Littérature scientifique sur le sujet « Automatic Stance Detection »
Créez une référence correcte selon les styles APA, MLA, Chicago, Harvard et plusieurs autres
Sommaire
Consultez les listes thématiques d’articles de revues, de livres, de thèses, de rapports de conférences et d’autres sources académiques sur le sujet « Automatic Stance Detection ».
À côté de chaque source dans la liste de références il y a un bouton « Ajouter à la bibliographie ». Cliquez sur ce bouton, et nous générerons automatiquement la référence bibliographique pour la source choisie selon votre style de citation préféré : APA, MLA, Harvard, Vancouver, Chicago, etc.
Vous pouvez aussi télécharger le texte intégral de la publication scolaire au format pdf et consulter son résumé en ligne lorsque ces informations sont inclues dans les métadonnées.
Articles de revues sur le sujet "Automatic Stance Detection"
1
Yan, Yilin, Jonathan Chen et Mei-Ling Shyu. « Efficient Large-Scale Stance Detection in Tweets ». International Journal of Multimedia Data Engineering and Management 9, no 3 (juillet 2018) : 1–16. http://dx.doi.org/10.4018/ijmdem.2018070101.
Texte intégralRésumé :
Stance detection is an important research direction which attempts to automatically determine the attitude (positive, negative, or neutral) of the author of text (such as tweets), towards a target. Nowadays, a number of frameworks have been proposed using deep learning techniques that show promising results in application domains such as automatic speech recognition and computer vision, as well as natural language processing (NLP). This article shows a novel deep learning-based fast stance detection framework in bipolar affinities on Twitter. It is noted that millions of tweets regarding Clinton and Trump were produced per day on Twitter during the 2016 United States presidential election campaign, and thus it is used as a test use case because of its significant and unique counter-factual properties. In addition, stance detection can be utilized to imply the political tendency of the general public. Experimental results show that the proposed framework achieves high accuracy results when compared to several existing stance detection methods.
2
Ghimire, Niroj, et Surendra Shrestha. « Fake News Stance Detection using Deep Neural Network ». Journal of Lumbini Engineering College 4, no 1 (7 décembre 2022) : 49–53. http://dx.doi.org/10.3126/lecj.v4i1.49366.
Texte intégralRésumé :
With the advancement of technology, fake news is more widely exposed to users. Fake news may be found on the Internet, news sources and social media platforms. The spread of the fake news has harmed both individuals and society. The way to observe fake news using the stance detection technique is the focus of this paper. Given a set of news body and headline pairs, stance detection is the task of automatic detection of relationships among pieces of text. Pre-trained GloVe word embedding is used for the word to vector representation as it can capture the inter-word semantic information. The LSTM neural network had been shown efficient in deep learning applications because it can capture sequential information of input data. In this paper, it is found that the LSTM-based encoding decoding model using pre-trained GloVe word embedding achieved 93.69% accuracy on the FNC-1 dataset.
3
Willemsen, A. T. M., F. Bloemhof et H. B. K. Boom. « Automatic stance-swing phase detection from accelerometer data for peroneal nerve stimulation ». IEEE Transactions on Biomedical Engineering 37, no 12 (1990) : 1201–8. http://dx.doi.org/10.1109/10.64463.
Texte intégral
4
Martínez, Rubén Yáñez, Guillermo Blanco et Anália Lourenço. « Spanish Corpora of tweets about COVID-19 vaccination for automatic stance detection ». Information Processing & ; Management 60, no 3 (mai 2023) : 103294. http://dx.doi.org/10.1016/j.ipm.2023.103294.
Texte intégral
5
Stede, Manfred. « Automatic argumentation mining and the role of stance and sentiment ». Journal of Argumentation in Context 9, no 1 (4 mai 2020) : 19–41. http://dx.doi.org/10.1075/jaic.00006.ste.
Texte intégralRésumé :
Abstract Argumentation mining is a subfield of Computational Linguistics that aims (primarily) at automatically finding arguments and their structural components in natural language text. We provide a short introduction to this field, intended for an audience with a limited computational background. After explaining the subtasks involved in this problem of deriving the structure of arguments, we describe two other applications that are popular in computational linguistics: sentiment analysis and stance detection. From the linguistic viewpoint, they concern the semantics of evaluation in language. In the final part of the paper, we briefly examine the roles that these two tasks play in argumentation mining, both in current practice, and in possible future systems.
6
Lidstone, Daniel E., Louise M. Porcher, Jessica DeBerardinis, Janet S. Dufek et Mohamed B. Trabia. « Concurrent Validity of an Automated Footprint Detection Algorithm to Measure Plantar Contact Area During Walking ». Journal of the American Podiatric Medical Association 109, no 6 (1 novembre 2019) : 416–25. http://dx.doi.org/10.7547/17-118.
Texte intégralRésumé :
Background: Monitoring footprints during walking can lead to better identification of foot structure and abnormalities. Current techniques for footprint measurements are either static or dynamic, with low resolution. This work presents an approach to monitor the plantar contact area when walking using high-speed videography. Methods: Footprint images were collected by asking the participants to walk across a custom-built acrylic walkway with a high-resolution digital camera placed directly underneath the walkway. This study proposes an automated footprint identification algorithm (Automatic Identification Algorithm) to measure the footprint throughout the stance phase of walking. This algorithm used coloration of the plantar tissue that was in contact with the acrylic walkway to distinguish the plantar contact area from other regions of the foot that were not in contact. Results: The intraclass correlation coefficient (ICC) demonstrated strong agreement between the proposed automated approach and the gold standard manual method (ICC = 0.939). Strong agreement between the two methods also was found for each phase of stance (ICC > 0.78). Conclusions: The proposed automated footprint detection technique identified the plantar contact area during walking with strong agreement with a manual gold standard method. This is the first study to demonstrate the concurrent validity of an automated identification algorithm to measure the plantar contact area during walking.
7
Houliston, B. R., A. F. Merry et D. T. Parry. « TADAA : Towards Automated Detection of Anaesthetic Activity ». Methods of Information in Medicine 50, no 05 (2011) : 464–71. http://dx.doi.org/10.3414/me11-02-0001.
Texte intégralRésumé :
SummaryBackground: Task analysis is a valuable research method for better understanding the activity of anaesthetists in the operating room (OR), providing evidence for designing and evaluating improvements to systems and processes. It may also assist in identifying potential error paths to adverse events, ultimately improving patient safety. Human observers are the current ‘gold standard’ for capturing task data, but they are expensive and have cognitive limitations.Objectives: Towards Automated Detection of Anaesthetic Activity (TADAA) – aims to produce an automated task analysis system, employing Radio Frequency Identification (RFID) technology to capture anaesthetists’ location, orientation and stance (LOS). This is the first stage in a scheme for automatic detection of activity.Methods: Active RFID tags were attached to anaesthetists and various objects in a high fidelity OR simulator, and anesthetic procedures performed. The anaesthetists’ LOSs were calculated using received signal strength (RSS) measurements combined with machine learning tools including Self-Organizing Maps (SOMs). These LOSs were compared to those derived from video recordings.Results: SOM clustering was effective at determining anaesthetists’ LOS from RSS data for each procedure. However cross-procedure comparison was less reliable, probably because of changes in the environment.Conclusions: Active RFID tags provide potentially useful information on LOS at a low cost and with minimal impact on the work environment. Machine learning techniques may be employed to handle the variable nature of RFID’s radio signals. Work on mapping LOS data to activities will involve integration with other sensors and task analysis techniques.
8
Omero, Paolo, Massimiliano Valotto, Riccardo Bellana, Ramona Bongelli, Ilaria Riccioni, Andrzej Zuczkowski et Carlo Tasso. « Writer’s uncertainty identification in scientific biomedical articles : a tool for automatic if-clause tagging ». Language Resources and Evaluation 54, no 4 (11 juin 2020) : 1161–81. http://dx.doi.org/10.1007/s10579-020-09491-8.
Texte intégralRésumé :
Abstract In a previous study, we manually identified seven categories (verbs, non-verbs, modal verbs in the simple present, modal verbs in the conditional mood, if, uncertain questions, and epistemic future) of Uncertainty Markers (UMs) in a corpus of 80 articles from the British Medical Journal randomly sampled from a 167-year period (1840–2007). The UMs detected on the base of an epistemic stance approach were those referring only to the authors of the articles and only in the present. We also performed preliminary experiments to assess the manual annotated corpus and to establish a baseline for the UMs automatic detection. The results of the experiments showed that most UMs could be recognized with good accuracy, except for the if-category, which includes four subcategories: if-clauses in a narrow sense; if-less clauses; as if/as though; if and whether introducing embedded questions. The unsatisfactory results concerning the if-category were probably due to both its complexity and the inadequacy of the detection rules, which were only lexical, not grammatical. In the current article, we describe a different approach, which combines grammatical and syntactic rules. The performed experiments show that the identification of uncertainty in the if-category has been largely double improved compared to our previous results. The complex overall process of uncertainty detection can greatly profit from a hybrid approach which should combine supervised Machine learning techniques with a knowledge-based approach constituted by a rule-based inference engine devoted to the if-clause case and designed on the basis of the above mentioned epistemic stance approach.
9
Karande, Hema, Rahee Walambe, Victor Benjamin, Ketan Kotecha et TS Raghu. « Stance detection with BERT embeddings for credibility analysis of information on social media ». PeerJ Computer Science 7 (14 avril 2021) : e467. http://dx.doi.org/10.7717/peerj-cs.467.
Texte intégralRésumé :
The evolution of electronic media is a mixed blessing. Due to the easy access, low cost, and faster reach of the information, people search out and devour news from online social networks. In contrast, the increasing acceptance of social media reporting leads to the spread of fake news. This is a minacious problem that causes disputes and endangers the societal stability and harmony. Fake news spread has gained attention from researchers due to its vicious nature. proliferation of misinformation in all media, from the internet to cable news, paid advertising and local news outlets, has made it essential for people to identify the misinformation and sort through the facts. Researchers are trying to analyze the credibility of information and curtail false information on such platforms. Credibility is the believability of the piece of information at hand. Analyzing the credibility of fake news is challenging due to the intent of its creation and the polychromatic nature of the news. In this work, we propose a model for detecting fake news. Our method investigates the content of the news at the early stage i.e., when the news is published but is yet to be disseminated through social media. Our work interprets the content with automatic feature extraction and the relevance of the text pieces. In summary, we introduce stance as one of the features along with the content of the article and employ the pre-trained contextualized word embeddings BERT to obtain the state-of-art results for fake news detection. The experiment conducted on the real-world dataset indicates that our model outperforms the previous work and enables fake news detection with an accuracy of 95.32%.
10
Briggs, Eloise V., et Claudia Mazzà. « Automatic methods of hoof-on and -off detection in horses using wearable inertial sensors during walk and trot on asphalt, sand and grass ». PLOS ONE 16, no 7 (26 juillet 2021) : e0254813. http://dx.doi.org/10.1371/journal.pone.0254813.
Texte intégralRésumé :
Detection of hoof-on and -off events are essential to gait classification in horses. Wearable sensors have been endorsed as a convenient alternative to the traditional force plate-based method. The aim of this study was to propose and validate inertial sensor-based methods of gait event detection, reviewing different sensor locations and their performance on different gaits and exercise surfaces. Eleven horses of various breeds and ages were recruited to wear inertial sensors attached to the hooves, pasterns and cannons. Gait events detected by pastern and cannon methods were compared to the reference, hoof-detected events. Walk and trot strides were recorded on asphalt, grass and sand. Pastern-based methods were found to be the most accurate and precise for detecting gait events, incurring mean errors of between 1 and 6ms, depending on the limb and gait, on asphalt. These methods incurred consistent errors when used to measure stance durations on all surfaces, with mean errors of 0.1 to 1.16% of a stride cycle. In conclusion, the methods developed and validated here will enable future studies to reliably detect equine gait events using inertial sensors, under a wide variety of field conditions.
Thèses sur le sujet "Automatic Stance Detection"
1
Dias, Marcelo dos Santos. « Detecção não supervisionada de posicionamento em textos de tweets ». reponame:Biblioteca Digital de Teses e Dissertações da UFRGS, 2017. http://hdl.handle.net/10183/169098.
Texte intégralRésumé :
Detecção de posicionamento é a tarefa de automaticamente identificar se o autor de um texto é favorável, contrário, ou nem favorável e nem contrário a uma dada proposição ou alvo. Com o amplo uso do Twitter como plataforma para expressar opiniões e posicionamentos, a análise automatizada deste conteúdo torna-se de grande valia para empresas, organizações e figuras públicas. Em geral, os trabalhos que exploram tal tarefa adotam abordagens supervisionadas ou semi-supervisionadas. O presente trabalho propõe e avalia um processo não supervisionado de detecção de posicionamento em textos de tweets que tem como entrada apenas o alvo e um conjunto de tweets a rotular e é baseado em uma abordagem híbrida composta por 2 etapas: a) rotulação automática de tweets baseada em um conjunto de heurísticas e b) classificação complementar baseada em aprendizado supervisionado de máquina. A proposta tem êxito quando aplicada a figuras públicas, superando o estado-da-arte. Além disso, são avaliadas alternativas no intuito de melhorar seu desempenho quando aplicada a outros domínios, revelando a possibilidade de se empregar estratégias tais como o uso de alvos e perfis semente dependendo das características de cada domínio.Stance Detection is the task of automatically identifying if the author of a text is in favor of the given target, against the given target, or whether neither inference is likely. With the wide use of Twitter as a platform to express opinions and stances, the automatic analysis of this content becomes of high regard for companies, organizations and public figures. In general, works that explore such task adopt supervised or semi-supervised approaches. The present work proposes and evaluates a non-supervised process to detect stance in texts of tweets that has as entry only the target and a set of tweets to classify and is based on a hybrid approach composed by 2 stages: a) automatic labelling of tweets based on a set of heuristics and b) complementary classification based on supervised machine learning. The proposal succeeds when applied to public figures, overcoming the state-of-the-art. Beyond that, some alternatives are evaluated with the intention of increasing the performance when applied to other domains, revealing the possibility of use of strategies such as using seed targets and profiles depending on each domain characteristics.
2
Kanhere, Neeraj Krantiveer. « Vision-based detection, tracking and classification of vehicles using stable features with automatic camera calibration ». Connect to this title online, 2008. http://etd.lib.clemson.edu/documents/1219861574/.
Texte intégral
3
cignarella, alessandra teresa. « Dependency Syntax in the Automatic Detection of Irony and Stance ». Doctoral thesis, 2021. http://hdl.handle.net/2318/1873580.
Texte intégral
4
Renda, Alessandro. « Algorithms and techniques for data stream mining ». Doctoral thesis, 2021. http://hdl.handle.net/2158/1235915.
Texte intégralRésumé :
The abstraction of data streams encompasses a vast range of diverse applications that continuously generate data and therefore require dedicated algorithms and approaches for exploitation and mining. In this framework both unsupervised and supervised approaches are generally employed, depending on the task and on the availability of annotated data. This thesis proposes novel algorithms and techniques specifically tailored for the streaming setting and for knowledge discovery from Social Networks.
In the first part of this work we propose a novel clustering algorithm for data streams. Our investigation stems from the discussion of general challenges posed by cluster analysis and of those purely related to the streaming setting. First, we propose SF-DBSCAN (streaming fuzzy DBSCAN) a preliminary solution conceived as an extension of the popular DBSCAN algorithm. SF-DBSCAN handles the arrival of new objects and continuously updates the clustering result by taking advantage of concepts from fuzzy set theory. However, it gives equal importance to every collected object and therefore is not suitable to manage unbounded data streams and to adapt to evolving settings. Then, we introduce TSF-DBSCAN, a novel "temporal" adaptation of streaming fuzzy DBSCAN: it overcomes the limits of the previous proposal and proves to be effective in handling evolving and potentially unbounded data streams, discovering clusters with fuzzy overlapping borders.
In the second part of the thesis we explore a supervised learning application: the goal of our analysis is to discover the public opinion towards the vaccination topic in Italy, by exploiting the popular Twitter platform as data source. First, we discuss the design and development of a system for stance detection from text. The deployment of the classification model for the online monitoring of the public opinion, however, cannot ignore that tweets can be seen as a particular form of a temporal data stream. Then, we discuss the importance of leveraging user-related information, which enables the design of a set of techniques aimed at deepening and enhancing the analysis. Finally, we compare different learning schemes for addressing concept-drift, i.e. a change in the underlying data distribution, in a dynamic environment affected by the occurrence of real world context-related events. In this case study and throughout the thesis, the proposal of algorithms and techniques is supported by in-depth experimental analysis.
Chapitres de livres sur le sujet "Automatic Stance Detection"
1
Zhang, Yanjing, Jianming Cui et Ming Liu. « Research on Adversarial Patch Attack Defense Method for Traffic Sign Detection ». Dans Communications in Computer and Information Science, 199–210. Singapore : Springer Nature Singapore, 2022. http://dx.doi.org/10.1007/978-981-19-8285-9_15.
Texte intégralRésumé :
AbstractAccurate and stable traffic sign detection is a key technology to achieve L3 driving automation, and its performance has been significantly improved by the development of deep learning technology in recent years. However, the current traffic sign detection has inadequate difficulty resisting anti-attack ability and even does not have basic defense capability. To solve this critical issue, an adversarial patch attack defense model IYOLO-TS is proposed in this paper. The main innovation is to simulate the conditions of traffic signs being partially damaged, obscured or maliciously modified in real world by training the attack patches, and then add the attacked classes in the last layer of the YOLOv2 which are corresponding to the original detection categories, and finally the attack patch obtained from the training is used to complete the adversarial training of the detection model. The attack patch is obtained by first using RP2 algorithm to attack the detection model and then training on the blank patch. In order to verify the defense effective of the proposed IYOLO-TS model, we constructed a patch dataset LISA-Mask containing 50 different mask generation patches of 33000 sheets, and then training dataset by combining LISA and LISA-Mask datasets. The experiment results show that the mAP of the proposed IYOLO-TS is up to 98.12%. Compared with YOLOv2, it improved the defense ability against patch attacks and has the real-time detection ability. It can be considered that the proposed method has strong practicality and achieves a tradeoff between design complexity and efficiency.
2
Wang, Yan. « Fast Detection and Automatic Parameter Estimation of a Gravitational Wave Signal with a Novel Method ». Dans First-stage LISA Data Processing and Gravitational Wave Data Analysis, 205–15. Cham : Springer International Publishing, 2015. http://dx.doi.org/10.1007/978-3-319-26389-2_12.
Texte intégral
3
Yan, Yilin, Jonathan Chen et Mei-Ling Shyu. « Efficient Large-Scale Stance Detection in Tweets ». Dans Deep Learning and Neural Networks, 667–83. IGI Global, 2020. http://dx.doi.org/10.4018/978-1-7998-0414-7.ch037.
Texte intégralRésumé :
Stance detection is an important research direction which attempts to automatically determine the attitude (positive, negative, or neutral) of the author of text (such as tweets), towards a target. Nowadays, a number of frameworks have been proposed using deep learning techniques that show promising results in application domains such as automatic speech recognition and computer vision, as well as natural language processing (NLP). This article shows a novel deep learning-based fast stance detection framework in bipolar affinities on Twitter. It is noted that millions of tweets regarding Clinton and Trump were produced per day on Twitter during the 2016 United States presidential election campaign, and thus it is used as a test use case because of its significant and unique counter-factual properties. In addition, stance detection can be utilized to imply the political tendency of the general public. Experimental results show that the proposed framework achieves high accuracy results when compared to several existing stance detection methods.
4
Barker, Zoe E., Nick J. Bell, Jonathan R. Amory et Edward A. Codling. « Developments in automated systems for monitoring livestock health : lameness ». Dans Advances in precision livestock farming, 247–88. Burleigh Dodds Science Publishing, 2022. http://dx.doi.org/10.19103/as.2021.0090.10.
Texte intégralRésumé :
Lameness is a key issue for commercially managed livestock species such as dairy cattle. Lameness can lead to significant economic impacts for farmers and to ongoing health and welfare problems for animals. However, lameness detection remains a difficult and time-consuming task and there is a need for reliable automated methods to support farmers, especially where herd sizes are large. In this chapter we provide an overview of lameness and its impacts on animal health and behaviour, with a particular focus on dairy cows. We review existing methods for manual and automated detection of lameness, including approaches that detect changes and abnormalities in the gait or stance of the animal, and methods that directly or indirectly detect changes in individual and social behaviour. We highlight approaches that use automated technology such as video, accelerometers and spatial positioning systems, and discuss methods to analyse trends and signals in these data.
5
Zou, Dehua, Zhipeng Jiang, Minmin Qiao, Lanlan Liu, Wei Jiang et Qianwei Yi. « Analysis and Simulation of Dynamic Characteristics for Multi-Split Transmission Line Splicing Pipe Flaw Detection Robot ». Dans Advances in Transdisciplinary Engineering. IOS Press, 2022. http://dx.doi.org/10.3233/atde220494.
Texte intégralRésumé :
The splicing pipe is an important hardware on multi-split transmission lines. The quality of its crimping and internal structure directly affects the normal and stable operation of the transmission line. The online flaw detection robot for the splicing pipe is an important means to realize the automatic flaw detection of the splicing pipe. This paper on the basis of proposing a robot mechanism configuration suitable for multi-split transmission line splicing pipe flaw detection, aiming at the matching relationship between the imaging plate movement and the mechanism joint driving torque during the splicing pipe flaw detection process, a flaw detection method is established by using the Lagrangian method. The dynamic model of the operation process of the imaging board is installed, and the relationship between the joint motion of each mechanism and the pose change of the imaging board in the area under the jurisdiction of the four-split wire is analyzed. The displacement and velocity curves of each joint motion of the robot can be obtained. From the simulation results, it can be known that the joint motion of the robot is continuous and stable, and the motion of the imaging plate is driven continuously and smoothly, which realizes the coordinated pose control between the imaging probe and the imaging plate. Coupling pipe flaw detection operation, the research in this paper has important theoretical significance and practical application value for the design of the transmission line joint pipe flaw detection robot system, especially the joint drive mechanism design.
6
« Intrusion Detection Systems for (Wireless) Automation Systems ». Dans The State of the Art in Intrusion Prevention and Detection, 449–66. Auerbach Publications, 2014. http://dx.doi.org/10.1201/b16390-23.
Texte intégral
7
Tang, Yongmei, Xiangyun Liao, Weixin Si et Zhigang Ning. « Prediction of Alzheimer’s Disease Based on Coordinate-Dense Attention Network ». Dans Proceedings of CECNet 2021. IOS Press, 2021. http://dx.doi.org/10.3233/faia210390.
Texte intégralRésumé :
Alzheimer’s disease (AD) is a degenerative disease of the nervous system. Mild cognitive impairment (MCI) is a condition between brain aging and dementia. The prediction will be divided into stable sMCI and progressive pMCI as a binary task. Structural magnetic resonance imaging (sMRI) can describe structural changes in the brain and provide a diagnostic method for the detection and early prevention of Alzheimer’s disease. In this paper, an automatic disease prediction scheme based on MRI was designed. A dense convolutional network was used as the basic model. By adding a channel attention mechanism to the model, significant feature information in MRI images was extracted, and the unimportant features were ignored or suppressed. The proposed framework is compared with the most advanced methods, and better results are obtained.
8
Sklar, Larry A. « The Future of Flow Cytometry in Biotechnology : The Response to Diversity and Complexity ». Dans Flow Cytometry for Biotechnology. Oxford University Press, 2005. http://dx.doi.org/10.1093/oso/9780195183146.003.0004.
Texte intégralRésumé :
Flow cytometry is a mature technology: Instruments recognizable as having elements of modern flow cytometers date back at least 30 years. There are many good sources for information about the essential features of flow cytometers, how they operate, and how they have been used. For the purposes of this book, it is necessary to know that flow cytometers have fluidic, optical, electronic, computational, and mechanical features. The main function of the fluidic components is to use hydrodynamic focusing to create a stable particle stream in which particles are aligned in single file within a sheath stream, so that the particles can be analyzed and sorted. The main functions of the optical components are to allow the particles to be illuminated by one or more lasers or other light sources and to allow scattered light as well as multiple fluorescence signals to be resolved and be routed to individual detectors. The electronics coordinate these functions, from the acquisition of the signals (pulse collection, pulse analysis, triggering, time delay, data, gating, detector control) to forming and charging individual droplets, and to making sort decisions. The computational components are directed at postacquisition data display and analysis, analysis of multivariate populations and multiplexing assays, and calibration and analysis of time-dependent cell or reaction phenomena. Mechanical components are now being integrated with flow cytometers to handle plates of samples and to coordinate automation such as the movement of a cloning tray with the collection of the droplets. The reader is directed to a concise description of these processes in Robinson’s article in the Encyclopedia of Biomaterials and Biomedical Engineering. This book was conceived of to provide a perspective on the future of flow cytometry, and particularly its application to biotechnology. It attempts to answer the question I heard repeatedly, especially during my association with the National Institutes of Health–funded National Flow Cytometry Resource at Los Alamos National Laboratory: What is the potential for innovation in flow cytometer design and application? This volume brings together those approaches that identify the unique contributions of flow cytometry to the modern world of biotechnology.
Actes de conférences sur le sujet "Automatic Stance Detection"
1
Gupta, Anuradha, et Shikha Mehta. « Automatic Stance Detection for Twitter Data ». Dans 2022 1st International Conference on Informatics (ICI). IEEE, 2022. http://dx.doi.org/10.1109/ici53355.2022.9786920.
Texte intégral
2
Mohtarami, Mitra, Ramy Baly, James Glass, Preslav Nakov, Lluís Màrquez et Alessandro Moschitti. « Automatic Stance Detection Using End-to-End Memory Networks ». Dans Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics : Human Language Technologies, Volume 1 (Long Papers). Stroudsburg, PA, USA : Association for Computational Linguistics, 2018. http://dx.doi.org/10.18653/v1/n18-1070.
Texte intégral
3
Conforti, Costanza, Mohammad Taher Pilehvar et Nigel Collier. « Towards Automatic Fake News Detection : Cross-Level Stance Detection in News Articles ». Dans Proceedings of the First Workshop on Fact Extraction and VERification (FEVER). Stroudsburg, PA, USA : Association for Computational Linguistics, 2018. http://dx.doi.org/10.18653/v1/w18-5507.
Texte intégral
4
Christhie, William, Julio C. S. Reis, Fabrício Benevenuto Mirella M. Moro et Virgílio Almeida. « Detecção de Posicionamento em Tweets sobre Política no Contexto Brasileiro ». Dans VII Brazilian Workshop on Social Network Analysis and Mining. Sociedade Brasileira de Computação - SBC, 2018. http://dx.doi.org/10.5753/brasnam.2018.3583.
Texte intégralRésumé :
Opinions shared over the Web constitute big volumes of data. Moreover, they may contain stances that are expressed directly or indirectly. Hence, stance detection may help to define the polarity related to a target idea. Here, we present the characterization of a broad set of tweets in Portuguese about the 2018 Brazilian presidential race. Such a set serves as the basis for automatic stance detection through a semi-supervised approach. In our evaluation, we find clues on the presence of bots in the network. We also evaluate three classifiers with paired statistical test, and our results present F-Measure above 94%.
5
Santos, Patricia D., et Denise H. Goya. « Automatic Twitter Stance Detection on Politically Controversial Issues : A Study on Covid-19’s CPI ». Dans Encontro Nacional de Inteligência Artificial e Computacional. Sociedade Brasileira de Computação - SBC, 2021. http://dx.doi.org/10.5753/eniac.2021.18281.
Texte intégralRésumé :
Prever o posicionamento de usuários de mídias sociais sobre um tópico tema pode ser desafiador, especialmente para casos não supervisionados. Neste trabalho foram utilizadas postagens retuitadas como elementos de interação dos usuários, para calcular as semelhanças entre os mais ativos dentro de uma discussão. A detecção de posicionamento para esses usuários foi realizada usando técnicas de redução de dimensionalidade e clusterização, modelagem de tópicos usando embeddings contextualizados, e rotulação automática de clusters baseada em termos recorrentes em cada grupo. Esta abordagem produziu um pequeno número de clusters de usuários (entre 2 e 3), com uniformidade na rotulação dos usuários em diferentes bases superior a 98%.
6
Kotelnikov, Evgeny, Natalia Loukachevitch, Irina Nikishina et Alexander Panchenko. « RuArg-2022 : Argument Mining Evaluation ». Dans Dialogue. RSUH, 2022. http://dx.doi.org/10.28995/2075-7182-2022-21-333-348.
Texte intégralRésumé :
Argumentation analysis is a field of computational linguistics that studies methods for extracting arguments from texts and the relationships between them, as well as building argumentation structure of texts. This paper is a report of the organizers on the first competition of argumentation analysis systems dealing with Russian language texts within the framework of the Dialogue conference. During the competition, the participants were offered two tasks: stance detection and argument classification. A corpus containing 9,550 sentences (comments on social media posts) on three topics related to the COVID-19 pandemic (vaccination, quarantine, and wearing masks) was prepared, annotated, and used for training and testing. The system that won the first place in both tasks used the NLI (Natural Language Inference) variant of the BERT architecture, automatic translation into English to apply a specialized BERT model, retrained on Twitter posts discussing COVID-19, as well as additional masking of target entities. This system showed the following results: for the stance detection task an F1-score of 0.6968, for the argument classification task an F1-score of 0.7404. We hope that the prepared dataset and baselines will help to foster further research on argument mining for the Russian language.
7
Dow, Blaine, Pierrick Ferrando, N. I. Abolins, Tom Leonard, Ahmed Abuelaish, Nicolas Gallinal Cuenca, Jerry Hansen et Freddy Rojas Rodriguez. « Advancing Influx Detection Toward Automated Well Control ». Dans IADC/SPE International Drilling Conference and Exhibition. SPE, 2022. http://dx.doi.org/10.2118/208750-ms.
Texte intégralRésumé :
Abstract Beginning in the early 2000s, the industry has applied a lot of attention, effort and protocol toward drilling automation routines in the interest of advancing efficiency and consistency, and most of all, safety. The routines that have grown in maturity and gone mainstream blend data interpretation from advanced algorithms and apply actions to downhole and/or surface machines to achieve the desired outcome. Directional drilling stands our as one field that has proven successful. Well control has been pursued with less enthusiasm in the automation space. Some effort has gone into automating a segmented well control workflow. A fully closed loop automated workflow that detects and controls has not reached commercial maturity yet. A key challenge for the pursuit of a fully closed loop influx management routine is detection. The data signatures that present themselves remain difficult to interpret consistently. This may be due to the wide range of variables that influence the interpretation. The drilling fluid, reservoir fluid and pressure, and the drilling state when the influx initiates are only a few. This paper will describe and demonstrate new technology that improves upon a process used since 2014, targeting the most important step of advancing early kick detection. This new generation of algorithms and workflows reduce gain and loss detection thresholds, can enable kick tolerance reduction, and will also minimizes false alarms. As dynamic pressure management and primary well control techniques become more complex, so to, do the challenges associated with the prompt and accurate detection of gains and losses. The new algorithms and workflows have been developed to ensure compatibility with surface backpressure Managed Pressure Drilling (MPD) systems, as well other techniques such as riser annulus height control. With increasingly stringent environment regulations being implemented worldwide, it is becoming essential for drilling operations to not only detect gains and losses, but to also monitor subsea equipment to avoid unplanned releases of drilling fluid. A side benefit of accurate gain and loss detection also enables detection of such leaks from subsea equipment. The case studies presented here will focus on the results obtained while running this technology in both conventional mode, and MPD mode. Additionally, a case study will describe how the detection application is then coupled with applied backpressure MPD and previously developed work automating circulation of the influx in SPE-194089-MS, attempting to fully close the automated influx management loop. Beyond the technology, human factors remain a barrier. To break these barriers, validation under robust conditions is essential.
8
Krikor, Ara, Shreepad Purushottam Khambete, Paulinus Abhyudaya Bimastianto, Michael Bradley Cotten, Lucian Toader, Fernando Jose Landaeta Rivas, Shahid Yakubbhai Duivala et al. « Machine Learning Delivers Automated Feedback on Real Time Key Performance Indicators During Drilling Operations ». Dans ADIPEC. SPE, 2022. http://dx.doi.org/10.2118/211753-ms.
Texte intégralRésumé :
Abstract Drilling, tripping and running casing represents approximately fifty percent of the total well time, where the connection time KPI is the common performance indicator for those operations. Therefore, enabling real-time monitoring on drilling weight to weight and tripping connection time KPI's will add significant value through well time saving. The objective of this paper is to discuss the detailed implementation of machine learning to automate the detection and computation of the KPI's in real-time. The existing method for drilling performance monitoring requires extensive human data interpretation to calibrate the parameters required in this process. To overcome the complexity and reduce the human interaction, the automated Rig state and Drill state activity level were implemented based on Machine Learning (ML). The algorithm learns from the previous connections, drilling stand or tripping conditions to define the thresholds necessary to determine the current rig operation. With automatic rig activity detection, statistics to monitor the performance can be done in a systematic way. As a result, consistency of computation allows to compare performance and to improve it. The automated process using Machine Learning (ML) delivered consistent and powerful real time KPI computation, this helped to eliminate any human interpretation. This enabled real-time performance analysis delivery to rig site operations team. The machine learning model results were compared with the existing performance engine output and the comparison showed accurate and identical rig state/drill state detection and KPI's computation. The initial potential time saving with the implementation of this methodology is estimated around 15%, this was achieved through performance improvement on drilling and tripping connection KPI's. Further potential time saving can be achieved by extending the concept to track casing and liner running performance monitoring and other relevant drilling activities. This project introduces novel Rig state detection and KPI computation based on automated machine leaning model, demonstrating the benefits through improvement in drilling performance. The approach allows operators to mitigate data issues related with human interpretation and demonstrate real-time, high frequency and high-accuracy KPI's to significantly improve the drilling performance.
9
Al Radi, Muaz, Hamad Karki, Naoufel Werghi, Sajid Javed et Jorge Dias. « Video Analysis of Flare Stacks with an Autonomous Low-Cost Aerial System ». Dans ADIPEC. SPE, 2022. http://dx.doi.org/10.2118/211007-ms.
Texte intégralRésumé :
Abstract Objectives/Scope The inspection of flare stacks operation is a challenging task that requires time and human effort. Flare stack systems undergo various types of faults, including cracks in the flare stack's structure and incomplete combustion of the flared gas, which need to be monitored in a timely manner to avoid costly and dangerous accidents. Automating this inspection process via the application of autonomous robotic systems is a promising solution for minimizing the involved hazards and costs. Methods, Procedures, Process In this work, we present an autonomous low-cost aerial system to be used as a flare stack inspection system. The proposed UAV system uses the visual signal obtained from an on-board camera for analyzing the observed scene, guiding the UAV's movement, and obtaining spectral data measurements from the flare during operation of the inspected system. The UAV system uses a deep learning detection network for detecting the flare stack's structure and extracting visual features. These visual features are used simultaneously for guiding the UAV's movement along the structure inspection mission and computing combustion-related measures. Results, Observations, Conclusions The deep learning network was trained for inspecting the structure and monitoring the operation of the flare stack system. Simulations were conducted for evaluating the performance of the proposed structure and operation inspection technique and real images of flare stacks were used for testing the initial phases of the prototype. The developed system could autonomously collect an image database of the flare stack's structure for inspection purposes. Moreover, the trained deep learning detector could accurately detect combustion-related objects, such as flame and smoke, to give a conclusion about the current state of the flare stack system. Novel/Additive Information The current system introduces a novelty to combine 3D navigation using visual servoing and a deep learning detection network in an autonomous UAV system for automating the process of flare stacks inspection and monitoring. The implementation of such system is expected to lower the cost and minimize the human resource risks of flare stack inspection processes.
10
Kotonya, Neema, et Francesca Toni. « Gradual Argumentation Evaluation for Stance Aggregation in Automated Fake News Detection ». Dans Proceedings of the 6th Workshop on Argument Mining. Stroudsburg, PA, USA : Association for Computational Linguistics, 2019. http://dx.doi.org/10.18653/v1/w19-4518.
Texte intégralRapports d'organisations sur le sujet "Automatic Stance Detection"
1
Berney, Ernest, Andrew Ward et Naveen Ganesh. First generation automated assessment of airfield damage using LiDAR point clouds. Engineer Research and Development Center (U.S.), mars 2021. http://dx.doi.org/10.21079/11681/40042.
Texte intégralRésumé :
This research developed an automated software technique for identifying type, size, and location of man-made airfield damage including craters, spalls, and camouflets from a digitized three-dimensional point cloud of the airfield surface. Point clouds were initially generated from Light Detection and Ranging (LiDAR) sensors mounted on elevated lifts to simulate aerial data collection and, later, an actual unmanned aerial system. LiDAR data provided a high-resolution, globally positioned, and dimensionally scaled point cloud exported in a LAS file format that was automatically retrieved and processed using volumetric detection algorithms developed in the MATLAB software environment. Developed MATLAB algorithms used a three-stage filling technique to identify the boundaries of craters first, then spalls, then camouflets, and scaled their sizes based on the greatest pointwise extents. All pavement damages and their locations were saved as shapefiles and uploaded into the GeoExPT processing environment for visualization and quality control. This technique requires no user input between data collection and GeoExPT visualization, allowing for a completely automated software analysis with all filters and data processing hidden from the user.
2
Yan, Yujie, et Jerome F. Hajjar. Automated Damage Assessment and Structural Modeling of Bridges with Visual Sensing Technology. Northeastern University, mai 2021. http://dx.doi.org/10.17760/d20410114.
Texte intégralRésumé :
Recent advances in visual sensing technology have gained much attention in the field of bridge inspection and management. Coupled with advanced robotic systems, state-of-the-art visual sensors can be used to obtain accurate documentation of bridges without the need for any special equipment or traffic closure. The captured visual sensor data can be post-processed to gather meaningful information for the bridge structures and hence to support bridge inspection and management. However, state-of-the-practice data postprocessing approaches require substantial manual operations, which can be time-consuming and expensive. The main objective of this study is to develop methods and algorithms to automate the post-processing of the visual sensor data towards the extraction of three main categories of information: 1) object information such as object identity, shapes, and spatial relationships - a novel heuristic-based method is proposed to automate the detection and recognition of main structural elements of steel girder bridges in both terrestrial and unmanned aerial vehicle (UAV)-based laser scanning data. Domain knowledge on the geometric and topological constraints of the structural elements is modeled and utilized as heuristics to guide the search as well as to reject erroneous detection results. 2) structural damage information, such as damage locations and quantities - to support the assessment of damage associated with small deformations, an advanced crack assessment method is proposed to enable automated detection and quantification of concrete cracks in critical structural elements based on UAV-based visual sensor data. In terms of damage associated with large deformations, based on the surface normal-based method proposed in Guldur et al. (2014), a new algorithm is developed to enhance the robustness of damage assessment for structural elements with curved surfaces. 3) three-dimensional volumetric models - the object information extracted from the laser scanning data is exploited to create a complete geometric representation for each structural element. In addition, mesh generation algorithms are developed to automatically convert the geometric representations into conformal all-hexahedron finite element meshes, which can be finally assembled to create a finite element model of the entire bridge. To validate the effectiveness of the developed methods and algorithms, several field data collections have been conducted to collect both the visual sensor data and the physical measurements from experimental specimens and in-service bridges. The data were collected using both terrestrial laser scanners combined with images, and laser scanners and cameras mounted to unmanned aerial vehicles.
3
Fang, Chen. Unsettled Issues in Vehicle Autonomy, Artificial Intelligence, and Human-Machine Interaction. SAE International, avril 2021. http://dx.doi.org/10.4271/epr2021010.
Texte intégralRésumé :
Artificial intelligence (AI)-based solutions are slowly making their way into our daily lives, integrating with our processes to enhance our lifestyles. This is major a technological component regarding the development of autonomous vehicles (AVs). However, as of today, no existing, consumer ready AV design has reached SAE Level 5 automation or fully integrates with the driver. Unsettled Issues in Vehicle Autonomy, AI and Human-Machine Interaction discusses vital issues related to AV interface design, diving into speech interaction, emotion detection and regulation, and driver trust. For each of these aspects, the report presents the current state of research and development, challenges, and solutions worth exploring.
4
Seginer, Ido, Louis D. Albright et Robert W. Langhans. On-line Fault Detection and Diagnosis for Greenhouse Environmental Control. United States Department of Agriculture, février 2001. http://dx.doi.org/10.32747/2001.7575271.bard.
Texte intégralRésumé :
Background Early detection and identification of faulty greenhouse operation is essential, if losses are to be minimized by taking immediate corrective actions. Automatic detection and identification would also free the greenhouse manager to tend to his other business. Original objectives The general objective was to develop a method, or methods, for the detection, identification and accommodation of faults in the greenhouse. More specific objectives were as follows: 1. Develop accurate systems models, which will enable the detection of small deviations from normal behavior (of sensors, control, structure and crop). 2. Using these models, develop algorithms for an early detection of deviations from the normal. 3. Develop identifying procedures for the most important faults. 4. Develop accommodation procedures while awaiting a repair. The Technion team focused on the shoot environment and the Cornell University team focused on the root environment. Achievements Models: Accurate models were developed for both shoot and root environment in the greenhouse, utilizing neural networks, sometimes combined with robust physical models (hybrid models). Suitable adaptation methods were also successfully developed. The accuracy was sufficient to allow detection of frequently occurring sensor and equipment faults from common measurements. A large data base, covering a wide range of weather conditions, is required for best results. This data base can be created from in-situ routine measurements. Detection and isolation: A robust detection and isolation (formerly referred to as 'identification') method has been developed, which is capable of separating the effect of faults from model inaccuracies and disturbance effects. Sensor and equipment faults: Good detection capabilities have been demonstrated for sensor and equipment failures in both the shoot and root environment. Water stress detection: An excitation method of the shoot environment has been developed, which successfully detected water stress, as soon as the transpiration rate dropped from its normal level. Due to unavailability of suitable monitoring equipment for the root environment, crop faults could not be detected from measurements in the root zone. Dust: The effect of screen clogging by dust has been quantified. Implications Sensor and equipment fault detection and isolation is at a stage where it could be introduced into well equipped and maintained commercial greenhouses on a trial basis. Detection of crop problems requires further work. Dr. Peleg was primarily responsible for developing and implementing the innovative data analysis tools. The cooperation was particularly enhanced by Dr. Peleg's three summer sabbaticals at the ARS, Northem Plains Agricultural Research Laboratory, in Sidney, Montana. Switching from multi-band to hyperspectral remote sensing technology during the last 2 years of the project was advantageous by expanding the scope of detected plant growth attributes e.g. Yield, Leaf Nitrate, Biomass and Sugar Content of sugar beets. However, it disrupted the continuity of the project which was originally planned on a 2 year crop rotation cycle of sugar beets and multiple crops (com and wheat), as commonly planted in eastern Montana. Consequently, at the end of the second year we submitted a continuation BARD proposal which was turned down for funding. This severely hampered our ability to validate our findings as originally planned in a 4-year crop rotation cycle. Thankfully, BARD consented to our request for a one year extension of the project without additional funding. This enabled us to develop most of the methodology for implementing and running the hyperspectral remote sensing system and develop the new analytical tools for solving the non-repeatability problem and analyzing the huge hyperspectral image cube datasets. However, without validation of these tools over a ful14-year crop rotation cycle this project shall remain essentially unfinished. Should the findings of this report prompt the BARD management to encourage us to resubmit our continuation research proposal, we shall be happy to do so.
5
Galili, Naftali, Roger P. Rohrbach, Itzhak Shmulevich, Yoram Fuchs et Giora Zauberman. Non-Destructive Quality Sensing of High-Value Agricultural Commodities Through Response Analysis. United States Department of Agriculture, octobre 1994. http://dx.doi.org/10.32747/1994.7570549.bard.
Texte intégralRésumé :
The objectives of this project were to develop nondestructive methods for detection of internal properties and firmness of fruits and vegetables. One method was based on a soft piezoelectric film transducer developed in the Technion, for analysis of fruit response to low-energy excitation. The second method was a dot-matrix piezoelectric transducer of North Carolina State University, developed for contact-pressure analysis of fruit during impact. Two research teams, one in Israel and the other in North Carolina, coordinated their research effort according to the specific objectives of the project, to develop and apply the two complementary methods for quality control of agricultural commodities. In Israel: An improved firmness testing system was developed and tested with tropical fruits. The new system included an instrumented fruit-bed of three flexible piezoelectric sensors and miniature electromagnetic hammers, which served as fruit support and low-energy excitation device, respectively. Resonant frequencies were detected for determination of firmness index. Two new acoustic parameters were developed for evaluation of fruit firmness and maturity: a dumping-ratio and a centeroid of the frequency response. Experiments were performed with avocado and mango fruits. The internal damping ratio, which may indicate fruit ripeness, increased monotonically with time, while resonant frequencies and firmness indices decreased with time. Fruit samples were tested daily by destructive penetration test. A fairy high correlation was found in tropical fruits between the penetration force and the new acoustic parameters; a lower correlation was found between this parameter and the conventional firmness index. Improved table-top firmness testing units, Firmalon, with data-logging system and on-line data analysis capacity have been built. The new device was used for the full-scale experiments in the next two years, ahead of the original program and BARD timetable. Close cooperation was initiated with local industry for development of both off-line and on-line sorting and quality control of more agricultural commodities. Firmalon units were produced and operated in major packaging houses in Israel, Belgium and Washington State, on mango and avocado, apples, pears, tomatoes, melons and some other fruits, to gain field experience with the new method. The accumulated experimental data from all these activities is still analyzed, to improve firmness sorting criteria and shelf-life predicting curves for the different fruits. The test program in commercial CA storage facilities in Washington State included seven apple varieties: Fuji, Braeburn, Gala, Granny Smith, Jonagold, Red Delicious, Golden Delicious, and D'Anjou pear variety. FI master-curves could be developed for the Braeburn, Gala, Granny Smith and Jonagold apples. These fruits showed a steady ripening process during the test period. Yet, more work should be conducted to reduce scattering of the data and to determine the confidence limits of the method. Nearly constant FI in Red Delicious and the fluctuations of FI in the Fuji apples should be re-examined. Three sets of experiment were performed with Flandria tomatoes. Despite the complex structure of the tomatoes, the acoustic method could be used for firmness evaluation and to follow the ripening evolution with time. Close agreement was achieved between the auction expert evaluation and that of the nondestructive acoustic test, where firmness index of 4.0 and more indicated grade-A tomatoes. More work is performed to refine the sorting algorithm and to develop a general ripening scale for automatic grading of tomatoes for the fresh fruit market. Galia melons were tested in Israel, in simulated export conditions. It was concluded that the Firmalon is capable of detecting the ripening of melons nondestructively, and sorted out the defective fruits from the export shipment. The cooperation with local industry resulted in development of automatic on-line prototype of the acoustic sensor, that may be incorporated with the export quality control system for melons. More interesting is the development of the remote firmness sensing method for sealed CA cool-rooms, where most of the full-year fruit yield in stored for off-season consumption. Hundreds of ripening monitor systems have been installed in major fruit storage facilities, and being evaluated now by the consumers. If successful, the new method may cause a major change in long-term fruit storage technology. More uses of the acoustic test method have been considered, for monitoring fruit maturity and harvest time, testing fruit samples or each individual fruit when entering the storage facilities, packaging house and auction, and in the supermarket. This approach may result in a full line of equipment for nondestructive quality control of fruits and vegetables, from the orchard or the greenhouse, through the entire sorting, grading and storage process, up to the consumer table. The developed technology offers a tool to determine the maturity of the fruits nondestructively by monitoring their acoustic response to mechanical impulse on the tree. A special device was built and preliminary tested in mango fruit. More development is needed to develop a portable, hand operated sensing method for this purpose. In North Carolina: Analysis method based on an Auto-Regressive (AR) model was developed for detecting the first resonance of fruit from their response to mechanical impulse. The algorithm included a routine that detects the first resonant frequency from as many sensors as possible. Experiments on Red Delicious apples were performed and their firmness was determined. The AR method allowed the detection of the first resonance. The method could be fast enough to be utilized in a real time sorting machine. Yet, further study is needed to look for improvement of the search algorithm of the methods. An impact contact-pressure measurement system and Neural Network (NN) identification method were developed to investigate the relationships between surface pressure distributions on selected fruits and their respective internal textural qualities. A piezoelectric dot-matrix pressure transducer was developed for the purpose of acquiring time-sampled pressure profiles during impact. The acquired data was transferred into a personal computer and accurate visualization of animated data were presented. Preliminary test with 10 apples has been performed. Measurement were made by the contact-pressure transducer in two different positions. Complementary measurements were made on the same apples by using the Firmalon and Magness Taylor (MT) testers. Three-layer neural network was designed. 2/3 of the contact-pressure data were used as training input data and corresponding MT data as training target data. The remaining data were used as NN checking data. Six samples randomly chosen from the ten measured samples and their corresponding Firmalon values were used as the NN training and target data, respectively. The remaining four samples' data were input to the NN. The NN results consistent with the Firmness Tester values. So, if more training data would be obtained, the output should be more accurate. In addition, the Firmness Tester values do not consistent with MT firmness tester values. The NN method developed in this study appears to be a useful tool to emulate the MT Firmness test results without destroying the apple samples. To get more accurate estimation of MT firmness a much larger training data set is required. When the larger sensitive area of the pressure sensor being developed in this project becomes available, the entire contact 'shape' will provide additional information and the neural network results would be more accurate. It has been shown that the impact information can be utilized in the determination of internal quality factors of fruit. Until now,
6
Engel, Bernard, Yael Edan, James Simon, Hanoch Pasternak et Shimon Edelman. Neural Networks for Quality Sorting of Agricultural Produce. United States Department of Agriculture, juillet 1996. http://dx.doi.org/10.32747/1996.7613033.bard.
Texte intégralRésumé :
The objectives of this project were to develop procedures and models, based on neural networks, for quality sorting of agricultural produce. Two research teams, one in Purdue University and the other in Israel, coordinated their research efforts on different aspects of each objective utilizing both melons and tomatoes as case studies. At Purdue: An expert system was developed to measure variances in human grading. Data were acquired from eight sensors: vision, two firmness sensors (destructive and nondestructive), chlorophyll from fluorescence, color sensor, electronic sniffer for odor detection, refractometer and a scale (mass). Data were analyzed and provided input for five classification models. Chlorophyll from fluorescence was found to give the best estimation for ripeness stage while the combination of machine vision and firmness from impact performed best for quality sorting. A new algorithm was developed to estimate and minimize training size for supervised classification. A new criteria was established to choose a training set such that a recurrent auto-associative memory neural network is stabilized. Moreover, this method provides for rapid and accurate updating of the classifier over growing seasons, production environments and cultivars. Different classification approaches (parametric and non-parametric) for grading were examined. Statistical methods were found to be as accurate as neural networks in grading. Classification models by voting did not enhance the classification significantly. A hybrid model that incorporated heuristic rules and either a numerical classifier or neural network was found to be superior in classification accuracy with half the required processing of solely the numerical classifier or neural network. In Israel: A multi-sensing approach utilizing non-destructive sensors was developed. Shape, color, stem identification, surface defects and bruises were measured using a color image processing system. Flavor parameters (sugar, acidity, volatiles) and ripeness were measured using a near-infrared system and an electronic sniffer. Mechanical properties were measured using three sensors: drop impact, resonance frequency and cyclic deformation. Classification algorithms for quality sorting of fruit based on multi-sensory data were developed and implemented. The algorithms included a dynamic artificial neural network, a back propagation neural network and multiple linear regression. Results indicated that classification based on multiple sensors may be applied in real-time sorting and can improve overall classification. Advanced image processing algorithms were developed for shape determination, bruise and stem identification and general color and color homogeneity. An unsupervised method was developed to extract necessary vision features. The primary advantage of the algorithms developed is their ability to learn to determine the visual quality of almost any fruit or vegetable with no need for specific modification and no a-priori knowledge. Moreover, since there is no assumption as to the type of blemish to be characterized, the algorithm is capable of distinguishing between stems and bruises. This enables sorting of fruit without knowing the fruits' orientation. A new algorithm for on-line clustering of data was developed. The algorithm's adaptability is designed to overcome some of the difficulties encountered when incrementally clustering sparse data and preserves information even with memory constraints. Large quantities of data (many images) of high dimensionality (due to multiple sensors) and new information arriving incrementally (a function of the temporal dynamics of any natural process) can now be processed. Furhermore, since the learning is done on-line, it can be implemented in real-time. The methodology developed was tested to determine external quality of tomatoes based on visual information. An improved model for color sorting which is stable and does not require recalibration for each season was developed for color determination. Excellent classification results were obtained for both color and firmness classification. Results indicted that maturity classification can be obtained using a drop-impact and a vision sensor in order to predict the storability and marketing of harvested fruits. In conclusion: We have been able to define quantitatively the critical parameters in the quality sorting and grading of both fresh market cantaloupes and tomatoes. We have been able to accomplish this using nondestructive measurements and in a manner consistent with expert human grading and in accordance with market acceptance. This research constructed and used large databases of both commodities, for comparative evaluation and optimization of expert system, statistical and/or neural network models. The models developed in this research were successfully tested, and should be applicable to a wide range of other fruits and vegetables. These findings are valuable for the development of on-line grading and sorting of agricultural produce through the incorporation of multiple measurement inputs that rapidly define quality in an automated manner, and in a manner consistent with the human graders and inspectors.