Pub Date : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00036
Yu-Hsiang Chang, Hung-Chin Jang
The prosperity of the social economy, tourism, and entertainment industry are important factors to cause traffic congestion. In addition to commuting hours and holidays, if a large-scale event, such as a concert, a sporting event or an exhibition is held, it is easy to make traffic congestion even worse. If we know in advance the time and place of the large-scale event, then we can accurately forecast the future traffic flow and plan the driving route. It helps effectively relieve traffic flow, reduce travel time and carbon emissions. In this study, we used the Vehicle Detector (VD) [12] data from the Taipei City Government Open Data Platform as a source of regular traffic data as well as the data of Forecastable Sporadic Event (FSE), such as a large-scale event, to forecast traffic flow. The information of time and place of the FSE are collected from various information websites (ticketing websites, tourist websites, etc.) by web crawlers. We proposed a Long Short-Term Memory (LSTM) deep learning model for traffic flow forecast, which was trained with both VD and FSE data. We further used Adam Optimizer to adjust the weight and bias of the model to optimize the forecast accuracy. The implementation of the LSTM model was conducted in TensorFlow, a machine learning framework developed by Google. Finally, we evaluated the forecast accuracy of the model by Mean Absolute Percentage Error (MAPE) and analyzed the effectiveness of applying FSE data to traffic forecast.
{"title":"Traffic Flow Forecast for Traffic with Forecastable Sporadic Events","authors":"Yu-Hsiang Chang, Hung-Chin Jang","doi":"10.1109/Ubi-Media.2019.00036","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00036","url":null,"abstract":"The prosperity of the social economy, tourism, and entertainment industry are important factors to cause traffic congestion. In addition to commuting hours and holidays, if a large-scale event, such as a concert, a sporting event or an exhibition is held, it is easy to make traffic congestion even worse. If we know in advance the time and place of the large-scale event, then we can accurately forecast the future traffic flow and plan the driving route. It helps effectively relieve traffic flow, reduce travel time and carbon emissions. In this study, we used the Vehicle Detector (VD) [12] data from the Taipei City Government Open Data Platform as a source of regular traffic data as well as the data of Forecastable Sporadic Event (FSE), such as a large-scale event, to forecast traffic flow. The information of time and place of the FSE are collected from various information websites (ticketing websites, tourist websites, etc.) by web crawlers. We proposed a Long Short-Term Memory (LSTM) deep learning model for traffic flow forecast, which was trained with both VD and FSE data. We further used Adam Optimizer to adjust the weight and bias of the model to optimize the forecast accuracy. The implementation of the LSTM model was conducted in TensorFlow, a machine learning framework developed by Google. Finally, we evaluated the forecast accuracy of the model by Mean Absolute Percentage Error (MAPE) and analyzed the effectiveness of applying FSE data to traffic forecast.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134329774","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00014
Tipajin Thaipisutikul, Yi-Cheng Chen, Lin Hui, Sheng-Chih Chen, P. Mongkolwat, T. Shih
Reinforcement Learning (RL) is an extraordinarily paradigm that aims to solve a complex problem. This technique leverages the traditional feedforward networks with temporal-difference learning to overcome supervised and unsupervised real-world problems. However, RL is one of state-of-the-art topic due to the opaque aspects in design and implementation. Also, in which situation we will get performance gain from RL is still unclear. Therefore, This study firstly examines the impact of Experience Replay in Deep Q-Learning agent with Self-Driving Car application. Secondly, The impact of Eligibility Trace in RNN A3C agents with Breakout AI game application is studied. Our results indicated that these two techniques enhance RL performance by more than 20 percent as compared with traditional RL methods.
{"title":"The Matter of Deep Reinforcement Learning Towards Practical AI Applications","authors":"Tipajin Thaipisutikul, Yi-Cheng Chen, Lin Hui, Sheng-Chih Chen, P. Mongkolwat, T. Shih","doi":"10.1109/Ubi-Media.2019.00014","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00014","url":null,"abstract":"Reinforcement Learning (RL) is an extraordinarily paradigm that aims to solve a complex problem. This technique leverages the traditional feedforward networks with temporal-difference learning to overcome supervised and unsupervised real-world problems. However, RL is one of state-of-the-art topic due to the opaque aspects in design and implementation. Also, in which situation we will get performance gain from RL is still unclear. Therefore, This study firstly examines the impact of Experience Replay in Deep Q-Learning agent with Self-Driving Car application. Secondly, The impact of Eligibility Trace in RNN A3C agents with Breakout AI game application is studied. Our results indicated that these two techniques enhance RL performance by more than 20 percent as compared with traditional RL methods.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130234430","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00025
Juying Wu, Qing Han, Yi Li
With the unprecedented development of Computer Aided Instruction, integrating information technology into the teaching of Chinese character writing has become a trend. As an important part of Chinese character writing teaching supported by mobile platform, the correctness of judgment automatically plays an important role. On this basis, this paper designs and develops a method for judging the correctness of handwritten Chinese characters based on feature matrix. It firstly extracts the stroke features which includes stroke orientation, stroke length, absolute position and combination relationship of the stroke, then similarity matching is achieved by the feature matrix. This method can realize the one-to-one correspondence between the user's handwritten Chinese strokes and the standard ones, making the whole character correctness judgement and the specific error strokes and error types locating possible, which can be applied to Chinese character writing training and teaching.
{"title":"Study on Correctness Judgement of Handwritten Chinese Characters Based on Feature Matrix for Similarity Matching","authors":"Juying Wu, Qing Han, Yi Li","doi":"10.1109/Ubi-Media.2019.00025","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00025","url":null,"abstract":"With the unprecedented development of Computer Aided Instruction, integrating information technology into the teaching of Chinese character writing has become a trend. As an important part of Chinese character writing teaching supported by mobile platform, the correctness of judgment automatically plays an important role. On this basis, this paper designs and develops a method for judging the correctness of handwritten Chinese characters based on feature matrix. It firstly extracts the stroke features which includes stroke orientation, stroke length, absolute position and combination relationship of the stroke, then similarity matching is achieved by the feature matrix. This method can realize the one-to-one correspondence between the user's handwritten Chinese strokes and the standard ones, making the whole character correctness judgement and the specific error strokes and error types locating possible, which can be applied to Chinese character writing training and teaching.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121330908","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00042
Kuan-Hsien Wu, Wan-Lun Tsai, Tse-Yu Pan, Min-Chun Hu
DukeMTMCT is the largest and most completely labeled dataset in Multi-Target Multi-Camera Tracking (MTMCT). We investigate a state-of-the-art work on DukeMTMCT named DeepCC, and dig out two main problems. The first problem is that the openpose is prone to false detection, which seriously affects performance. The second problem is that two different persons may be assigned with the same ID. According to the corresponding problems, we not only propose a method to measure the similarity between detected bounding box and its original background avoiding false detection caused by OpenPose, but also design a strategy to correct the tracking trajectories which are affected by the unreliability of the correlation matrix clustering method proposed by DeepCC. Our method outperforms the state-of-the-art on DukeMTMCT.
{"title":"Improving Performance of DeepCC Tracker by Background Comparison and Trajectory Refinement","authors":"Kuan-Hsien Wu, Wan-Lun Tsai, Tse-Yu Pan, Min-Chun Hu","doi":"10.1109/Ubi-Media.2019.00042","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00042","url":null,"abstract":"DukeMTMCT is the largest and most completely labeled dataset in Multi-Target Multi-Camera Tracking (MTMCT). We investigate a state-of-the-art work on DukeMTMCT named DeepCC, and dig out two main problems. The first problem is that the openpose is prone to false detection, which seriously affects performance. The second problem is that two different persons may be assigned with the same ID. According to the corresponding problems, we not only propose a method to measure the similarity between detected bounding box and its original background avoiding false detection caused by OpenPose, but also design a strategy to correct the tracking trajectories which are affected by the unreliability of the correlation matrix clustering method proposed by DeepCC. Our method outperforms the state-of-the-art on DukeMTMCT.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130573698","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 : 2019-08-01DOI: 10.1109/Ubi-Media.2019.00054
Chun-Hsiung Tseng, Yung-Hui Chen, Jia-Rou Lin
In this research, we proposed some modules for physical signal collection. Despite of the fact that there are already quite a few code examples for programming in microcontrollers, it is still not easy to adopt these code snippets directly and some manual adjustments may be needed. In this manuscript, we proposed some modules to simplify the building of physical signal collection applications. Specifically, we proposed the following modules as scaffolds: a set of pre-built data reading modules, an executable script, a development tool, a Web service for I/O, and some GUI modules.
{"title":"A Tool-Set for Physical Signal Collection","authors":"Chun-Hsiung Tseng, Yung-Hui Chen, Jia-Rou Lin","doi":"10.1109/Ubi-Media.2019.00054","DOIUrl":"https://doi.org/10.1109/Ubi-Media.2019.00054","url":null,"abstract":"In this research, we proposed some modules for physical signal collection. Despite of the fact that there are already quite a few code examples for programming in microcontrollers, it is still not easy to adopt these code snippets directly and some manual adjustments may be needed. In this manuscript, we proposed some modules to simplify the building of physical signal collection applications. Specifically, we proposed the following modules as scaffolds: a set of pre-built data reading modules, an executable script, a development tool, a Web service for I/O, and some GUI modules.","PeriodicalId":259542,"journal":{"name":"2019 Twelfth International Conference on Ubi-Media Computing (Ubi-Media)","volume":"87 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126178331","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：