Wiphada Wettayaprasit

Wiphada Wettayaprasit

Auto-generated profile, based on publicly available information

Wiphada Wettayaprasit

Prince of Songkla University

Research Focus & Metrics

Academic Career Length

25Years (2002-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 July 2019
Verified
Conference Paper
Closed access
WSWannipa Sae-Lim
Wannipa Sae-Lim
WWWiphada Wettayaprasit
Wiphada Wettayaprasit
PAPattara Aiyarak
Pattara Aiyarak
Skin lesion classification is a particular interesting area of research in dermatoscopic lesion image processing. In this paper, we present a skin lesion classification approach based on the light weight deep Convolutional Neural Networks (CNNs), called MobileNet. We employed MobileNet and proposed the modified MobileNet for skin lesion classification. For the evaluation of our model, we had used the official dataset of Human Against Machine with 10,000 training images (HAM 10000) which was a collection of multisource dermatoscopic images. Data up-sampling and data augmentation method were used in our study for improving the efficiency of the classifier. The comparison results showed that our modified model had achieved higher accuracy, specificity, sensitivity, and F1-score than the traditional MobileNet.

1 November 2018 in Indonesian Journal of Electrical Engineering and Computer Science
Verified
Journal Article
Open access
PDF file
WMWin Win Myo
Win Win Myo
WWWiphada Wettayaprasit
Wiphada Wettayaprasit
PAPattara Aiyarak
Pattara Aiyarak
Step counter, being an active area of human daily physical activity, is an essential role in human activity determination research. As the current smartphones come with many different sensors and powerful processing capabilities, the step counting using built-in sensors in a smartphone is increasingly becoming a vital factor among many researchers. However, the step counting with a smartphone has still challenging due to many different walking behaviors and mobile phone positions. In this study, we introduce a more reliable step counter’s technique using Accelerometer sensor in a smart phone. The objective of this study is to get the accurate steps of three different walking activities in four different mobile positions. In order to achieve this, a new reliable technique based on peak is attracting considerable in our work using average acceleration. The experimental result shows 99.02% as an overall step counting performance that the proposed method reliably detects the steps under varying walking speed in different devices modes. This result is encouraging to facilitate among of the complex walking activities using built-in sensors in smartphone.

17 December 2018 in E-Learning and Digital Media
Verified
Journal Article
Closed access
ANAthitaya Nitchot
Athitaya Nitchot
WWWiphada Wettayaprasit
Wiphada Wettayaprasit
LGLester Gilbert
Lester Gilbert
The aim of this research is to introduce a method of constructing knowledge structures which represent the relationships among the knowledge items in a pedagogically useful way. A prototype, “MyTeLeMap,” was implemented to support both learners and instructors. Learners could visualize the knowledge structures of different knowledge domains and obtain links to corresponding study materials. The system also offered learning paths and recommendations, for example, to related structures. Instructors could create and share their knowledge structures. An experiment compared the learning outcomes of learners using MyTeLeMap and those using a free-browsing mode. The results showed that MyTeLeMap helped learners more than free browsing. Future work includes the incorporation of a search application for learners and of management features for instructors to upload and share learning materials linked to specific knowledge domains.

1 June 2021 in Indonesian Journal of Electrical Engineering and Computer Science
Verified
Journal Article
Open access
PDF file
TTTran Anh Tuan
Tran Anh Tuan
JDJarunee Duangsuwan
Jarunee Duangsuwan
WWWiphada Wettayaprasit
Wiphada Wettayaprasit
Aspect-based online information on social media plays a vital role in influencing people’s opinions when consumers concern with their decisions to make a purchase, or companies intend to pursue opinions on their product or services. Determining aspect-based opinions from the online information is necessary for business intelligence to support users in reaching their objectives. In this study, we propose the new aspect extraction and scoring system which has three procedures. The first procedure is normalizing and tagging part-of-speech for sentences of datasets. The second procedure is extracting aspects with pattern rules. The third procedure is assigning scores for aspects with SentiWordNet. In the experiments, benchmark datasets of customer reviews are used for evaluation. The performance evaluation of our proposed system shows that our proposed system has high accuracy when compared to other systems.

9 July 2018 in Education and Information Technologies
Verified
Journal Article
Closed access
PDF file
ANAthitaya Nitchot
Athitaya Nitchot
WWWiphada Wettayaprasit
Wiphada Wettayaprasit
LGLester Gilbert
Lester Gilbert

1 February 2019
Verified
Conference Paper
Closed access
WMWin Win Myo
Win Win Myo
WWWiphada Wettayaprasit
Wiphada Wettayaprasit
PAPattara Aiyarak
Pattara Aiyarak
Human Activity Recognition (HAR) using built-in sensors in a mobile phone is a very active research area to infer the human daily activities in recent years. The machine-learning algorithm is an important role to determine human activities. Although Artificial Neural Network (ANN) is effective classifier to determine the human activity recognition, its performance mainly depends on selection of hidden neuron nodes. In this study, we design a classifier using Multi-Layer Perceptron (MLP) for human activity recognition using UCIHAR dataset. The study introduces a new method to find the hidden neuron nodes for ANN classifier to achieve a better performance of HAR. In this scenario, we also consider the optimal hidden layers and learning rate of ANN after trained the HAR dataset. The simulation result of proposed method 98.32% accuracy is that the proposed model is a better classifier with minimal errors and fixation of hidden neurons for human activity recognition process.