Khanabhorn Kawattikul

Mr. Khanabhorn Kawattikul

orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

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

Representative Publications

View All Publications
27 August 2026 in Applied Sciences
Verified
Journal Article
Open access
PDF file
‪Chatklaw Jareanpon‪Chatklaw Jareanpon
‪Chatklaw Jareanpon
KKKhanabhorn Kawattikul
Khanabhorn Kawattikul
Joint-Embedding Predictive Architectures (JEPAs) have recently emerged as a promising paradigm for self-supervised representation learning by predicting latent embeddings from partial observations rather than reconstructing raw inputs. Although JEPA has demonstrated considerable success in computer vision and large-scale language models, the design of lightweight JEPA-inspired representation learning frameworks for natural language remains insufficiently understood. This paper proposes a lightweight span-masked JEPA-inspired framework that formulates language representation learning as latent semantic prediction from partially observed contextual inputs. The framework predicts target embeddings generated by a momentum-updated encoder while avoiding direct token reconstruction, thereby emphasizing semantic abstraction over lexical recovery. To investigate the influence of objective function design, cosine similarity and Variance–Invariance–Covariance Regularization (VICReg) are systematically compared under identical architectural and training conditions, with and without auxiliary masked language modeling (MLM) supervision. The proposed framework is evaluated on one controlled dataset and six benchmark datasets, including AG News, SST-2, IMDb, TREC, DBPedia, and Yelp Polarity. Experimental results consistently demonstrate that the cosine-based JEPA objective without auxiliary MLM supervision provides the best balance between representation quality, computational efficiency, and downstream linear probing performance. Statistical significance analysis using the Friedman test and pairwise Wilcoxon signed-rank tests with Holm correction further confirms that the cosine-based objective significantly outperforms the remaining configurations, while no significant difference is observed between VICReg with and without MLM supervision. The findings indicate that latent semantic prediction alone is sufficient for learning robust language representations under the proposed lightweight framework, whereas auxiliary token-level reconstruction does not consistently improve downstream performance despite increasing computational cost. Overall, this work provides reproducible empirical evidence and practical guidance for objective function design in lightweight JEPA-inspired language representation learning, offering an efficient experimental foundation for future self-supervised language models.

1 April 2022 in Indonesian Journal of Electrical Engineering and Computer Science
Verified
Journal Article
Open access
PDF file
KKKhanabhorn Kawattikul
Khanabhorn Kawattikul
KSKodchanipa Sermsai
Kodchanipa Sermsai
PCPhatthanaphong Chomphuwiset
Phatthanaphong Chomphuwiset
Whole slide image (WSI) processing is a common technique used in the analysis process performed by pathologists. Identifying precise and accurate regions of cancerous in the tissue is an important process in the disease diagnosis modality. This work proposes an automated technique for identifying invasive ductal carcinoma (IDC) in histology images using. An image is divided into small non-overlapped patches (or image windows). Then, the task is to classify the image patches into different classes, i.e., i) IDC and ii) non-IDC. We employ a two-stage classification-based to classify the patches, as to identify IDC regions in the tissue. In the first stage (patch-level classification), image patch classification is carried out using a conventional handcrafted feature and deep-learning technique are explored. The second stage (post-processing) undergoes a refinement process, which considers the spatial relationships between the neighboring patches. This stage aims to amend some of miss-classified patches. Markov random field (MRF) is implemented in this stage to examine the relationships of the patches and their neighborhoods. The experiments are conducted on public dataset. The experimental results show the post-processing can improve the performance of the classification in the first stage using the handcrafted-based technique and deep learning.

25 May 2026 in Applied Sciences
Verified
Journal Article
Open access
PDF file
PSPanida Songram
Panida Songram
Suchart KhummaneeSuchart Khummanee
Suchart Khummanee
JMJirapond Muangprathub
Jirapond Muangprathub
KKKhanabhorn Kawattikul
Khanabhorn Kawattikul
Thai sentiment analysis on social media remains challenging due to the lack of explicit word boundaries, informal expressions, code-mixing, short text length, and severe class imbalance, particularly in the WISESIGHT benchmark dataset. Although transformer-based models such as WangchanBERTa provide strong contextual representations for Thai language processing, conventional hybrid architectures using fixed convolutional kernels may fail to capture diverse local semantic patterns across different sentiment classes. To address this limitation, this study proposes WangchanBERTa with Instance-Class-Aware Dynamic Kernel Selection (WangchanBERTa-IC-DKS), a novel hybrid framework that integrates contextualized embeddings with multi-kernel convolutional neural networks and dynamic kernel selection at both the instance and class levels. The proposed model adaptively determines kernel importance according to sentence semantics and class-specific sentiment characteristics, enabling more effective local feature extraction and improved minority-class recognition. Experiments were conducted on the WISESIGHT sentiment benchmark using four kernel candidates: [2, 3, 4], [3, 4, 5], [2, 3, 4, 5], and [1, 2, 3, 4, 5]. The kernel candidate [2, 3, 4, 5] achieves the best performance, with a mean macro-F1 of 63.93% averaged across three independent runs using different random seeds. Ablation studies further show that combining both class-aware and instance-aware dynamic kernel selection improves balanced classification performance compared with fixed multi-kernel CNN baselines. The proposed model also outperforms the previous state-of-the-art Parallel Hybrid model by 1.23 percentage points on the WISESIGHT dataset and by 1.36 percentage points on the 40 Thai children’s tales dataset. These findings demonstrate that dynamic kernel selection is more effective than fixed convolutional kernels across the two evaluated Thai sentiment benchmarks, particularly for short, noisy, and imbalanced Thai texts.

17 March 2026 in IEEE Access
Verified
Journal Article
Open access
KKKhanabhorn Kawattikul
Khanabhorn Kawattikul
Kazuhiko HamamotoKazuhiko Hamamoto
Kazuhiko Hamamoto
Synthesizing realistic and anatomically consistent B-mode ultrasound (US) images under arbitrary ultrasound probe (US-probe) motion remains a challenging problem due to the conflicting requirements of pose-dependent anatomical alignment and stochastic speckle texture modeling. Generative models must preserve structural consistency governed by six-degree-of-freedom (6-DoF) US-probe poses while simultaneously reproducing realistic ultrasound speckle statistics. This study investigate architectural decomposition strategies for pose-conditioned ultrasound image synthesis. We propose a Three-Generator conditional GAN (CGAN) that explicitly separates translation-aware anatomical layout, rotation-dependent structural refinement, and speckle-specific texture generation into dedicated generator branches. To evaluate the effectiveness of this design, we conduct extensive comparisons against Single- and Two-Generator CGAN variants, as well as a conditional diffusion baseline, evaluated under a unified evaluation protocol. Model performance is assessed using a comprehensive evaluation protocol covering standard image fidelity metrics, ultrasound-specific realism measures, anatomy-preservation metrics, pose–image alignment, and efficiency of computation. Experimental results on a phantom dataset demonstrate that architectural decomposition consistently improves anatomical fidelity, speckle realism, and pose consistency. In particular, the proposed Three-Generator pose-conditioned CGAN achieves the most balanced performance across all evaluation criteria while maintaining stable and efficient inference. The results indicate that explicit architectural separation provides an effective framework for high-fidelity, pose-aware ultrasound image synthesis.

1 May 2023 in Indonesian Journal of Electrical Engineering and Computer Science
Verified
Journal Article
Open access
PDF file
PSPanida Songram
Panida Songram
Suchart KhummaneeSuchart Khummanee
Suchart Khummanee
PCPhatthanaphong Chomphuwiset
Phatthanaphong Chomphuwiset
‪Chatklaw Jareanpon‪Chatklaw Jareanpon
‪Chatklaw Jareanpon
LBLaor Boongasame
Laor Boongasame
KKKhanabhorn Kawattikul
Khanabhorn Kawattikul
Online deception has become a major problem affecting people, society, the economy, and national security. It is mostly done by spreading deceptive messages because message are quickly spread on social networks and are easily accessed by anyone. Detecting deceptive messages is challenging as the messages are unstructured, informal, and complex; this extends into Thai language messages. In this paper, various deep learning models are proposed to detect deceptive messages under two feature extraction trials. A balanced two-class dataset of deceptive and truthful Thai messages (n=2378) is collected from Facebook pages. Instance features are encoded using word embeddings (Thai2Fit) and one-hot encoding techniques. Five classification models, convolutional neural network (CNN), bidirectional long short-term memory (BiLSTM), bidirectional gated recurrent units (BiGRU), CNN-BiLSTM, and CNN-BiGRU, are proposed and evaluated upon the dataset with each feature extraction technique. The experimental results show that all the proposed models had excellent accuracy (95.59% to 98.74%) and BiLSTM with one-hot encoding gave the best performance, achieving 98.74% accuracy.

9 April 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
PSPanida Songram
Panida Songram
Suchart KhummaneeSuchart Khummanee
Suchart Khummanee
JMJirapond Muangprathub
Jirapond Muangprathub
KKKhanabhorn Kawattikul
Khanabhorn Kawattikul
Sentiment analysis of Thai social media texts remains challenging due to the lack of explicit word boundaries, informal language, and high linguistic variability, further exacerbated by class imbalance, short texts, and ambiguous categories. Recent transformer-based models, particularly WangchanBERTa, have shown strong performance in Thai NLP; however, their ability to extract fine-grained, sentiment-specific local features remains limited when applied to noisy, short social media texts. This study proposes a lightweight hybrid framework that integrates WangchanBERTa with multiple neural architectures, including convolutional neural networks (CNN), bidirectional long short-term memory (BiLSTM), and bidirectional gated recurrent units (BiGRU). A comparative evaluation is conducted to assess their effectiveness for four-class Thai sentiment classification. In addition, the impact of CNN kernel size configurations on sentiment feature extraction is systematically investigated, an aspect that has received limited attention in prior Thai NLP research. Experiments conducted on the WISESIGHT benchmark demonstrate that the WangchanBERTa–CNN model with medium kernel sizes [2, 3, 4] achieves the best performance. It achieves a macro-average F1-score of 65.80%, outperforming the current state-of-the-art on the WISESIGHT benchmark by 4.94%. In addition, it achieves a macro-average F1-score of 94.11% on the class-balanced dataset. These results are achieved while maintaining a relatively low parameter count. These findings confirm that combining contextualized global embeddings with local n-gram feature extraction provides an effective and efficient solution for sentiment classification of short, noisy, and linguistically diverse Thai social media texts.