Southeast Asia (SEA) is a region of extraordinary linguistic and cultural diversity, yet it remains significantly underrepresented in vision-language (VL) research. This often results in artificial intelligence (AI) models that fail to capture SEA cultural nuances. To fill this gap, we present SEA-VL, an open-source initiative dedicated to developing high-quality, culturally relevant data for SEA languages. By involving contributors from SEA countries, SEA-VL aims to ensure better cultural relevance and diversity, fostering greater inclusivity of underrepresented languages in VL research. Beyond crowdsourcing, our initiative goes one step further in the exploration of the automatic collection of culturally relevant images through crawling and image generation. First, we find that image crawling achieves approximately ~85% cultural relevance while being more cost- and time-efficient than crowdsourcing. Second, despite the substantial progress in generative vision models, synthetic images remain unreliable in accurately reflecting SEA cultures. The generated images often fail to reflect the nuanced traditions and cultural contexts of the region. Collectively, we gather 1.28M SEA culturally-relevant images, more than 50 times larger than other existing datasets. Through SEA-VL, we aim to bridge the representation gap in SEA, fostering the development of more inclusive AI systems that authentically represent diverse cultures across SEA.

In this paper, we introduce ConversaSynth, a framework designed to generate synthetic conversation audio using large language models (LLMs) with multiple persona settings. The framework first creates diverse and coherent text-based dialogues across various topics, which are then converted into audio using text-to-speech (TTS) systems. Our experiments demonstrate that ConversaSynth effectively generates highquality synthetic audio datasets, which can significantly enhance the training and evaluation of models for audio tagging, audio classification, and multi-speaker speech recognition. The results indicate that the synthetic datasets generated by ConversaSynth exhibit substantial diversity and realism, making them suitable for developing robust, adaptable audio-based AI systems.

Most super-resolution (SR) models struggle with real-world low-resolution (LR) images. This issue arises because the degradation characteristics in the synthetic datasets differ from those in real-world LR images. Since SR models are trained on pairs of high-resolution (HR) and LR images generated by downsampling, they are optimized for simple degradation. However, real-world LR images contain complex degradation caused by factors such as the imaging process and JPEG compression. Due to these differences in degradation characteristics, most SR models perform poorly on real-world LR images. This study proposes a dataset generation method using undertrained image reconstruction models. These models have the property of reconstructing low-quality images with diverse degradation from input images. By leveraging this property, this study generates LR images with diverse degradation from HR images to construct the datasets. Fine-tuning pre-trained SR models on our generated datasets improves noise removal and blur reduction, enhancing performance on real-world LR images. Furthermore, an analysis of the datasets reveals that degradation diversity contributes to performance improvements, whereas color differences between HR and LR images may degrade performance. 11 pages, (11 figures and 2 tables)

Diabetes is one of the most prevalent diseases worldwide, characterized by persistently high blood sugar levels, capable of damaging various internal organs and systems. Diabetes patients require routine check-ups, resulting in a time series of laboratory records, such as hemoglobin A1c, which reflects each patient's health behavior over time and informs their doctor's recommendations. Clustering patients into groups based on their entire time series data assists doctors in making recommendations and choosing treatments without the need to review all records. However, time series clustering of this type of dataset introduces some challenges; patients visit their doctors at different time points, making it difficult to capture and match trends, peaks, and patterns. Additionally, two aspects must be considered: differences in the levels of laboratory results and differences in trends and patterns. To address these challenges, we introduce a new clustering algorithm called Time and Trend Traveling Time Series Clustering (4TaStiC), using a base dissimilarity measure combined with Euclidean and Pearson correlation metrics. We evaluated this algorithm on artificial datasets, comparing its performance with that of seven existing methods. The results show that 4TaStiC outperformed the other methods on the targeted datasets. Finally, we applied 4TaStiC to cluster a cohort of 1,989 type 2 diabetes patients at Siriraj Hospital. Each group of patients exhibits clear characteristics that will benefit doctors in making efficient clinical decisions. Furthermore, the proposed algorithm can be applied to contexts outside the medical field.

Accurate theoretical results for interdigitated array of electrodes (IDAE) in semi-infinite cells can be found in the literature. However, these results are not always applicable when using finite cells. In this study, theoretical expressions for IDAE in a finite geometry cell are presented. At known current density, transient and steady state concentration profiles were obtained as well as the response time to a current step. Concerning the diffusion limited current, a lower bound was derived from the concentration profile and an upper bound was obtained from the limiting current of the semi-infinite case. The lower bound, which is valid when Kirchhoff's current law applies to the unit cell, can be useful to ensure a minimum current level during the design of the electrochemical cell. Finally, a criterion was developed defining when the behaviors of finite and semi-infinite cells are comparable. This allows to obtain higher current levels in finite cells, approaching that of the semi-infinite case. Examples with simulations were performed in order to illustrate and validate the theoretical results.

This study presents a deep learning system for breast cancer detection in mammography, developed using a modified EfficientNetV2 architecture with enhanced attention mechanisms. The model was trained on mammograms from a major Thai medical center and validated on three distinct datasets: an in-domain test set (9,421 cases), a biopsy-confirmed set (883 cases), and an out-of-domain generalizability set (761 cases) collected from two different hospitals. For cancer detection, the model achieved AUROCs of 0.89, 0.96, and 0.94 on the respective datasets. The system's lesion localization capability, evaluated using metrics including Lesion Localization Fraction (LLF) and Non-Lesion Localization Fraction (NLF), demonstrated robust performance in identifying suspicious regions. Clinical validation through concordance tests showed strong agreement with radiologists: 83.5% classification and 84.0% localization concordance for biopsy-confirmed cases, and 78.1% classification and 79.6% localization concordance for out-of-domain cases. Expert radiologists' acceptance rate also averaged 96.7% for biopsy-confirmed cases, and 89.3% for out-of-domain cases. The system achieved a System Usability Scale score of 74.17 for source hospital, and 69.20 for validation hospitals, indicating good clinical acceptance. These results demonstrate the model's effectiveness in assisting mammogram interpretation, with the potential to enhance breast cancer screening workflows in clinical practice.

We investigate how the initial and final exchange symmetries between the two-coin states influence the spatial entanglement dynamics between the two corresponding quantum walkers. Notably, when the initial state is anti-symmetric and the final measurement on the coins yields symmetric outcomes, all the initial entanglement will be transferred to the spatial degrees of freedom, regardless of when the coins are measured. Conversely, if the final outcomes are anti-symmetric, the spatial entanglement exhibits damped oscillation with a period ($T$) being inversely proportional to the coin operator parameter ($\theta$). These behaviours are reversed for symmetric initial states. Moreover, we also observe the same spatial entanglement damping regardless of the initial state when the post-selected results lack symmetry. Our findings reveal how symmetries affect the entanglement dynamics in quantum walks, offering potential insights for applications in quantum technology.

We present further developments on the Lagrangian 1-form description for one-dimensional integrable systems in both discrete and continuous levels. A key feature of integrability in this context called a closure relation will be derived from the local variation of the action on the space of independent variables. The generalised Euler-Lagrange equations and constraint equations are derived directly from the variation of the action on the space of dependent variables. This set of Lagrangian equations gives rise to a crucial property of integrable systems known as the multidimensional consistency. Alternatively, the closure relation can be obtained from generalised Stokes' theorem exhibiting a path independent property of the systems on the space of independent variables. The homotopy structure of paths suggests that the space of independent variables is simply connected. Furthermore, the N\"{o}ether charges, invariants in the context of Liouville integrability, can be obtained directly from the non-local variation of the action on the space of dependent variables.

Visual data can be understood at different levels of granularity, where global features correspond to semantic-level information and local features correspond to texture patterns. In this work, we propose a framework, called SPLIT, which allows us to disentangle local and global information into two separate sets of latent variables within the variational autoencoder (VAE) framework. Our framework adds generative assumption to the VAE by requiring a subset of the latent variables to generate an auxiliary set of observable data. This additional generative assumption primes the latent variables to local information and encourages the other latent variables to represent global information. We examine three different flavours of VAEs with different generative assumptions. We show that the framework can effectively disentangle local and global information within these models leads to improved representation, with better clustering and unsupervised object detection benchmarks. Finally, we establish connections between SPLIT and recent research in cognitive neuroscience regarding the disentanglement in human visual perception. The code for our experiments is at this https URL .

Customer temporal behavioral data was represented as images in order to perform churn prediction by leveraging deep learning architectures prominent in image classification. Supervised learning was performed on labeled data of over 6 million customers using deep convolutional neural networks, which achieved an AUC of 0.743 on the test dataset using no more than 12 temporal features for each customer. Unsupervised learning was conducted using autoencoders to better understand the reasons for customer churn. Images that maximally activate the hidden units of an autoencoder trained with churned customers reveal ample opportunities for action to be taken to prevent churn among strong data, no voice users.

Despite much promising research in the area of artificial intelligence for medical image diagnosis, there has been no large-scale validation study done in Thailand to confirm the accuracy and utility of such algorithms when applied to local datasets. Here we present a wide-reaching development and testing of a deep learning algorithm for automated thoracic disease detection, utilizing 421,859 local chest radiographs. Our study shows that convolutional neural networks can achieve remarkable performance in detecting 13 common abnormality conditions on chest X-ray, and the incorporation of local images into the training set is key to the model's success. This paper presents a state-of-the-art model for CXR abnormality detection, reaching an average AUROC of 0.91. This model, if integrated to the workflow, can result in up to 55.6% work reduction for medical practitioners in the CXR analysis process. Our work emphasizes the importance of investing in local research of medical diagnosis algorithms to ensure safe and efficient usage within the intended region.

Machine learning is gaining significant attention as a diagnostic tool in medical imaging, particularly in the analysis of retinal fundus images. However, this approach is not yet clinically applicable, as it still depends on human validation from a professional. Therefore, we present the design for a mobile application that monitors metrics related to retinal fundus images correlating to age-related conditions. The purpose of this platform is to observe for a change in these metrics over time, offering early insights into potential ocular diseases without explicitly delivering diagnostics. Metrics analysed include vessel tortuosity, as well as signs of glaucoma, retinopathy and macular edema. To evaluate retinopathy grade and risk of macular edema, a model was trained on the Messidor dataset and compared to a similar model trained on the MAPLES-DR dataset. Information from the DeepSeeNet glaucoma detection model, as well as tortuosity calculations, is additionally incorporated to ultimately present a retinal fundus image monitoring platform. As a result, the mobile application permits monitoring of trends or changes in ocular metrics correlated to age-related conditions with regularly uploaded photographs.

The behaviors of null geodesics in the spherical symmetric black holes in Einstein-Maxwell-dilaton (EMD) theory with coupling function $f(\Phi)=e^{-2\alpha \Phi}$ are meticulously analyzed. We investigate the effects of coupling constant $\alpha$ on the effective potential of photon trajectories within three ranges, namely 0<\alpha <1, $\alpha =1$ and \alpha >1. We find that the thicknesses of lensing and photon rings are smaller at larger $\alpha$ and fixed electric charge in the unit of mass $q$, whereas they are larger at fixed $\alpha$ and larger $q$. This behavior can be described by using the angular Lyapunov exponent $\gamma$ in the vicinity of the critical curve. Remarkably, the behaviors of photon trajectories are found to be more interesting when \alpha>1. Namely, the radius of the black hole shadow $R_\text{s}$ becomes to be smaller than the photon sphere radius $r_\text{ph}$ when \alpha > 1 and q>q^*. Moreover, $R_\text{s}$ goes to zero as $q$ saturates the extremal limit, beyond which the photon orbit becomes absent. Furthermore, we construct the optical appearance of black holes surrounded by optically and geometrically thin accretion disk with three cases of Gralla-Lupsasca-Marrone (GLM) emission profile. Our results indicate that the observed flux originating from the lensing and photon rings exhibits suppression as $\alpha$ increases, while it undergoes amplification with the increasing parameter $q$.

Selecting the appropriate number of clusters is a critical step in applying clustering algorithms. To assist in this process, various cluster validity indices (CVIs) have been developed. These indices are designed to identify the optimal number of clusters within a dataset. However, users may not always seek the absolute optimal number of clusters but rather a secondary option that better aligns with their specific applications. This realization has led us to introduce a Bayesian cluster validity index (BCVI), which builds upon existing indices. The BCVI utilizes either Dirichlet or generalized Dirichlet priors, resulting in the same posterior distribution. We evaluate our BCVI using the Wiroonsri index for hard clustering and the Wiroonsri-Preedasawakul index for soft clustering as underlying indices. We compare the performance of our proposed BCVI with that of the original underlying indices and several other existing CVIs, including Davies-Bouldin, Starczewski, Xie-Beni, and KWON2 indices. Our BCVI offers clear advantages in situations where user expertise is valuable, allowing users to specify their desired range for the final number of clusters. To illustrate this, we conduct experiments classified into three different scenarios. Additionally, we showcase the practical applicability of our approach through real-world datasets, such as MRI brain tumor images. These tools will be published as a new R package 'BayesCVI'.

We propose an algorithm for calculating the cardiothoracic ratio (CTR) from chest X-ray films. Our approach applies a deep learning model based on U-Net with VGG16 encoder to extract lung and heart masks from chest X-ray images and calculate CTR from the extents of obtained masks. Human radiologists evaluated our CTR measurements, and $76.5\%$ were accepted to be included in medical reports without any need for adjustment. This result translates to a large amount of time and labor saved for radiologists using our automated tools.

This research introduces a novel dual-attention transformer architecture for predicting soil electrical resistivity, a critical parameter for high-voltage substation construction. Our model employs attention mechanisms operating across both features and data batches, enhanced by feature embedding layers that project inputs into higher-dimensional spaces. We implements Particle Swarm Optimization for hyperparameter tuning, systematically optimizing embedding dimensions, attention heads, and neural network architecture. The proposed architecture achieves superior predictive performance (Mean Absolute Percentage Error: 0.63%) compared to recent state of the art models for tabular data. Crucially, our model maintains explainability through SHapley Additive exPlanations value analysis, revealing that fine particle content and dry density are the most influential parameters affecting soil resistivity. We developes a web-based application implementing this model to provide engineers with an accessible decision support framework that bridges geotechnical and electrical engineering requirements for the Electricity Generating Authority of Thailand. This integrated approach satisfies both structural stability and electrical safety standards, improving construction efficiency and safety compliance in high-voltage infrastructure implementation.

We consider Borwein-Preiss and Ekeland variational principles using distance functions that neither is symmetric nor enjoy the triangular inequality. All the given results rely exclusively on the convergence and continuity behaviors induced synthetically by the distance function itself without any topological implications. At the end of the paper, we also present two applications; the Caristi fixed point theorem and an existence theorem for equilibrium problems.

Recent low-redshift observations give value of the present-time Hubble parameter $H_{0}\simeq 74~\rm{km s}^{-1}\rm{Mpc}^{-1}$, roughly 10\% higher than the predicted value $H_{0}=67.4~\rm{km s}^{-1}\rm{Mpc}^{-1}$ from Planck's observations of the Cosmic Microwave Background radiation~(CMB) and the $\Lambda$CDM model. Phenomenologically, we show that by adding an extra component X with negative density in the Friedmann equation, it can relieve the Hubble tension without changing the Planck's constraint on the matter and dark energy densities. For the extra negative density to be sufficiently small, its equation-of-state parameter must satisfy $1/3\leq w_{X}\leq1$. We propose a quintom model of two scalar fields that realizes this condition and potentially alleviate the Hubble tension. One scalar field acts as a quintessence while another "phantom" scalar conformally couples to matter in such a way that viable cosmological scenario can be achieved. The model depends only on two parameters, $\lambda_{\phi}$ and $\delta$ which represent rolling tendency of the self-interacting potential of the quintessence and the strength of conformal phantom-matter coupling respectively. The toy quintom model with $H_{0}=73.4~\rm{km s}^{-1}\rm{Mpc}^{-1}$~(Quintom I) gives good Supernovae-Ia luminosity fits, decent $r_{\rm BAO}$ fit, but slightly small acoustic multipole $\ell_{A}=285.54$. Full parameter scan reveals that quintom model provide better model than the $\Lambda$CDM model in certain region of the parameter space, 0.02<\delta<0.10, \Omega_{m}^{(0)}<0.31, while significantly relieving Hubble tension even though not completely resolving it. A benchmark quintom model, Quintom II, is presented as an example.