KMITL

Permanent URI for this communityhttps://dspace.kmitl.ac.th/handle/123456789/1

Browse

Search Results

Now showing 1 - 10 of 29

Some of the metrics are blocked by your
consent settings
Enhancing Industrial Machine Sound Anomaly Detection Using STFT Integrated with DWT and Autoencoder-Based Neural Networks
(2026-01-23)
Mattanaweerapong, Tawan
;
Jearanaitanakij, Kietikul
This study proposes a hybrid feature extraction approach that integrates the Discrete Wavelet Transform (DWT) with the Short-Time Fourier Transform (STFT) to improve the accuracy of anomalous sound detection in industrial machines. Conventional STFT-based methods, while effective in representing time–frequency characteristics, exhibit limitations in handling non-stationary noise and transient variations, which often lead to reduced anomaly detection performance in practical industrial environments. To address this problem, the proposed method incorporates multiresolution analysis through DWT, enhancing the system’s capability to capture both spectral and temporal information with improved noise robustness. The MIMII dataset (valve,-6 dB, ID02) was used to evaluate the model, where the DWT–STFT feature, representation was applied to an autoencoder for unsupervised anomaly detection. Experimental results demonstrate that the integration of DWT effectively enhanced noise robustness and improved classification metrics, achieving higher AUC and F1-scores compared to the baseline STFT-based approach. In conclusion, the proposed DWT–STFT fusion provides a more resilient and discriminative feature representation, making it a promising technique for practical industrial anomaly detection systems.
Some of the metrics are blocked by your
consent settings
Stock Clustering Framework using Financial Ratios: A Case Study in the Stock Exchange of Thailand
(2025-10-01)
Jearanaitanakij, Kietikul
;
Poonpon, Natdanai
;
Wongtep, Chanidapa
;
Buriyameathakul, Kittaporn
;
Pimsupaporn, Artitaya
Value investors typically seek undervalued stocks that align with specific financial criteria to maximize their margin of safety. However, manually analyzing the financial data of all listed stocks is a time-intensive process. Furthermore, the market price of a target stock may exceed its intrinsic value, introducing potential investment risks. To address these challenges, this study proposes a stock clustering framework that groups equities based on financial ratio similarity. The proposed framework is designed to streamline the investment decision-making process by recommending stocks with comparable financial profiles as alternatives to those currently attracting investor interest but that may already be overvalued. Multiple clustering algorithms are evaluated to determine the most effective grouping strategy. Empirical back testing using four years of data from the Stock Exchange of Thailand reveals that the Gaussian Mixture Model (GMM) achieves the highest composite performance metric among the tested methods. Additionally, the HDBSCAN algorithm is employed to detect and exclude outlier stocks, thereby enhancing the reliability of the clustering results.
Some of the metrics are blocked by your
consent settings
Quality Classification of Sunglasses Lens by Deep Learning
(2025-01-01)
Theskham, Charin
;
Jearanaitanakij, Kietikul
Sunglasses lens are medical devices designed to correct human vision while also providing protection against ultraviolet (UV) radiation. Sunglasses lenses come in various colors, each offering different light-filtering properties. Currently, the quality classification of sunglasses lenses during the production process still relies on human vision. Therefore, this research aims to study the design and quality classification of sunglasses lens with both evenness and unevenness colors using machine vision and deep learning techniques. However, from the review of existing studies on sunglasses lens quality inspection, informal research has been conducted in this area. This work is considered a new contribution with potential for practical application in the industry. In this study, a Convolutional Neural Network (CNN) will be used. To save research time, the researchers employed transfer learning techniques using models such as VGG16, VGG19, InceptionV3, Xception, DenseNet121, ResNet50, EfficientNetB0, EfficientNet -B7, and EfficientNetV2L. The classification results are divided into 2 categories for evenness colors lens and unevenness colors lens. The dataset used real photos of sunglasses lens from Hoya Lens Thailand as the data source. A dataset of 1,250 real images of sunglasses lens was used, comprising 625 images of evenness colors lens and 625 images of unevenness colors lens. Data augmentation was performed by rotating the images 90, 180, and 270 degrees, as well as vertically flipping the sunglasses lens images. This process yielded a total of 10,000 images, with 5,000 images each for lenses with evenness and unevenness colors. The results of applying Transfer Learning of each model for classifying the quality of sunglasses lens that the DenseNet121 model achieved the highest performance, with an accuracy of 82.23 % and precision of 82.32 %
Some of the metrics are blocked by your
consent settings
An Ensemble Model of Dual Learning for Gambling and Pornographic Websites Classification
(2025-01-01)
Thianphan, Sirapat
;
Jearanaitanakij, Kietikul
The rapid proliferation of pornographic and gambling websites poses significant challenges, as these platforms increasingly employ sophisticated techniques to evade detection. Traditional classification approaches that rely on a single feature often fail to achieve high detection rates due to the diverse strategies these websites use to bypass detection systems. To address this limitation, this study introduces an ensemble model for classifying pornographic and gambling websites by integrating two key features: URLs and textual content. A webscraping script was developed to extract textual data from HTML elements of 3,000 websites, evenly distributed among benign, pornographic, and gambling categories, specifically curated for Thai users. The URLs undergo preprocessing to capture their meaningful semantic properties, which reflect the characteristics of the corresponding websites. Separate classifiers were then trained on each feature before being integrated into an ensemble model for final prediction. This approach achieved an outstanding accuracy of 96.83%, significantly surpassing single-feature classifiers. Moreover, the findings demonstrate the proposed model's robustness against obfuscation techniques and anti-crawling mechanisms, underscoring its potential for effective automated detection.
Some of the metrics are blocked by your
consent settings
Fast Hybrid Approach for Thai News Summarization
(2024-07-01)
Jearanaitanakij, Kietikul
;
Boonpong, Suratan
;
Teainnagrm, Kirttiphoom
;
Thonglor, Thanakrit
;
Kullawan, Tiwat
News summarization presents a significant challenge in Natural Language Processing (NLP). Lengthy news articles not only consume valuable time but also lead to confusion regarding key points. The ideal news summarization should swiftly produce a succinct summary while retaining the essence of the information conveyed by the news writer. While intelligent chatbots like ChatGPT and Gemini offer user-friendly text summarization, their embedded Large Language Model (LLM) cannot be downloaded for private use. Moreover, implementing them in a business process can be expensive, both in terms of pay-per-use costs and response time. The objective of this research is to develop a private Thai news summarization that effectively extracts sentences encapsulating the main idea and abstractly summarizes them. The proposed model consists of two components. The first extracts a contiguous region containing important sentences using the TextRank algorithm, while the second employs the finetuned mBART as an LLM to generate the abstractive summary from the previously extracted sentences. In other words, the proposed model extracts an important news region before passing it to mBART. This approach produces a news summary with key information and a syntactic style akin to the natural Thai language. We evaluate the summarization quality by ROUGE scores and BERTScore (precision, recall, and F1-score). The evaluation metrics Experimental results on the ThaiSum dataset show relatively high ROUGE scores and BERTScore for the proposed model compared to most of the other approaches. Furthermore, it significantly reduces the runtime, keeping it within a reasonable limit.
Some of the metrics are blocked by your
consent settings
Thai Question-Answering System Using Similarity Search and LLM
(2024-07-01)
Jearanaitanakij, Kietikul
;
Srithongdee, Chananchida
;
Ketkham, Sirinoot
;
Ardsana, Onwanya
;
Kullawan, Tiwat
A question-answering (QA) system is essential to an organization where numerous QA pairs respond to customer queries. Choosing the right pair corresponding to the query is a complex task. Although the QA system from a commercial product like ChatGPT provides an excellent solution, it is costly, and the fine-tuned Large Language Model (LLM) cannot be downloaded for private use at the local site. In addition, the cost of using such LLM may significantly increase when the number of users grows. We propose a Thai QA system that can swiftly respond and correctly match the user query to the reference answer in the QA dataset. The proposed system encodes both QA pairs and a query into individual embeddings and finds a couple of QA pairs that are most related to the query by using the fast similarity search called Faiss (Facebook AI Similarity Search.) Afterward, the relevant QA pairs and the query are fed to the fine-tuned LLM (WangchanBERTa-pretraining multilingual transformer-based) to choose the single best match QA pair. The fine-tuned WangchanBERTa can retrieve the correct answer and respond to the query naturally. The experiment conducted on the Thai Wiki QA dataset indicates the superior ROUGE values, precision, recall, F1-score, and runtime of the proposed system against other strategies.
Some of the metrics are blocked by your
consent settings
Compact Thai Sign Language Translation by Deep Learning
(2024-01-01)
Jearanaitanakij, Kietikul
;
Choojan, Piyathida
;
Thongtem, Piyada
Sign language translation is a challenging problem in natural language processing. Its principle involves machine translation from sign language images to spoken language text. Designing a good translation is not a trivial task since there are a large number of both input image pixels and output classes. We propose the deep learning model to translate static gestures of Thai sign language (TSL) to the corresponding Thai spoken words. The main objective is to design a compact model that delivers high performance so that it can be implemented on mobile devices. Several mobile convolutional neural networks (CNN) are investigated to find the best backbone architecture. We also attach additional layers to the selected CNN architecture to fine-tune its performance. The experiments on the dataset collected from twenty-four volunteers indicate excellent results; in terms of precision, recall, and f1-score, of the proposed model. The comparisons with the state-of-the-art models and the feature visualizations from convolution layers endorse its effectiveness.
Some of the metrics are blocked by your
consent settings
DEEP LEARNING DEVELOPMENTAL MODEL FOR EMBRYO GRADING IN THAI POPULATION
(2024-01-01)
Wiriyasirivaj, Budsaba
;
Limkiatsataporn, Sawit
;
Pukinghin, Apisit
;
Kuekulkomain, Phatrapron
;
Promrungrueng, Pornprom
In light of the growing challenges associated with infertility, an increasing number of researchers are resorting to assisted reproductive technologies such as In Vitro Fertilization (IVF). Embryo grading is a crucial step in the IVF process that requires embryologists’ expertise. However, their limited availability has led to the exploration of technological alternatives. This study aims to use deep learning for human embryo grading, with models specifically designed for the dataset in Thailand at Vajira Hospital. The process of IVF at Vajira Hospital presents its own set of challenges since its embryo classification extends beyond the Istanbul consensus. Furthermore, classes that occur infrequently are removed and classes with similarities are merged. We apply transfer learning to pretrained deep learning models like VGG19, VGG16, ResNet152, Resnet101, Xception, InceptionV3, and EfficientNet, to create a system capable of accurately classifying embryo quality scores. Experimental results from the embryo dataset collected from Vajira Hospital in Thailand demonstrate the proposed classifier’s accuracy, precision, recall, f1-score, and AUC superiority. This research contributes to the field of IVF in Thailand by potentially reducing human errors and addressing the demand for skilled embryologists.
Some of the metrics are blocked by your
consent settings
An adaptive traffic light control system using reinforcement learning
(2022-07-01)
Jearanaitanakij, Kietikul
;
Jamkhaw, Chanayut
;
Puangpipat, Nattapat
;
Worasrivisal, Tot
Traffic signal control (TSC) is a challenging issue in managing an urban transportation system. A fixed time TSC is easy to implement but has drawbacks in such measures as flow rate, waiting time, and traffic density. The situation gets worse when the arrival rates of vehicles periodically change over time, which is usual in most urban cities. We propose adaptive reinforcement learning (RL) to manage TSC with varying vehicle arrival rates. Our objectives are to improve the averages of flow rate and waiting time and reduce the wasteful green light problem by considering the vehicle densities of the current lane and the downstream directions. Experiments were conducted by Simulation of Urban MObility (SUMO) under three traffic layouts and various vehicle arrival rates. The proposed method not only reduced on average traffic density, waiting time, and queue length, but also increased the average flow rate and average speed, relative to the other algorithms tested.
Some of the metrics are blocked by your
consent settings
Improving the ID3 Algorithm By Filtering Out Attributes With Values Of 0 or 1
(2022-01-01)
Chuenprasertsuk, Pinyarat
;
Jearanaitanakij, Kietikul
The iterative Dichotomiser 3 (ID3) algorithm is a classification algorithm that generates a decision tree, and it is one of the most simple and well-known tools. However, there is still room for improvement. This paper aims to reduce the running time of generating a decision tree by ignoring attributes that have an information gain value of 0 or 1 and the experiment result shows that the improved ID3 algorithm can significantly reduce the running time by more than 10 percent when compared with the traditional ID3 algorithm.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results