Pub Date : 2020-09-01DOI: 10.1109/TransAI49837.2020.00012
Kevin Ma
Recently, artificial intelligence has made huge strides in sports analysis. This paper attempts to focus this technology into table tennis with a real-time machine learning system that enables individual ping pong players to have independent training. This system enables table tennis players to maintain the benefits of training with a coach, without the physical presence of one. This, of course, also helps to practice social distancing under present situations. Our system uses a SensorTile development hardware and embedded workbench software to collect real time sensor data, using a variety of MEMS sensors such as accelerometers, gyroscopes, and magnetometers. Therefore, the mounted SensorTile system can detect the motion and orientation of the table tennis racket. We used machine learning (ML) methods to perform real-time table tennis stroke classification producing accurate classification results. Using this proposed machine learning system, players now have an effective training machine that is able to tell them if their strokes are accurate. This also reduces private coaching time in an attempt to limit unnecessary exposure, while still allowing players to receive feedback to improve their game.
{"title":"Artificial Intelligence Aided Training in Ping Pong Sport Education","authors":"Kevin Ma","doi":"10.1109/TransAI49837.2020.00012","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00012","url":null,"abstract":"Recently, artificial intelligence has made huge strides in sports analysis. This paper attempts to focus this technology into table tennis with a real-time machine learning system that enables individual ping pong players to have independent training. This system enables table tennis players to maintain the benefits of training with a coach, without the physical presence of one. This, of course, also helps to practice social distancing under present situations. Our system uses a SensorTile development hardware and embedded workbench software to collect real time sensor data, using a variety of MEMS sensors such as accelerometers, gyroscopes, and magnetometers. Therefore, the mounted SensorTile system can detect the motion and orientation of the table tennis racket. We used machine learning (ML) methods to perform real-time table tennis stroke classification producing accurate classification results. Using this proposed machine learning system, players now have an effective training machine that is able to tell them if their strokes are accurate. This also reduces private coaching time in an attempt to limit unnecessary exposure, while still allowing players to receive feedback to improve their game.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123857552","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}
Pub Date : 2020-09-01DOI: 10.1109/TransAI49837.2020.00014
Mrudula Murali, Katerina Potika, C. Pollett
A community in a network is a set of nodes that are densely and closely connected within the set, yet sparsely connected to nodes outside of it. Detecting communities in large networks helps solve many real-world problems. However, detecting such communities in a complex network by focusing on the whole network is costly. Instead, one can focus on finding overlapping communities starting from one or more seed nodes of interest. Moreover, on the online setting the network is given as a stream of higher order structures, i.e., triangles of nodes to be clustered into communities.In this paper, we propose an on online local graph community detection algorithm that uses motifs, such as triangles of nodes. We provide experimental results and compare it to another algorithm named COEUS. We use two public datasets, one of Amazon data and the other of DBLP data. Furthermore, we create and experiment on a new dataset that consists of web pages and their links by using the Internet Archive. This latter dataset provides insights to better understand how working with motifs is different than working with edges.
{"title":"Online local communities with motifs","authors":"Mrudula Murali, Katerina Potika, C. Pollett","doi":"10.1109/TransAI49837.2020.00014","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00014","url":null,"abstract":"A community in a network is a set of nodes that are densely and closely connected within the set, yet sparsely connected to nodes outside of it. Detecting communities in large networks helps solve many real-world problems. However, detecting such communities in a complex network by focusing on the whole network is costly. Instead, one can focus on finding overlapping communities starting from one or more seed nodes of interest. Moreover, on the online setting the network is given as a stream of higher order structures, i.e., triangles of nodes to be clustered into communities.In this paper, we propose an on online local graph community detection algorithm that uses motifs, such as triangles of nodes. We provide experimental results and compare it to another algorithm named COEUS. We use two public datasets, one of Amazon data and the other of DBLP data. Furthermore, we create and experiment on a new dataset that consists of web pages and their links by using the Internet Archive. This latter dataset provides insights to better understand how working with motifs is different than working with edges.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126056621","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}
Pub Date : 2020-09-01DOI: 10.1109/TransAI49837.2020.00007
Chenxi Zhang, Ning Zhang, Dechun Wang, Yu Cao, Benyuan Liu
Gastrointestinal cancer is a common and deadly disease that affects many people in the world. In 2019, Gastrointestinal cancer was the most common cancer and the second leading cause of death in the US. Detecting gastrointestinal cancer during the early stage is the most effective way to improve the survival rate. One of the commonly used clinical procedures for early detection of gastrointestinal cancer is endoscopy. The main challenge of a high-quality endoscopy operation is the presence of various forms of artifacts during the operation, e.g., pixel saturation, motion blur, defocus, specular reflections, bubbles, fluid, debris. These artifacts not only increase the difficulty in examining the underlying tissues during diagnosis but also affect the post-analysis methods required for follow-ups (e.g., video mosaicking for follow-ups and archival purposes and video-frame retrieval for reporting). Also, the presence of these artifacts often interferes with the computer-aided diagnosis of various lesions in endoscopy. The Convolutional Neural Network (CNN) based object detection methods have proved to be an effective approach for nature image object detection and colonoscopy applications (e.g., polyp detection). However, fewer efforts have been devoted to endoscopic artifact detection due to the lack of training data. In this paper, we use data from the EAD2019 challenge and investigate the performance of two improved CNN-based methods for seven-class endoscopic artifact detection (EAD). Experiment results show that our proposed objection detectors based on SSD and Faster-RCNN significantly outperform the baseline.
{"title":"Artifact Detection in Endoscopic Video with Deep Convolutional Neural Networks","authors":"Chenxi Zhang, Ning Zhang, Dechun Wang, Yu Cao, Benyuan Liu","doi":"10.1109/TransAI49837.2020.00007","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00007","url":null,"abstract":"Gastrointestinal cancer is a common and deadly disease that affects many people in the world. In 2019, Gastrointestinal cancer was the most common cancer and the second leading cause of death in the US. Detecting gastrointestinal cancer during the early stage is the most effective way to improve the survival rate. One of the commonly used clinical procedures for early detection of gastrointestinal cancer is endoscopy. The main challenge of a high-quality endoscopy operation is the presence of various forms of artifacts during the operation, e.g., pixel saturation, motion blur, defocus, specular reflections, bubbles, fluid, debris. These artifacts not only increase the difficulty in examining the underlying tissues during diagnosis but also affect the post-analysis methods required for follow-ups (e.g., video mosaicking for follow-ups and archival purposes and video-frame retrieval for reporting). Also, the presence of these artifacts often interferes with the computer-aided diagnosis of various lesions in endoscopy. The Convolutional Neural Network (CNN) based object detection methods have proved to be an effective approach for nature image object detection and colonoscopy applications (e.g., polyp detection). However, fewer efforts have been devoted to endoscopic artifact detection due to the lack of training data. In this paper, we use data from the EAD2019 challenge and investigate the performance of two improved CNN-based methods for seven-class endoscopic artifact detection (EAD). Experiment results show that our proposed objection detectors based on SSD and Faster-RCNN significantly outperform the baseline.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132087247","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}
Pub Date : 2020-09-01DOI: 10.1109/TransAI49837.2020.00025
Felicity Gerry, Joseph R. Barr, Peter Shaw
This poster uses the mathematics of networks in the novel context of corporate reporting of slavery in supply chains as a method to meet corporate obligations to respect human rights. For those corporates considering risks such as liability for slavery in supply chains, using graph theory, which is capable of sampling affinity in data-bases, can ‘value add’ due diligence by scoring identity and veracity.
{"title":"Graph Theory–The Case for Investigating Corruption and Modern Slavery through Suspicious Employment Data","authors":"Felicity Gerry, Joseph R. Barr, Peter Shaw","doi":"10.1109/TransAI49837.2020.00025","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00025","url":null,"abstract":"This poster uses the mathematics of networks in the novel context of corporate reporting of slavery in supply chains as a method to meet corporate obligations to respect human rights. For those corporates considering risks such as liability for slavery in supply chains, using graph theory, which is capable of sampling affinity in data-bases, can ‘value add’ due diligence by scoring identity and veracity.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129056719","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}
Pub Date : 2020-09-01DOI: 10.1109/TransAI49837.2020.00030
Bruce X. B. Yu, Yan Liu, Keith C. C. Chan
Human action abnormality detection has been attempted by various sensors for application domains like rehabilitation, healthcare, and assisted living. Since the release of motion sensors that ease the human body skeleton retrieval, skeleton-based human action recognition has recently been an active topic in the area of artificial intelligence. Unlike human action recognition, human action abnormality detection is an emerging field that aims to detect the incorrect action from the same action class. Graph convolutional network has been widely adopted for human action recognition. However, to the best of our knowledge, whether it could be effective for the task of human action abnormality detection has not been attempted. To advance prior work in the emerging field of human action abnormality detection, we propose a novel method that uses graph convolutional network to detect abnormal actions in skeleton data. To validate the effectiveness of our proposed method, we conduct extensive experiments on a public dataset called UI-PRMD. Based on the experimental results, our proposed method achieved superior action abnormality detection performance comparing with existing deep learning methods.
{"title":"Skeleton-Based Detection of Abnormalities in Human Actions Using Graph Convolutional Networks","authors":"Bruce X. B. Yu, Yan Liu, Keith C. C. Chan","doi":"10.1109/TransAI49837.2020.00030","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00030","url":null,"abstract":"Human action abnormality detection has been attempted by various sensors for application domains like rehabilitation, healthcare, and assisted living. Since the release of motion sensors that ease the human body skeleton retrieval, skeleton-based human action recognition has recently been an active topic in the area of artificial intelligence. Unlike human action recognition, human action abnormality detection is an emerging field that aims to detect the incorrect action from the same action class. Graph convolutional network has been widely adopted for human action recognition. However, to the best of our knowledge, whether it could be effective for the task of human action abnormality detection has not been attempted. To advance prior work in the emerging field of human action abnormality detection, we propose a novel method that uses graph convolutional network to detect abnormal actions in skeleton data. To validate the effectiveness of our proposed method, we conduct extensive experiments on a public dataset called UI-PRMD. Based on the experimental results, our proposed method achieved superior action abnormality detection performance comparing with existing deep learning methods.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122222911","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}
Pub Date : 2020-09-01DOI: 10.1109/TransAI49837.2020.00026
Vinesh Ravuri, Projna Paromita, Karel Mundnich, Amrutha Nadarajan, Brandon M. Booth, Shrikanth S. Narayanan, Theodora Chaspari
Healthcare workers often experience stress and burnout due to the demanding job responsibilities and long work hours. Ambulatory monitoring devices, such as wearable and environmental sensors, combined with machine learning algorithms can afford us a better understanding of the naturalistic onset and evolution of stress and emotional reactivity in real-life with valuable implications in behavioral interventions. However, the typically large degree of inter-subject variability, due to individual differences in responses and behaviors, makes it difficult for machine learning models to robustly learn behavioral signal patterns and adequately generalize to unseen individuals. In this study, we design group-specific models of well-being (i.e., stress, sleep, positive affect, negative affect) and contextual outcomes (i.e., type of activity) based on real-life multimodal longitudinal data collected in situ from healthcare workers in a hospital environment. Group-specific models are constructed by learning an initial model based on all individuals and subsequently refining the model for a specific group of participants. Participants are originally grouped based on the feature space constructed by the multimodal data, while the original grouping is iteratively refined using the learned multimodal representations of the group-specific models. The results from this study indicate that in the majority of cases the proposed group-specific models, learned through iterative participant clustering, outperform the baseline systems, which involve general models learned based on all participants, as well as group-specific models without iterative participant clustering. This study provides promising results for predicting psychological and behavioral factors that affect the well-being of healthcare workers and lays the foundation toward ambulatory real-life assessment and interventions.
{"title":"Group-specific models of healthcare workers’ well-being using iterative participant clustering","authors":"Vinesh Ravuri, Projna Paromita, Karel Mundnich, Amrutha Nadarajan, Brandon M. Booth, Shrikanth S. Narayanan, Theodora Chaspari","doi":"10.1109/TransAI49837.2020.00026","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00026","url":null,"abstract":"Healthcare workers often experience stress and burnout due to the demanding job responsibilities and long work hours. Ambulatory monitoring devices, such as wearable and environmental sensors, combined with machine learning algorithms can afford us a better understanding of the naturalistic onset and evolution of stress and emotional reactivity in real-life with valuable implications in behavioral interventions. However, the typically large degree of inter-subject variability, due to individual differences in responses and behaviors, makes it difficult for machine learning models to robustly learn behavioral signal patterns and adequately generalize to unseen individuals. In this study, we design group-specific models of well-being (i.e., stress, sleep, positive affect, negative affect) and contextual outcomes (i.e., type of activity) based on real-life multimodal longitudinal data collected in situ from healthcare workers in a hospital environment. Group-specific models are constructed by learning an initial model based on all individuals and subsequently refining the model for a specific group of participants. Participants are originally grouped based on the feature space constructed by the multimodal data, while the original grouping is iteratively refined using the learned multimodal representations of the group-specific models. The results from this study indicate that in the majority of cases the proposed group-specific models, learned through iterative participant clustering, outperform the baseline systems, which involve general models learned based on all participants, as well as group-specific models without iterative participant clustering. This study provides promising results for predicting psychological and behavioral factors that affect the well-being of healthcare workers and lays the foundation toward ambulatory real-life assessment and interventions.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":" 28","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120832209","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}
Pub Date : 2020-04-18DOI: 10.1109/TransAI49837.2020.00009
Saeed Moazami, P. Doerschuk
Although recent model-free reinforcement learning algorithms have been shown to be capable of mastering complicated decision-making tasks, the sample complexity of these methods has remained a hurdle to utilizing them in many real-world applications. In this regard, model-based reinforcement learning proposes some remedies. Yet, inherently, model-based methods are more computationally expensive and susceptible to sub-optimality. One reason is that model-generated data are always less accurate than real data, and this often leads to inaccurate transition and reward function models. With the aim to mitigate this problem, this work presents the notion of survival by discussing cases in which the agent’s goal is to survive and its analogy to maximizing the expected rewards. To that end, a substitute model for the reward function approximator is introduced that learns to avoid terminal states rather than to maximize accumulated rewards from safe states. Focusing on terminal states, as a small fraction of state-space, reduces the training effort drastically. Next, a model-based reinforcement learning method is proposed (Survive) to train an agent to avoid dangerous states through a safety map model built upon temporal credit assignment in the vicinity of terminal states. Finally, the performance of the presented algorithm is investigated, along with a comparison between the proposed and current methods.
{"title":"Modeling Survival in model-based Reinforcement Learning","authors":"Saeed Moazami, P. Doerschuk","doi":"10.1109/TransAI49837.2020.00009","DOIUrl":"https://doi.org/10.1109/TransAI49837.2020.00009","url":null,"abstract":"Although recent model-free reinforcement learning algorithms have been shown to be capable of mastering complicated decision-making tasks, the sample complexity of these methods has remained a hurdle to utilizing them in many real-world applications. In this regard, model-based reinforcement learning proposes some remedies. Yet, inherently, model-based methods are more computationally expensive and susceptible to sub-optimality. One reason is that model-generated data are always less accurate than real data, and this often leads to inaccurate transition and reward function models. With the aim to mitigate this problem, this work presents the notion of survival by discussing cases in which the agent’s goal is to survive and its analogy to maximizing the expected rewards. To that end, a substitute model for the reward function approximator is introduced that learns to avoid terminal states rather than to maximize accumulated rewards from safe states. Focusing on terminal states, as a small fraction of state-space, reduces the training effort drastically. Next, a model-based reinforcement learning method is proposed (Survive) to train an agent to avoid dangerous states through a safety map model built upon temporal credit assignment in the vicinity of terminal states. Finally, the performance of the presented algorithm is investigated, along with a comparison between the proposed and current methods.","PeriodicalId":151527,"journal":{"name":"2020 Second International Conference on Transdisciplinary AI (TransAI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123729642","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：