KMITL

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

Browse

Search Results

Now showing 1 - 10 of 22

Some of the metrics are blocked by your
consent settings
YOLO-augment strategy with diffusion-based inpainting for enhanced traffic sign detection
(2026-01-01)
Sub-r-pa, Chayanon
;
Pavarangkoon, Praphan
;
Huang, Su Wen
;
Fan, Ming Zhong
;
Chen, Rung Ching
Traffic sign datasets often suffer from data scarcity and class imbalance, which challenge the development of robust autonomous driving systems. This article proposes a novel dataset augmentation method that leverages Stable Diffusion inpainting to generate realistic synthetic traffic signs. The method fine-tunes a Stable Diffusion model and introduces an object-size-based crop (OSB-crop) technique with mask adjustments to ensure high-quality augmentations that maintain contextual consistency. Evaluations using the Fréchet Inception Distance (FID) show average scores of 195.85 for the DFG-T10 subset and 247.077 for the DFG-B10 subset, demonstrating the ability to produce realistic inpainted signs, particularly for more represented minority classes. Qualitative analyses further highlight seamless integration into real-world scenes, although challenges remain for extremely underrepresented classes and ensuring perfect visual fidelity. The benefits of this approach include its potential to enhance traffic sign datasets, address class imbalances, and improve the potential for training more reliable autonomous driving systems by providing more diverse and realistic training data. This study focuses on evaluating the quality of the generated data itself as a foundational step toward enhancing downstream detection models. However, limitations include the computational cost of fine-tuning and the difficulty in achieving high-quality inpainting for all underrepresented classes, especially those with poor initial data quality. This work lays a strong foundation for advancing dataset augmentation techniques for real-world applications.
Some of the metrics are blocked by your
consent settings
Heuristic Approaches for Cache Node Placement in Content-Centric Networking Under Maximum Link Traffic Minimization
(2026-01-01)
Pavarangkoon, Praphan
;
Nakajima, Shohei
;
Kitsuwan, Nattapong
This paper proposes heuristic approaches for cache node placement in Content-Centric Networking (CCN) with the objective of minimizing maximum link traffic. The work builds on an exact Integer Linear Programming (ILP) formulation from our earlier study, which jointly models routing and caching decisions but becomes computationally expensive for large-scale networks. To address this limitation, we develop scalable heuristic algorithms that approximate the ILP solution while requiring much lower computation time. We introduce four heuristic approaches, comprising two Linear Programming (LP)-based local search algorithms, 2Swap and GreedySwap, which exploit the fractional solution of the LP relaxation, and two proxy-based heuristics, Population-Weighted Closeness (PWC) and Population-Weighted Betweenness (PWB), which estimate cache utility using population-weighted centrality measures without solving any optimization model. Experimental evaluations on four network topologies show that the LP-based heuristics remain within 0-6% of the ILP optimum while reducing computation time by factors ranging from about 40× to more than 160×. Compared with the proxy-based heuristics, the LP-based heuristics consistently yield lower maximum link traffic across all settings. These results indicate that the proposed heuristic approaches provide effective and scalable solutions for cache node placement in practical CCN deployments.
Some of the metrics are blocked by your
consent settings
Prediction of Medication-Related Osteonecrosis of the Jaw in Patients Receiving Antiresorptive Therapy Using Machine Learning Models
(2025-03-01)
Warin, Kritsasith
;
Lochanachit, Sirasit
;
Pavarangkoon, Praphan
;
Techapanurak, Engkarat
;
Somyanonthanakul, Rachasak
Background: Medication-related osteonecrosis of the jaw (MRONJ) is a serious complication associated with the use of antiresorptive agents, impacting patient quality of life and treatment outcomes. Predictive modeling may aid in a better understanding of MRONJ development. Purpose: The study aimed to evaluate machine learning (ML)–based models for predicting MRONJ in patients receiving antiresorptive therapy. Study Design, Setting, Sample: This retrospective in silico study analyzed electronic medical records from Thammasat University Hospital, covering the period from January 2012 to December 2022. The sample included subjects receiving antiresorptive therapy, excluding those with a history of radiation therapy or metastatic jaw disease. Predictor Variables: The primary predictor variable was the predicted probability of MRONJ development from the ML models. Outcome Variables: The outcome variable was MRONJ status coded as present or absent based on chart review. Covariates: Covariates included demographic data, MRONJ occurrence, location and staging of MRONJ, comorbidities, diseases related to antiresorptive agents, types of antiresorptive agents, therapy duration, concurrent medications, blood calcium levels, and dental factors. Analyses: Model performance was assessed via accuracy, sensitivity, specificity, positive and negative predictive values, and the area under the receiver operating characteristic curve. Additionally, univariate and multivariate Cox regression analyses were conducted to identify factors significantly associated with MRONJ development. P ≤ .05 was statistically significant. Results: The study analyzed data from 5,305 subjects with a mean age of 75 ± 11.1 years, predominantly female. MRONJ was observed in 81 cases (1.5%), with a median time to development of 33 months (interquartile range = 3). Among the 6 models tested, the best-performing model had an accuracy of 0.95 and an area under the receiver operating characteristic curve of 0.89-0.90. Significant predictors identified through Cox regression included metabolic syndrome (hazard ratio = 14.064, 95% confidence interval = 1.111-178.067, P = .041) and patients receiving intravenous pamidronate (hazard ratio = 5.932, 95% confidence interval = 1.755-20.051, P = .004), indicating their association with MRONJ development. Conclusions and Relevance: ML-based predictive and time-to-event models effectively predict MRONJ risk, aiding in the strategic prevention and management for patients undergoing antiresorptive therapy.
Some of the metrics are blocked by your
consent settings
Fairness-Aware Computation Offloading for Mobile Edge Computing with Energy Harvesting
(2025-01-01)
Triyanto, Dedi
;
Wayan Mustika, I.
;
Widyawan
;
Pavarangkoon, Praphan
Mobile edge computing (MEC) improves network performance by minimizing latency and assigning computing tasks to edge servers. Nonetheless, delegating computations in environments with high device density poses considerable difficulties. Ensuring fairness in resource distribution among users is essential for preserving network stability and user satisfaction in these contexts. This research formulates the Fairness-aware Computation Offloading Optimization (FACOO) algorithm. The Lyapunov approach and sequential least squares quadratic programming (SLSQP) are used to ascertain the best offloading ratio, transmission power, and CPU frequency while complying with signal-to-interference-plus-noise ratio (SINR) limitations. Energy harvesting (EH) is built into FACOO to prolong device battery life and to ensure that MEC systems, which have limited resources, are more sustainable. The results show that FACOO greatly improves throughput and fairness while using significantly less energy, especially in settings with numerous nodes dispersed across large areas. Comprehensive simulations demonstrate that the method effectively balances fairness, throughput, and energy use, making it a workable way to improve resource allocation in MEC systems.
Some of the metrics are blocked by your
consent settings
Automatic Smoke/Forest Fire Detecting System based on Visual IoT
(2025-01-01)
Tungpimolrut, Kanokvate
;
Karnjana, Jessada
;
Chatpoj, Montri
;
Kitbutrawat, Nathavuth
;
Pavarangkoon, Praphan
In this paper, a visual IoT system or IP camera system for early detection of smoke/forest fire has been investigated. The overall system implementation including hardware and software as well as system installation in targeted areas in Chiangmai, Thailand have been also described. The dataset construction, the preliminary model development and testing have been conducted based on YOLOv5. The model improvement based on Fast Segment Anything Model with YOLOv11 has been proposed to improve False Positive. The results show good performance of Mean Square Error.
Some of the metrics are blocked by your
consent settings
Energy-Efficient and Fair Computation Offloading for Multi-user MEC with EH Devices
(2025-01-01)
Mustika, I. Wayan
;
Triyanto, Dedi
;
Halimah, Noor Siti
;
Pavarangkoon, Praphan
The increasing demand for low-latency and energy-efficient mobile applications has propelled the development of Mobile Edge Computing (MEC), enabling the offloading of computational activities from resource-constrained Mobile Devices (MDs) to nearby edge servers. This study examines a joint problem of computation offloading and resource allocation issue in a wireless multi-user, multi-server MEC system with Energy Harvesting (EH) capabilities. Our objective is to reduce long-term energy consumption while adhering to limitations related to latency, energy causality, server capacity, and Signal-To-Interference-Plus-Noise Ratio (SINR). To address the complexities of system dynamics and uncertainty in energy arrivals, we propose a low-complexity online approach utilizing Lyapunov optimization. The proposed method dynamically modifies offloading ratios, transmission power, CPU frequencies, and server allocations without requiring future data. The simulation results show that our method achieves significant energy savings, has low delays, and ensures fairness among users, even in highly congested scenarios. A comparative analysis with benchmark algorithms validates the efficacy and resilience of the proposed framework in real MEC situations.
Some of the metrics are blocked by your
consent settings
Evaluation and Optimization of LLM and RAG Components for a Post-Operative Oral Surgery Consultation Chatbot
(2025-01-01)
Lochanachit, Sirasit
;
Bunlaue, Patcharamon
;
Kaewmuneechoke, Chanapat
;
Wilairatanaporn, Nopasorn
;
Trachoo, Vorapat
The increasing demand for dental services highlights the need for efficient post-operative oral surgery consultations. Many patients experience anxiety due to limited knowledge of oral care and treatment. This study introduces a chatbot prototype integrating Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs) to provide accurate, context-aware responses. The research evaluates various LLMs, embedding models, and chunking techniques to enhance chatbot performance. The multilingual-e5-large embedding model excelled in retrieval tasks due to its multilingual training, instruction tuning, and contrastive pre-training, ensuring high retrieval precision. The Hybrid Chunking method was selected for its ability to segment text contextually, combining Markdown-based, token-based, and semantic segmentation for optimal chunk relevance. The Llama3.3 (70B) model was chosen for its superior fluency, relevance, and ability to handle complex dependencies. The results demonstrate that combining the multilingual-e5-large embedding model, Hybrid Chunking technique, and Llama3.3 (70B) model improves retrieval precision, response accuracy, and relevance, enhancing patient care and operational effectiveness of dental staffs.
Some of the metrics are blocked by your
consent settings
Impact of Rhythm, Tempo, and Rest Variations on Pitch Detection in Deep Learning-Based Piano Transcription Models
(2024-01-01)
Pangwapee, Priyakorn
;
Mekkoktanphira, Juthakan
;
Dilokthanakul, Nat
;
Lochanachit, Sirasit
;
Kanungsukkasem, Nont
This paper investigates the impact of rhythm, tempo, and rest variations on pitch detection in deep learning-based models for piano transcription. We conducted a series of experiments using GRU and Transformer architectures, manipulating note lengths, rhythmic patterns, and rest intervals to assess their effect on pitch transcription accuracy. Our findings indicate that model performance is significantly influenced by these musical factors. The experiment with GRU shows notable sensitivity to rhythmic and rest changes. However, the Transformer model handles varied conditions more robustly. These findings help refine our approach to music transcription software, particularly in improving pitch recognition across varied rhythmic patterns, tempos and rests.
Some of the metrics are blocked by your
consent settings
Cache Node Placement Scheme Considering Maximum Traffic in Content-Centric Networks
(2023-01-01)
Pavarangkoon, Praphan
;
Nakajima, Shohei
;
Kitsuwan, Nattapong
This paper proposes a cache node placement scheme considering the maximum traffic in content-centric networks (CCNs). The cache node placement problem is considered to satisfy the user's requirement in CCNs. Traffic utilization is one of the most common requirements. Reduced traffic allows more additional traffic on links. In this paper, a scheme to minimize the maximum traffic and the number of hops is proposed. The mathematical model for the cache node placement problem is formulated. The dynamic routing is considered in this model. Numerical result shows that the proposed scheme outperforms the conventional scheme. It suggests that the proposed scheme provides reference values to support the implementation of heuristic algorithms for the cache node placement problem.
Some of the metrics are blocked by your
consent settings
Reducing Bandwidth Blocking Rate in Elastic Optical Networks Through Scale-Based Slicer Placement Strategy
(2023-01-01)
Kitsuwan, Nattapong
;
Pavarangkoon, Praphan
In this paper, we present a scale-based slicer placement strategy aimed at decreasing the bandwidth blocking rate (BBR) in elastic optical networks (EONs). EONs adopt slicing-and-stitching technology to bypass the requirement of consecutive spectrum slots on the same link for a single request. This is achieved through the use of slicers to split the requested spectrum band into multiple sub-spectrum components. However, a uniform distribution of slicers across all nodes may not guarantee low BBR due to varying traffic volumes on each node. Our proposed scale-based policy determines the necessary number of slicers for each node based on our investigation. Simulation results demonstrate that our strategy leads to a 91% reduction in BBR compared to conventional methods.

KMITL

Browse

Filters

Settings

Sort By

Results per page

Search Results