Pub Date : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378628
Gábor Kertész
Metric embedding learning is a special form of supervised learning: instead of regression or classification a similarity value is predicted based on embedded vector distance. To implement such a behavior, first the Siamese architecture was introduced, where training is based on two input samples, and the transformation model seeks to minimize distance between same-category samples, and increase distance between different samples. To deal with the problem of overtraining, the triplet loss was introduced in 2015, considering three input samples at a training step. Triplet networks also highlighted a novel problem: sample selection is important to eliminate those training triplets, where the measured distance based similarity results in zero loss. To deal with this phenomena, triplet mining techniques are analyzed, while other researchers discussed the possibility of different triplet-based loss functions. In this paper, the so-called lossless triplet loss function is compared with the original triplet loss method, while applying different negative sampling methods.
{"title":"Different triplet sampling techniques for lossless triplet loss on metric similarity learning","authors":"Gábor Kertész","doi":"10.1109/SAMI50585.2021.9378628","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378628","url":null,"abstract":"Metric embedding learning is a special form of supervised learning: instead of regression or classification a similarity value is predicted based on embedded vector distance. To implement such a behavior, first the Siamese architecture was introduced, where training is based on two input samples, and the transformation model seeks to minimize distance between same-category samples, and increase distance between different samples. To deal with the problem of overtraining, the triplet loss was introduced in 2015, considering three input samples at a training step. Triplet networks also highlighted a novel problem: sample selection is important to eliminate those training triplets, where the measured distance based similarity results in zero loss. To deal with this phenomena, triplet mining techniques are analyzed, while other researchers discussed the possibility of different triplet-based loss functions. In this paper, the so-called lossless triplet loss function is compared with the original triplet loss method, while applying different negative sampling methods.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129431688","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378640
Qin Sun, Hongxu Li, Yingchao Zhang, Yuxian Xie, Chengyu Liu
Preliminary progress has been made in the assessment of unmanned aerial vehicle (UAV) swarm resilience based on complex networks. However, the evaluation results mostly use the initial performance state as the evaluation baseline, which is unreasonable. When UAV swarm performs a mission, as long as the network performance during the life cycle is sufficient to meet the mission requirements, it can be considered that UAV swarm has the resilience required to complete the mission. Therefore, a baseline assessment method of UAV swarm resilience based on complex networks is proposed in this paper. First, the baseline assessment method of UAV swarm resilience based on complex networks is characterized and investigated. Second, the effectiveness of the baseline assessment method is verified by simulation. The result shows that the baseline evaluation can effectively relax the evaluation result in a mission-oriented manner, and no longer use the initial state as the standard performance to measure the completion of the mission of UAV swarm. When UAV swarm performs a mission, it only needs to maintain or restore the resilience needed to complete the mission.
{"title":"A Baseline Assessment Method of UAV Swarm Resilience Based on Complex Networks*","authors":"Qin Sun, Hongxu Li, Yingchao Zhang, Yuxian Xie, Chengyu Liu","doi":"10.1109/SAMI50585.2021.9378640","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378640","url":null,"abstract":"Preliminary progress has been made in the assessment of unmanned aerial vehicle (UAV) swarm resilience based on complex networks. However, the evaluation results mostly use the initial performance state as the evaluation baseline, which is unreasonable. When UAV swarm performs a mission, as long as the network performance during the life cycle is sufficient to meet the mission requirements, it can be considered that UAV swarm has the resilience required to complete the mission. Therefore, a baseline assessment method of UAV swarm resilience based on complex networks is proposed in this paper. First, the baseline assessment method of UAV swarm resilience based on complex networks is characterized and investigated. Second, the effectiveness of the baseline assessment method is verified by simulation. The result shows that the baseline evaluation can effectively relax the evaluation result in a mission-oriented manner, and no longer use the initial state as the standard performance to measure the completion of the mission of UAV swarm. When UAV swarm performs a mission, it only needs to maintain or restore the resilience needed to complete the mission.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115455711","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378696
William Steingartner
In this paper, we present a software module developed for the course Semantics of Programming Languages. This software will be part of the intended comprehensive software package to simplify and make the teaching of formal principles in theoretical computer science more attractive. During the pandemic, the need to support illustrative and illustrative online teaching increased. Our software is designed to illustrate and visualize the translation of a higher-level language into code for the Abstract Machine, the definition of which is based on the structural operational semantics of programs.
{"title":"Compiler Module of Abstract Machine Code for Formal Semantics Course","authors":"William Steingartner","doi":"10.1109/SAMI50585.2021.9378696","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378696","url":null,"abstract":"In this paper, we present a software module developed for the course Semantics of Programming Languages. This software will be part of the intended comprehensive software package to simplify and make the teaching of formal principles in theoretical computer science more attractive. During the pandemic, the need to support illustrative and illustrative online teaching increased. Our software is designed to illustrate and visualize the translation of a higher-level language into code for the Abstract Machine, the definition of which is based on the structural operational semantics of programs.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"200 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115814088","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378645
H. Issa, J. Tar
In various technical applications the local minimum of a differentiable cost function must be found under constraints that are interpreted as embedded hypersurfaces in the whole space of search. Generally Lagrange's “Reduced Gradient Method” can be applied for solving such problems in which the Lagrange multipliers associated with the individual constraint equations have important physical interpretation, therefore it is desirable to compute them. Though in special cases this algorithm can be replaced by closed form calculations via considering the “Auxiliary Function”, in other cases the algorithmic realization cannot be avoided. In this paper it is shown that via avoiding the calculation of the individual Lagrange multipliers the algorithm can be made considerably faster especially if the constraint equations are appropriately handled. Simulation investigations are presented to substantiate the suggested method.
{"title":"Speeding up the Reduced Gradient Method for Constrained Optimization","authors":"H. Issa, J. Tar","doi":"10.1109/SAMI50585.2021.9378645","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378645","url":null,"abstract":"In various technical applications the local minimum of a differentiable cost function must be found under constraints that are interpreted as embedded hypersurfaces in the whole space of search. Generally Lagrange's “Reduced Gradient Method” can be applied for solving such problems in which the Lagrange multipliers associated with the individual constraint equations have important physical interpretation, therefore it is desirable to compute them. Though in special cases this algorithm can be replaced by closed form calculations via considering the “Auxiliary Function”, in other cases the algorithmic realization cannot be avoided. In this paper it is shown that via avoiding the calculation of the individual Lagrange multipliers the algorithm can be made considerably faster especially if the constraint equations are appropriately handled. Simulation investigations are presented to substantiate the suggested method.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115901261","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378654
Senem Tanberk, D. Tükel
The recent advance in artificial neural networks is an inspiration for automatic music generation. Deep learning algorithms help to produce pleasing melodies. They lead the creativity of musicians to be reproduced in digital environments. The proposed system learns from the Turkish popular music and then produces new music. In this study, our goal is to generate melody with a specific style, such as unforgettable soundtracks admired widely. We proposed a novel combination of convolutional neural network (CNN) and long short-term memory (LSTM) network for music generation. The experimental results reveal that the proposed combined deep model exhibits remarkable music quality compared to the lstm-only deep model or cnn-only deep model. We also conducted a survey to evaluate the quality of the generated music. The survey results show that the introduced model is capable of producing better quality and more pleasant music compared to other state-of-the-art music generation methods.
{"title":"Style-Specific Turkish Pop Music Composition with CNN and LSTM Network","authors":"Senem Tanberk, D. Tükel","doi":"10.1109/SAMI50585.2021.9378654","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378654","url":null,"abstract":"The recent advance in artificial neural networks is an inspiration for automatic music generation. Deep learning algorithms help to produce pleasing melodies. They lead the creativity of musicians to be reproduced in digital environments. The proposed system learns from the Turkish popular music and then produces new music. In this study, our goal is to generate melody with a specific style, such as unforgettable soundtracks admired widely. We proposed a novel combination of convolutional neural network (CNN) and long short-term memory (LSTM) network for music generation. The experimental results reveal that the proposed combined deep model exhibits remarkable music quality compared to the lstm-only deep model or cnn-only deep model. We also conducted a survey to evaluate the quality of the generated music. The survey results show that the introduced model is capable of producing better quality and more pleasant music compared to other state-of-the-art music generation methods.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"191 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121842399","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378694
B. Štefanovič, M. Michalíková, L. Bednarčíková, T. Tóth, R. Hudák, J. Živčák
The present article deals with assistive technologies for influencing upper limb tremor. The introduction defines the basic concepts and theoretical knowledge needed to address the issue, which is the design of a writing aid for compensation of upper limb tremor. The method section describes the principle of operation through a block diagram of the device with a subsequent description of the basic components of which the device consists. The model itself was created in SOLIDWORKS CAD software and can be manufactured by 3D printing. A preliminary price calculation of individual elements and subsequently the price value of the complete equipment was also calculated. Emphasis was placed on the weight of the device and the final design.
{"title":"KALEM- a writing aid for patients with upper limb tremor","authors":"B. Štefanovič, M. Michalíková, L. Bednarčíková, T. Tóth, R. Hudák, J. Živčák","doi":"10.1109/SAMI50585.2021.9378694","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378694","url":null,"abstract":"The present article deals with assistive technologies for influencing upper limb tremor. The introduction defines the basic concepts and theoretical knowledge needed to address the issue, which is the design of a writing aid for compensation of upper limb tremor. The method section describes the principle of operation through a block diagram of the device with a subsequent description of the basic components of which the device consists. The model itself was created in SOLIDWORKS CAD software and can be manufactured by 3D printing. A preliminary price calculation of individual elements and subsequently the price value of the complete equipment was also calculated. Emphasis was placed on the weight of the device and the final design.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123498006","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378641
J. Palša, Martin Havrilla
The aim of this paper is to create intelligent environment based on prototype platform, which is applicable in the area of the Internet of Things. Created intelligent environment allows to receive and send data in real time between the given setting and the user. Communication takes place between prototype platform Raspberry Pi created by us which communicates via script in Python and subsequently sends measured data to cloud service Firebase. Part of the intelligent environment consists of multiple sensors, which are used to control and collect data. Part of the implementation is also a web interface with the help of which the user can display the current measured values, which are visualized for better clarity. We can confirm that we do not need overpriced equipment to create appropriate intelligent environment and that we can substitute them with cheaper alternative, which this research proves.
{"title":"Transfer and Visualization of the Data in Intelligent Environment","authors":"J. Palša, Martin Havrilla","doi":"10.1109/SAMI50585.2021.9378641","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378641","url":null,"abstract":"The aim of this paper is to create intelligent environment based on prototype platform, which is applicable in the area of the Internet of Things. Created intelligent environment allows to receive and send data in real time between the given setting and the user. Communication takes place between prototype platform Raspberry Pi created by us which communicates via script in Python and subsequently sends measured data to cloud service Firebase. Part of the intelligent environment consists of multiple sensors, which are used to control and collect data. Part of the implementation is also a web interface with the help of which the user can display the current measured values, which are visualized for better clarity. We can confirm that we do not need overpriced equipment to create appropriate intelligent environment and that we can substitute them with cheaper alternative, which this research proves.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130521954","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378624
Jaromír Šulc, Katefina Šulcová
The goal of this paper is to enhance RAPTOR journey planning algorithm to return a comprehensive list of journeys to the end user, offering a wide range of meaningful alternatives. We optimize for minimal length of the journeys, while considering the number of transfers. The RAPTOR journey planning algorithm optimizes for two criteria: arrival time and number of transfers. However, when optimizing for the minimal arrival time, RAPTOR doesn't maximize the departure time, neither it finds any later alternatives. Both properties are important for the end user satisfaction with a journey planner. Skipping in time to obtain the next alternative journey can degrade or significantly slow down the algorithm. In this paper we are analyzing model situations and impact of two approaches. Firstly, cycle management of general RAPTOR, and secondly, specific setting of RAPTOR's extension - rRAPTOR - which leads to provisioning set of journeys within a time range having minimal travel time for given number of transfers. The resulting journey planner is already used in public transit routing systems in the Czech Republic. We still receive customer complaints on the amount and variability of alternative journeys provided, however the complaint rate is steadily very low, around 1 complaint raised per quarter.
{"title":"Usage of RAPTOR for travel time minimizing journey planner","authors":"Jaromír Šulc, Katefina Šulcová","doi":"10.1109/SAMI50585.2021.9378624","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378624","url":null,"abstract":"The goal of this paper is to enhance RAPTOR journey planning algorithm to return a comprehensive list of journeys to the end user, offering a wide range of meaningful alternatives. We optimize for minimal length of the journeys, while considering the number of transfers. The RAPTOR journey planning algorithm optimizes for two criteria: arrival time and number of transfers. However, when optimizing for the minimal arrival time, RAPTOR doesn't maximize the departure time, neither it finds any later alternatives. Both properties are important for the end user satisfaction with a journey planner. Skipping in time to obtain the next alternative journey can degrade or significantly slow down the algorithm. In this paper we are analyzing model situations and impact of two approaches. Firstly, cycle management of general RAPTOR, and secondly, specific setting of RAPTOR's extension - rRAPTOR - which leads to provisioning set of journeys within a time range having minimal travel time for given number of transfers. The resulting journey planner is already used in public transit routing systems in the Czech Republic. We still receive customer complaints on the amount and variability of alternative journeys provided, however the complaint rate is steadily very low, around 1 complaint raised per quarter.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"76 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113993159","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378639
Melánia Puskás, Dániel András Drexler
Mathematical models of tumor growth and the effect of therapy is fundamental for personalizing and optimizing anticancer therapies. The aim of the research is to provide a good estimation of personalized tumor model parameters based on measurements carried out on the patient. We use in silico experiments to create a large set of training data in a span that covers real-life scenarios. The data are used to train neural networks which provide a good initial guess for the model parameters. The estimated parameters can be used as initial estimations for more sophisticated, but local identification algorithms.
{"title":"Parameter identification of a tumor model using artificial neural networks","authors":"Melánia Puskás, Dániel András Drexler","doi":"10.1109/SAMI50585.2021.9378639","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378639","url":null,"abstract":"Mathematical models of tumor growth and the effect of therapy is fundamental for personalizing and optimizing anticancer therapies. The aim of the research is to provide a good estimation of personalized tumor model parameters based on measurements carried out on the patient. We use in silico experiments to create a large set of training data in a span that covers real-life scenarios. The data are used to train neural networks which provide a good initial guess for the model parameters. The estimated parameters can be used as initial estimations for more sophisticated, but local identification algorithms.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"86 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115863456","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 : 2021-01-21DOI: 10.1109/SAMI50585.2021.9378689
Abhijeet Bhattacharya, Tanmay Baweja
This manuscript presents the first-ever dataset made for supervised learning on reconstructing missing data in remotely sensed data. The types of noises present in this dataset are 1) Salt and pepper noise, caused by an error in transmission, analog-digital converter error, 2) The Landsat ETM + Scan Line Corrector (SLC)-of a problem, caused because of the poor performance of satellite sensors, cross-talk between sensors, etc. 3) Presence of thick clouds in its view due to poor atmospheric conditions. Usually, the remotely sensed data suffer an information loss because of satellite sensors' internal malfunction or poor atmospheric conditions such as thick clouds. Losing any pixel due to any external/internal error leads to a huge information loss in the images due to high spatial resolution and further tasks like detection, classification, segmentation, and many more to be applied to it. Therefore, it becomes an important task to regain the lost data before applying any other algorithm. This dataset contains a total of 21,080 images with a spatial resolution of 0.3m and 1.5m. The dataset is accessible at https://sites.google.com/view/clearviewdataset.
{"title":"Clear-View: A dataset for missing data in Remote Sensing Images","authors":"Abhijeet Bhattacharya, Tanmay Baweja","doi":"10.1109/SAMI50585.2021.9378689","DOIUrl":"https://doi.org/10.1109/SAMI50585.2021.9378689","url":null,"abstract":"This manuscript presents the first-ever dataset made for supervised learning on reconstructing missing data in remotely sensed data. The types of noises present in this dataset are 1) Salt and pepper noise, caused by an error in transmission, analog-digital converter error, 2) The Landsat ETM + Scan Line Corrector (SLC)-of a problem, caused because of the poor performance of satellite sensors, cross-talk between sensors, etc. 3) Presence of thick clouds in its view due to poor atmospheric conditions. Usually, the remotely sensed data suffer an information loss because of satellite sensors' internal malfunction or poor atmospheric conditions such as thick clouds. Losing any pixel due to any external/internal error leads to a huge information loss in the images due to high spatial resolution and further tasks like detection, classification, segmentation, and many more to be applied to it. Therefore, it becomes an important task to regain the lost data before applying any other algorithm. This dataset contains a total of 21,080 images with a spatial resolution of 0.3m and 1.5m. The dataset is accessible at https://sites.google.com/view/clearviewdataset.","PeriodicalId":402414,"journal":{"name":"2021 IEEE 19th World Symposium on Applied Machine Intelligence and Informatics (SAMI)","volume":"235 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122443610","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：