Pub Date : 2022-05-24DOI: 10.48550/arXiv.2205.11830
Iason Katsamenis, Eleni Eirini Karolou, Agapi Davradou, Eftychios E. Protopapadakis, A. Doulamis, N. Doulamis, D. Kalogeras
Substantial progress has been made in the field of object detection in road scenes. However, it is mainly focused on vehicles and pedestrians. To this end, we investigate traffic cone detection, an object category crucial for road effects and maintenance. In this work, the YOLOv5 algorithm is employed, in order to find a solution for the efficient and fast detection of traffic cones. The YOLOv5 can achieve a high detection accuracy with the score of IoU up to 91.31%. The proposed method is been applied to an RGB roadwork image dataset, collected from various sources.
{"title":"TraCon: A novel dataset for real-time traffic cones detection using deep learning","authors":"Iason Katsamenis, Eleni Eirini Karolou, Agapi Davradou, Eftychios E. Protopapadakis, A. Doulamis, N. Doulamis, D. Kalogeras","doi":"10.48550/arXiv.2205.11830","DOIUrl":"https://doi.org/10.48550/arXiv.2205.11830","url":null,"abstract":"Substantial progress has been made in the field of object detection in road scenes. However, it is mainly focused on vehicles and pedestrians. To this end, we investigate traffic cone detection, an object category crucial for road effects and maintenance. In this work, the YOLOv5 algorithm is employed, in order to find a solution for the efficient and fast detection of traffic cones. The YOLOv5 can achieve a high detection accuracy with the score of IoU up to 91.31%. The proposed method is been applied to an RGB roadwork image dataset, collected from various sources.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116300210","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Andrianthi Kapetanaki, Akrivi Krouska, C. Troussas, C. Sgouropoulou
Augmented Reality (AR) is an emerging technology thriving in recent years. The implementation of AR in education offers great opportunities to enhance educational environments achieving better learning outcomes. As students with learning disabilities struggle with reading comprehension, an AR learning environment provides them support to better understand texts they actually read. Even though few studies have tried to explore the impact of AR technology to reading comprehension for students with learning disabilities in Secondary Education, there is a lack of research grounded in the incorporation of learning theories and personalization technologies. The goal of this paper is to present an AR educational environment capable of supporting meaningful learning outcomes by taking into consideration each student special educational needs and learning style. The novelty of this study lies in the student-centered and personalized design, which leads to improved understanding, student interaction and self-learning.
{"title":"A Novel Framework Incorporating Augmented Reality and Pedagogy for Improving Reading Comprehension in Special Education","authors":"Andrianthi Kapetanaki, Akrivi Krouska, C. Troussas, C. Sgouropoulou","doi":"10.3233/faia210081","DOIUrl":"https://doi.org/10.3233/faia210081","url":null,"abstract":"Augmented Reality (AR) is an emerging technology thriving in recent years. The implementation of AR in education offers great opportunities to enhance educational environments achieving better learning outcomes. As students with learning disabilities struggle with reading comprehension, an AR learning environment provides them support to better understand texts they actually read. Even though few studies have tried to explore the impact of AR technology to reading comprehension for students with learning disabilities in Secondary Education, there is a lack of research grounded in the incorporation of learning theories and personalization technologies. The goal of this paper is to present an AR educational environment capable of supporting meaningful learning outcomes by taking into consideration each student special educational needs and learning style. The novelty of this study lies in the student-centered and personalized design, which leads to improved understanding, student interaction and self-learning.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117004411","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Subjective cognitive decline is an early state of Alzheimer’s Disease which affects almost 10 million people every year. It results from negative emotions such as frustration which are more present than healthy adults. For this reason, our work focuses on relaxing subjective cognitive decline patients using virtual reality environments to improve their memory performance. We proposed in our previous work a neurofeedback approach which adapts the virtual environment to each patient according to their emotions using a Neural Agent. We found that the Neural Agent can adapt the environment to each participant but have limitations. This work is a continuation of our approach in which we propose a Limbic Agent able to monitor the interactions between the Neural Agent and patients’ emotional reactions, learn from these interactions, and modify the Neural Agent in order to enhance the adaptation to each patient with an Intelligent Cognitive Control System. Our goal is to create a system able to support the Limbic System which is the main area in charge of controlling emotions and creating memory in the human brain. We used data collected form our previous work to train the Limbic Agent and results showed that the agent is capable of modifying the weight of existing rules, generating new intervention rules, and predicting if they will work or not.
{"title":"Toward Personalizing Alzheimer's Disease Therapy Using an Intelligent Cognitive Control System","authors":"H. Abdessalem, C. Frasson","doi":"10.3233/faia210077","DOIUrl":"https://doi.org/10.3233/faia210077","url":null,"abstract":"Subjective cognitive decline is an early state of Alzheimer’s Disease which affects almost 10 million people every year. It results from negative emotions such as frustration which are more present than healthy adults. For this reason, our work focuses on relaxing subjective cognitive decline patients using virtual reality environments to improve their memory performance. We proposed in our previous work a neurofeedback approach which adapts the virtual environment to each patient according to their emotions using a Neural Agent. We found that the Neural Agent can adapt the environment to each participant but have limitations. This work is a continuation of our approach in which we propose a Limbic Agent able to monitor the interactions between the Neural Agent and patients’ emotional reactions, learn from these interactions, and modify the Neural Agent in order to enhance the adaptation to each patient with an Intelligent Cognitive Control System. Our goal is to create a system able to support the Limbic System which is the main area in charge of controlling emotions and creating memory in the human brain. We used data collected form our previous work to train the Limbic Agent and results showed that the agent is capable of modifying the weight of existing rules, generating new intervention rules, and predicting if they will work or not.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121973503","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Z. Kanetaki, C. Stergiou, C. Troussas, C. Sgouropoulou
The COVID-19 pandemic struck humanity in February 2020. Closures of educational institutions, worldwide, resulted to the creation of emergency remote teaching environments as a substitute to face to face learning. The disruption caused in the academic community has stimulated innovative learning methods within all levels of the educational sector. New parameters affecting knowledge transmission are getting involved while students follow courses apart on a common virtual learning environment. This research is based on a first-semester Mechanical Engineering CAD module in tertiary education. A learning strategy has been applied by reforming the traditional face-to-face leaning mode to a fully remote learning environment. The methods applied have been tested using statistical analysis and have shown to contribute significantly in students’ spatial perception in 2-Dimentional Drawings. The outcomes of this research reveal a novel teaching strategy that improved students’ academic achievements in CAD during the lockdown. Specific aspects can be considered sustainable on their return back to normality.
{"title":"Development of an Innovative Learning Methodology Aiming to Optimise Learners' Spatial Conception in an Online Mechanical CAD Module During COVID-19 Pandemic","authors":"Z. Kanetaki, C. Stergiou, C. Troussas, C. Sgouropoulou","doi":"10.3233/faia210072","DOIUrl":"https://doi.org/10.3233/faia210072","url":null,"abstract":"The COVID-19 pandemic struck humanity in February 2020. Closures of educational institutions, worldwide, resulted to the creation of emergency remote teaching environments as a substitute to face to face learning. The disruption caused in the academic community has stimulated innovative learning methods within all levels of the educational sector. New parameters affecting knowledge transmission are getting involved while students follow courses apart on a common virtual learning environment. This research is based on a first-semester Mechanical Engineering CAD module in tertiary education. A learning strategy has been applied by reforming the traditional face-to-face leaning mode to a fully remote learning environment. The methods applied have been tested using statistical analysis and have shown to contribute significantly in students’ spatial perception in 2-Dimentional Drawings. The outcomes of this research reveal a novel teaching strategy that improved students’ academic achievements in CAD during the lockdown. Specific aspects can be considered sustainable on their return back to normality.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132527181","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The work presents a new approach to the study of problems associated with the initial definition of the concept of “intelligence”. In work, intelligence is a property that ensures, at some level, the successful interaction of a subject or system with their environment, which is specified in the form of a subject area. For this, the subject area is modeled (smart modeling). Modeling is the process of representing a domain in which the subject must solve the corresponding problems. The domain model is presented as a body of knowledge covering the constituent elements of this domain. Knowledge is a means of describing this model and is determined using a language that includes the logic of describing the corresponding entities of the subject area. Language properties are used to model some abstract formal structure in the form of a mathematical representation. The modeling process consists in a sequential transition from the description of the subject area to an interconnected structure of knowledge, associated with the constituent elements of this area, which are highlighted in the modeling process.
{"title":"Intelligence and Intelligent Simulation","authors":"V. Meytus","doi":"10.3233/faia210083","DOIUrl":"https://doi.org/10.3233/faia210083","url":null,"abstract":"The work presents a new approach to the study of problems associated with the initial definition of the concept of “intelligence”. In work, intelligence is a property that ensures, at some level, the successful interaction of a subject or system with their environment, which is specified in the form of a subject area. For this, the subject area is modeled (smart modeling). Modeling is the process of representing a domain in which the subject must solve the corresponding problems. The domain model is presented as a body of knowledge covering the constituent elements of this domain. Knowledge is a means of describing this model and is determined using a language that includes the logic of describing the corresponding entities of the subject area. Language properties are used to model some abstract formal structure in the form of a mathematical representation. The modeling process consists in a sequential transition from the description of the subject area to an interconnected structure of knowledge, associated with the constituent elements of this area, which are highlighted in the modeling process.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116061488","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Charalampos Zafeiropoulos, M. Bimpas, Eftychios E. Protopapadakis, Emmanuel Sardis, N. Doulamis, A. Doulamis, C. Maksimovic, Stanislava Boskovic, Ranko Bozovic, Maja Lalic
This paper introduces euPOLIS; an EU funded project, which emphasize on the appropriate development of urban ecosystems in a way that enhance Public Health (PH) and Well-Being (WB) without significant Life-Cycle costs. Such an approach has has the potential to regenerate urban ecosystems addressing multiple challenges, such as low environmental quality, fragmentation and low biodiversity in public spaces, water-stressed resources, undervalued use of space in deprived areas resulting in an improved urban livability. The proposed methodology is expected to improve people’s quality of life, providing them with pleasant socializing open areas that stimulate social exchange while monitoring the impact of all those interventions to PH and WB of citizens. The euPOLIS suggested solutions will be demonstrated in 4 European cities: Belgrade, Lodz, Piraeus and Gladsaxe.
{"title":"An Introduction to the euPOLIS Project","authors":"Charalampos Zafeiropoulos, M. Bimpas, Eftychios E. Protopapadakis, Emmanuel Sardis, N. Doulamis, A. Doulamis, C. Maksimovic, Stanislava Boskovic, Ranko Bozovic, Maja Lalic","doi":"10.3233/faia210094","DOIUrl":"https://doi.org/10.3233/faia210094","url":null,"abstract":"This paper introduces euPOLIS; an EU funded project, which emphasize on the appropriate development of urban ecosystems in a way that enhance Public Health (PH) and Well-Being (WB) without significant Life-Cycle costs. Such an approach has has the potential to regenerate urban ecosystems addressing multiple challenges, such as low environmental quality, fragmentation and low biodiversity in public spaces, water-stressed resources, undervalued use of space in deprived areas resulting in an improved urban livability. The proposed methodology is expected to improve people’s quality of life, providing them with pleasant socializing open areas that stimulate social exchange while monitoring the impact of all those interventions to PH and WB of citizens. The euPOLIS suggested solutions will be demonstrated in 4 European cities: Belgrade, Lodz, Piraeus and Gladsaxe.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116683561","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Antonia Kaouni, Georgia Theodoropoulou, Alexandros Bousdekis, A. Voulodimos, G. Miaoulis
The increasing amounts of data have affected conceptual modeling as a research field. In this context, process mining involves a set of techniques aimed at extracting a process schema from an event log generated during process execution. While automatic algorithms for process mining and analysis are needed to filter out irrelevant data and to produce preliminary results, visual inspection, domain knowledge, human judgment and creativity are needed for proper interpretation of the results. Moreover, a process discovery on an event log usually results in complicated process models not easily comprehensible by the business user. To this end, visual analytics has the potential to enhance process mining towards the direction of explainability, interpretability and trustworthiness in order to better support human decisions. In this paper we propose an approach for identifying bottlenecks in business processes by analyzing event logs and visualizing the results. In this way, we exploit visual analytics in the process mining context in order to provide explainable and interpretable analytics results for business processes without exposing to the user complex process models that are not easily comprehensible. The proposed approach was applied to a manufacturing business process and the results show that visual analytics in the context of process mining is capable of identifying bottlenecks and other performance-related issues and exposing them to the business user in an intuitive and non-intrusive way.
{"title":"Visual Analytics in Process Mining for Supporting Business Process Improvement","authors":"Antonia Kaouni, Georgia Theodoropoulou, Alexandros Bousdekis, A. Voulodimos, G. Miaoulis","doi":"10.3233/faia210089","DOIUrl":"https://doi.org/10.3233/faia210089","url":null,"abstract":"The increasing amounts of data have affected conceptual modeling as a research field. In this context, process mining involves a set of techniques aimed at extracting a process schema from an event log generated during process execution. While automatic algorithms for process mining and analysis are needed to filter out irrelevant data and to produce preliminary results, visual inspection, domain knowledge, human judgment and creativity are needed for proper interpretation of the results. Moreover, a process discovery on an event log usually results in complicated process models not easily comprehensible by the business user. To this end, visual analytics has the potential to enhance process mining towards the direction of explainability, interpretability and trustworthiness in order to better support human decisions. In this paper we propose an approach for identifying bottlenecks in business processes by analyzing event logs and visualizing the results. In this way, we exploit visual analytics in the process mining context in order to provide explainable and interpretable analytics results for business processes without exposing to the user complex process models that are not easily comprehensible. The proposed approach was applied to a manufacturing business process and the results show that visual analytics in the context of process mining is capable of identifying bottlenecks and other performance-related issues and exposing them to the business user in an intuitive and non-intrusive way.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"121 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114162369","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nowadays human activities are incoming at a digitalization stage. The introduction of information technology along with new forms of communication, influence a variety of forms of human action and focus mainly on the integration and the convergence of the digital and physical worlds. The use of more intelligent – electronic solutions, improves the lives of people around the world, according to studies carried out on the ingress of new smart technologies. Artificial and Ambient intelligence nowadays getting more and more attention about the development of smart, digital environments. The Smart Cities designed for All must aim to arrange the disparity in cities through smart technology, making cities both smart and accessible to a range of users regardless of their abilities or disabilities. The birth of “Artificial Intelligence” (AI) has facilitated the complex computations for reality simulation the new communication era of wireless 5G, all combined have given the hope for a new and better future, to reverse disability to empower the humans with more capabilities, to be faster than they can ever be, stronger than they can ever dream. This paper provides an overview of Ambient Intelligence and smart environments, as well as how technological advancements will benefit everyday usage by devices in common spaces such as homes or offices, and how they will interact and serve as a part of an intelligent ecosystem by bringing together resources such as networks, sensors, human-computer interfaces, pervasive computing, and so on.
{"title":"Ambient Intelligence and Smart Environments: A Preliminary Overview","authors":"Maria Poli","doi":"10.3233/faia210074","DOIUrl":"https://doi.org/10.3233/faia210074","url":null,"abstract":"Nowadays human activities are incoming at a digitalization stage. The introduction of information technology along with new forms of communication, influence a variety of forms of human action and focus mainly on the integration and the convergence of the digital and physical worlds. The use of more intelligent – electronic solutions, improves the lives of people around the world, according to studies carried out on the ingress of new smart technologies. Artificial and Ambient intelligence nowadays getting more and more attention about the development of smart, digital environments. The Smart Cities designed for All must aim to arrange the disparity in cities through smart technology, making cities both smart and accessible to a range of users regardless of their abilities or disabilities. The birth of “Artificial Intelligence” (AI) has facilitated the complex computations for reality simulation the new communication era of wireless 5G, all combined have given the hope for a new and better future, to reverse disability to empower the humans with more capabilities, to be faster than they can ever be, stronger than they can ever dream. This paper provides an overview of Ambient Intelligence and smart environments, as well as how technological advancements will benefit everyday usage by devices in common spaces such as homes or offices, and how they will interact and serve as a part of an intelligent ecosystem by bringing together resources such as networks, sensors, human-computer interfaces, pervasive computing, and so on.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122429028","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Efforts toward COVID-19 proximity tracking in closed environments focus on efficient proximity identification by combining it with indoor localization theory for location activity monitoring and proximity detection. But these are met with concerns based on existing considerations of the localization theory like costly infrastructure, multi-story support, and over-reliance on sensor networks. Semantic location identities (SLI), or location data stored with additional meaningful context, has become a feasible localizing factor especially in locations that have multiple spaces with different usage from each other. There is also a novel method of classification framework, called hierarchical classification, that leverages the hierarchical structure of the labels to reduce model complexity. The research aims to provide a solution to proximity analysis and location activity monitoring considering guidelines released in a Philippine context that addresses concerns of indoor localization and handling of geospatial data by implementing a hybrid hierarchical indoor semantic location identity classification that focuses on observable events within context-unique locations.
{"title":"Implementing Hierarchical Indoor Semantic Location Identity Classification: A Case Study for COVID-19 Proximity Tracking in the Philippines","authors":"I.K.P. Paderes, L. L. Figueroa, R. Feria","doi":"10.3233/faia210087","DOIUrl":"https://doi.org/10.3233/faia210087","url":null,"abstract":"Efforts toward COVID-19 proximity tracking in closed environments focus on efficient proximity identification by combining it with indoor localization theory for location activity monitoring and proximity detection. But these are met with concerns based on existing considerations of the localization theory like costly infrastructure, multi-story support, and over-reliance on sensor networks. Semantic location identities (SLI), or location data stored with additional meaningful context, has become a feasible localizing factor especially in locations that have multiple spaces with different usage from each other. There is also a novel method of classification framework, called hierarchical classification, that leverages the hierarchical structure of the labels to reduce model complexity. The research aims to provide a solution to proximity analysis and location activity monitoring considering guidelines released in a Philippine context that addresses concerns of indoor localization and handling of geospatial data by implementing a hybrid hierarchical indoor semantic location identity classification that focuses on observable events within context-unique locations.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127759923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Lazaros Toumanidis, P. Kasnesis, Christos Chatzigeorgiou, Michail Feidakis, C. Patrikakis
A widespread practice in machine learning solutions is the continuous use of human intelligence to increase their quality and efficiency. A common problem in such solutions is the requirement of a large amount of labeled data. In this paper, we present a practical implementation of the human-in-the-loop computing practice, which includes the combination of active and transfer learning for sophisticated data sampling and weight initialization respectively, and a cross-platform mobile application for crowdsourcing data annotation tasks. We study the use of the proposed framework to a post-event building reconnaissance scenario, where we utilized the implementation of an existing pre-trained computer vision model, an image binary classification solution built on top of it, and max entropy and random sampling as uncertainty sampling methods for the active learning step. Multiple annotations with majority voting as quality assurance are required for new human-annotated images to be added on the train set and retrain the model. We provide the results and discuss our next steps.
{"title":"ActiveCrowds: A Human-in-the-Loop Machine Learning Framework","authors":"Lazaros Toumanidis, P. Kasnesis, Christos Chatzigeorgiou, Michail Feidakis, C. Patrikakis","doi":"10.3233/faia210090","DOIUrl":"https://doi.org/10.3233/faia210090","url":null,"abstract":"A widespread practice in machine learning solutions is the continuous use of human intelligence to increase their quality and efficiency. A common problem in such solutions is the requirement of a large amount of labeled data. In this paper, we present a practical implementation of the human-in-the-loop computing practice, which includes the combination of active and transfer learning for sophisticated data sampling and weight initialization respectively, and a cross-platform mobile application for crowdsourcing data annotation tasks. We study the use of the proposed framework to a post-event building reconnaissance scenario, where we utilized the implementation of an existing pre-trained computer vision model, an image binary classification solution built on top of it, and max entropy and random sampling as uncertainty sampling methods for the active learning step. Multiple annotations with majority voting as quality assurance are required for new human-annotated images to be added on the train set and retrain the model. We provide the results and discuss our next steps.","PeriodicalId":234167,"journal":{"name":"International Conference on Novelties in Intelligent Digital Systems","volume":"9 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129299056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：