Search Results

Now showing 1 - 5 of 5

Some of the metrics are blocked by your
consent settings
Comparison of Data Augmentation Techniques for Thai Text Sentiment Analysis
(2023-01-01)
Rongsawad, Kanda
;
Chatwiriya, Watchara
Many text corpora e.g., Thai text, have problems with low amounts of data or imbalanced data. Data augmentation is one technique used to solve this problem by generating synthetic data. Various data augmentation techniques are applied to text, each method has different algorithms to generate new text. This paper will compare the performance of three well-known data augmentation techniques: Easy Data Augmentation, Embedding Replacement, and Replacement by the masked language model. We apply these data augmentation techniques to the Thai text sentiment corpus to generate synthetic text, then the synthetic text will be used to train the sentiment classifier. After that, the performances of classifiers are analyzed and compared. Three sentiment classifiers used in this paper are Convolutional Neural Network (CNN), Bidirectional Long-Short Term Memory (Bi-LSTM), and Gated Recurrent Unit (GRU). From the results, we found that data augmentation techniques can improve sentiment classifiers’ performance. The EDA technique outperformed the others in the Cosmetics Drink and All topics sets but in Restaurant sets, MLM is the best technique that can improve Thai sentiment classification performance. The results show that EDA can be used as a baseline for further study of data augmentation.
Some of the metrics are blocked by your
consent settings
Improvement on vehicle trajectory reconstruction using Geometric Road Network (GRN)
(2018-11-17)
Chomputawat, Anuthep
;
Chatwiriya, Watchara
;
Chomputawat, Sirikanya
This paper presents the technique using geometric road network (GRN) to improve data fusion method for vehicle trajectory analysis and reconstruction. The experiment used two methods of data fusion which are Linear-interpolation-based one and Kalman-filter-based one and used the simulation of detection and localization of vehicle moving in straight, big zigzag, small zigzag and arbitrary trajectories, with localization errors and multi-level loss of data. By comparing the data fusion methods with GRN and the ones without GRN, the result shows that the data fusion methods with GRN can reduce localization errors better than the ones without GRN in all level of data loss and all type of trajectories. In conclusion, using GRN can improve the efficiency of vehicle trajectory analysis and reconstruction.
1
Some of the metrics are blocked by your
consent settings
Applying epidemic algorithm for financial service based on blockchain technology
(2018-08-13)
Poonpakdee, Pasu
;
Koiwanit, Jarotwan
;
Yuangyai, Chumpol
;
Chatwiriya, Watchara
Our global market is emerging transformation strategy that can make the difference between success and failure. Smart contract systems through developments of technological innovations are increasingly seen as alternative technologies to impact transactional processes significantly. Blockchain is a smart contract protocol with trust offering the potential for creating new transaction platforms and thus shows a radical change of the current core value creation in third parties. These results in enormous cost and time savings and the reduced risk for the parties. This study proposed a method to improve the efficiency of distributed consensus in blockchains using epidemic algorithm. The results showed that epidemic protocols can distribute the information similar to blockchain.
3
Some of the metrics are blocked by your
consent settings
Mobile Crime Incident Reporting System using UX dimensions guideline
(2015-05-21)
Vatanasuk, Natruja
;
Chomputawat, Anuthep
;
Chomputawat, Sirikanya
;
Chatwiriya, Watchara
Crime incidents tend to increase steadily. To effectively respond to the incidents, police officers need to have a highly efficient incident reporting system. However, the telephone-based incident reporting system currently used has some limitations. It cannot cope with current problems and meet all requirements. Since mobile devices, especially smartphones, are prevalent, widely used and can transmit various types of information, they thus become an additional channel to the telephone-based incident reporting system. This paper proposes Mobile Crime Incident Reporting System, which comprises two parts. The incident report part was developed as a mobile application. To make it suitable for reporting crime or emergency incidents, UX dimensions guideline was used in the following order; 1) Identification 2) Emotion 3) Stimulation 4) Social relatedness/Co-experience 5) Visual and aesthetic experience and 6) Meaning and Value. The incident response part was developed as a web application. The system was tested by two groups of police officers; one acted as incident reporters, the other as incident response officers. The test result showed that the officers were satisfied with the application's user interface. They could easily learn how to use the system. The system also helped reduce the officers' response time because the system provided enough information which the officers needed to make quick and appropriate decision.
Some of the metrics are blocked by your
consent settings
Forest fire smoke detection in video based on digital image processing approach with static and dynamic characteristic analysis
(2011-08-18)
Surit, Surapong
;
Chatwiriya, Watchara
In this paper, propose a method for forest fire smoke detection in video based on digital image processing approach with static and dynamic characteristic analysis. The proposed method is composed of four steps. The first step is to detect the area of change in the current input frame against the background image. Second step is to locate regions of interest (ROIs) by connected component algorithm, calculate area of ROI by convexhull algorithm and segment the area of change from image. Third step is to analysis and calculate static and dynamic characteristic. Finally, we decide whether the objects that have changed in that picture is the smoke or not. The experimental result with simulated forest fire smoke and the moving object video show that our method effectively and accuracy detect fire smoke. © 2011 IEEE.
1

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results