patients that high mutational burden localized at the membrane (mTMB), correlate with ICI responsiveness, and improved overall survival in multiple cancer types. We anticipate that our results will ameliorate predictability of cancer patient response to ICI with potential implications in clinical guidelines to tailor ICI treatment. This would not only increase patient survival for those receiving ICI, but also patients’ quality of life by reducing the number of patients enduring non-effective ICI treatments.

Keywords: cancer, immunotherapy, membrane neoantigens, efficacy prediction, biomarkers

Procedia PDF Downloads 83

Authors: Fariza Yunus, Jasmee Jaafar, Zamalia Mahmud, Nurul Nisa’ Khairul Azmi, Nursalleh K. Chang, Nursalleh K. Chang

Abstract:

Variation of air temperature from one place to another is cause by air temperature controls. In general, the most important control of air temperature is elevation. Another significant independent variable in estimating air temperature is the location of meteorological stations. Distances to coastline and land use type are also contributed to significant variations in the air temperature. On the other hand, in homogeneous terrain direct interpolation of discrete points of air temperature work well to estimate air temperature values in un-sampled area. In this process the estimation is solely based on discrete points of air temperature. However, this study presents that air temperature controls also play significant roles in estimating air temperature over homogenous terrain of Peninsular Malaysia. An Inverse Distance Weighting (IDW) interpolation technique was adopted to generate continuous data of air temperature. This study compared two different datasets, observed mean monthly data of T, and estimation error of T–T’, where T’ estimated value from a multiple regression model. The multiple regression model considered eight independent variables of elevation, latitude, longitude, coastline, and four land use types of water bodies, forest, agriculture and build up areas, to represent the role of air temperature controls. Cross validation analysis was conducted to review accuracy of the estimation values. Final results show, estimation values of T–T’ produced lower errors for mean monthly mean air temperature over homogeneous terrain in Peninsular Malaysia.

Keywords: air temperature control, interpolation analysis, peninsular Malaysia, regression model, air temperature

Procedia PDF Downloads 352