Kitsuchart Pasupa

Sarmistha Das, Vaibhav Vishal, Shreyas Guha et al.

In the contemporary epoch of multilingual education, learning idioms provides a fascinating gateway towards creativity, cultural values, historical context, and diverse perspectives inherent to various linguistic traditions. This paper showcases the navigation of retaining figurative and cultural semantics in low-resource Southeast Asian languages such as Hindi, Bengali, and Thai, where culturally rich idioms pose significant obstacles for computational modeling and cross-linguistic transfer due to their deep metaphorical complexity. To tackle such complexity, we present Varnika, a reconstructed multimodal idiom corpus comprising 3,533 multilingual idioms, enriched with seven idiomatic tones aligned with both textual and visual representations. Additionally, to infer informative idiomatic understanding, we introduce a Hybrid Mixture-of-Experts (HybridMoE) framework that embeds multiple idiomatic expert opinions while mitigating expert sparsity by integrating outputs from both selected and unselected experts through controlled hybridization, further augmented with Idiomatic Property Signals via masked multimodal embeddings. To analyze the performance across multiple dimensions, we propose the IDIO-TONE and Idiomatic Validation Score, a three-stage evaluation pipeline measuring (i) literal translation fidelity, (ii) visual-semantic alignment, and (iii) idiomatic meaning retention. Empirical evaluations highlight that HybridMoE achieves 5--6\% performance gains across advanced vision language models, demonstrating improved representation of figurative language and culturally embedded meaning in multilingual multimodal settings

Sarmistha Das, Shreyas Guha, Suvrayan Bandyopadhyay et al.

Idiomatic reasoning, deeply intertwined with metaphor and culture, remains a blind spot for contemporary language models, whose progress skews toward surface-level lexical and semantic cues. For instance, the Bengali idiom \textit{\foreignlanguage{bengali}{\char"0986\char"0999\char"09CD\char"0997\char"09C1 \char"09B0 \char"09AB\char"09B2 \char"099F\char"0995}} (angur fol tok, ``grapes are sour''): it encodes denial-driven rationalization, yet naive models latch onto the literal fox-and-grape imagery. Addressing this oversight, we present ``Mediom,'' a multilingual, multimodal idiom corpus of 3,533 Hindi, Bengali, and Thai idioms, each paired with gold-standard explanations, cross-lingual translations, and carefully aligned text--image representations. We benchmark both large language models (textual reasoning) and vision-language models (figurative disambiguation) on Mediom, exposing systematic failures in metaphor comprehension. To mitigate these gaps, we propose ``HIDE,'' a Hinting-based Idiom Explanation framework that leverages error-feedback retrieval and targeted diagnostic cues for iterative reasoning refinement. Collectively, Mediom and HIDE establish a rigorous test bed and methodology for culturally grounded, multimodal idiom understanding embedded with reasoning hints in next-generation AI systems.

Arkadeep Acharya, Akash Ghosh, Pradeepika Verma et al.

With the increasing use of RetrievalAugmented Generation (RAG), strong retrieval models have become more important than ever. In healthcare, multimodal retrieval models that combine information from both text and images offer major advantages for many downstream tasks such as question answering, cross-modal retrieval, and multimodal summarization, since medical data often includes both formats. However, there is currently no standard benchmark to evaluate how well these models perform in medical settings. To address this gap, we introduce M3Retrieve, a Multimodal Medical Retrieval Benchmark. M3Retrieve, spans 5 domains,16 medical fields, and 4 distinct tasks, with over 1.2 Million text documents and 164K multimodal queries, all collected under approved licenses. We evaluate leading multimodal retrieval models on this benchmark to explore the challenges specific to different medical specialities and to understand their impact on retrieval performance. By releasing M3Retrieve, we aim to enable systematic evaluation, foster model innovation, and accelerate research toward building more capable and reliable multimodal retrieval systems for medical applications. The dataset and the baselines code are available in this github page https://github.com/AkashGhosh/M3Retrieve.

Sarmistha Das, Priya Mathur, Ishani Sharma et al.

The exponential technological breakthrough of the FinTech industry has significantly enhanced user engagement through sophisticated advisory chatbots. However, large-scale fine-tuning of LLMs can occasionally yield unprofessional or flippant remarks, such as ``With that money, you're going to change the world,'' which, though factually correct, can be contextually inappropriate and erode user trust. The scarcity of domain-specific datasets has led previous studies to focus on isolated components, such as reasoning-aware frameworks or the enhancement of human-like response generation. To address this research gap, we present Fin-Solution 2.O, an advanced solution that 1) introduces the multi-turn financial conversational dataset, Fin-Vault, and 2) incorporates a unified model, Fin-Ally, which integrates commonsense reasoning, politeness, and human-like conversational dynamics. Fin-Ally is powered by COMET-BART-embedded commonsense context and optimized with a Direct Preference Optimization (DPO) mechanism to generate human-aligned responses. The novel Fin-Vault dataset, consisting of 1,417 annotated multi-turn dialogues, enables Fin-Ally to extend beyond basic account management to provide personalized budgeting, real-time expense tracking, and automated financial planning. Our comprehensive results demonstrate that incorporating commonsense context enables language models to generate more refined, textually precise, and professionally grounded financial guidance, positioning this approach as a next-generation AI solution for the FinTech sector. Dataset and codes are available at: https://github.com/sarmistha-D/Fin-Ally

Punit Kumar Singh, Nishant Kumar, Akash Ghosh et al.

Language Models (LMs) are primarily evaluated on globally popular sports, often overlooking regional and indigenous sporting traditions. To address this gap, we introduce \textbf{\textit{CultSportQA}}, a benchmark designed to assess LMs' understanding of traditional sports across 60 countries and 6 continents, encompassing four distinct cultural categories. The dataset features 33,000 multiple-choice questions (MCQs) across text and image modalities, each of which is categorized into three key types: history-based, rule-based, and scenario-based. To evaluate model performance, we employ zero-shot, few-shot, and chain-of-thought (CoT) prompting across a diverse set of Large Language Models (LLMs), Small Language Models (SLMs), and Multimodal Large Language Models (MLMs). By providing a comprehensive multilingual and multicultural sports benchmark, \textbf{\textit{CultSportQA}} establishes a new standard for assessing AI's ability to understand and reason about traditional sports.

Wasu Kudisthalert, Kitsuchart Pasupa, Aythami Morales et al.

Extreme Learning Machine is a powerful classification method very competitive existing classification methods. It is extremely fast at training. Nevertheless, it cannot perform face verification tasks properly because face verification tasks require comparison of facial images of two individuals at the same time and decide whether the two faces identify the same person. The structure of Extreme Leaning Machine was not designed to feed two input data streams simultaneously, thus, in 2-input scenarios Extreme Learning Machine methods are normally applied using concatenated inputs. However, this setup consumes two times more computational resources and it is not optimized for recognition tasks where learning a separable distance metric is critical. For these reasons, we propose and develop a Siamese Extreme Learning Machine (SELM). SELM was designed to be fed with two data streams in parallel simultaneously. It utilizes a dual-stream Siamese condition in the extra Siamese layer to transform the data before passing it along to the hidden layer. Moreover, we propose a Gender-Ethnicity-Dependent triplet feature exclusively trained on a variety of specific demographic groups. This feature enables learning and extracting of useful facial features of each group. Experiments were conducted to evaluate and compare the performances of SELM, Extreme Learning Machine, and DCNN. The experimental results showed that the proposed feature was able to perform correct classification at 97.87% accuracy and 99.45% AUC. They also showed that using SELM in conjunction with the proposed feature provided 98.31% accuracy and 99.72% AUC. They outperformed the well-known DCNN and Extreme Leaning Machine methods by a wide margin.

Kitsuchart Pasupa, Supawit Vatathanavaro, Suchat Tungjitnob

Morphologies of red blood cells are normally interpreted by a pathologist. It is time-consuming and laborious. Furthermore, a misclassified red blood cell morphology will lead to false disease diagnosis and improper treatment. Thus, a decent pathologist must truly be an expert in classifying red blood cell morphology. In the past decade, many approaches have been proposed for classifying human red blood cell morphology. However, those approaches have not addressed the class imbalance problem in classification. A class imbalance problem---a problem where the numbers of samples in classes are very different---is one of the problems that can lead to a biased model towards the majority class. Due to the rarity of every type of abnormal blood cell morphology, the data from the collection process are usually imbalanced. In this study, we aimed to solve this problem specifically for classification of dog red blood cell morphology by using a Convolutional Neural Network (CNN)---a well-known deep learning technique---in conjunction with a focal loss function, adept at handling class imbalance problem. The proposed technique was conducted on a well-designed framework: two different CNNs were used to verify the effectiveness of the focal loss function and the optimal hyper-parameters were determined by 5-fold cross-validation. The experimental results show that both CNNs models augmented with the focal loss function achieved higher $F_{1}$-scores, compared to the models augmented with a conventional cross-entropy loss function that does not address class imbalance problem. In other words, the focal loss function truly enabled the CNNs models to be less biased towards the majority class than the cross-entropy did in the classification task of imbalanced dog red blood cell data.

Wanthanee Rathasamuth, Kitsuchart Pasupa, Sissades Tongsima

A panel of large number of common Single Nucleotide Polymorphisms (SNPs) distributed across an entire porcine genome has been widely used to represent genetic variability of pig. With the advent of SNP-array technology, a genome-wide genetic profile of a specimen can be easily observed. Among the large number of such variations, there exist a much smaller subset of the SNP panel that could equally be used to correctly identify the corresponding breed. This work presents a SNP selection heuristic that can still be used effectively in the breed classification process. The proposed feature selection was done by the approach of combining a filter method and a wrapper method--information gain method and genetic algorithm--plus a feature frequency selection step, while classification was done by support vector machine. The approach was able to reduce the number of significant SNPs to 0.86 % of the total number of SNPs in a swine dataset and provided a high classification accuracy of 94.80 %.

Zakria Hussain, Arto Klami, Jussi Kujala et al.

This paper describes PinView, a content-based image retrieval system that exploits implicit relevance feedback collected during a search session. PinView contains several novel methods to infer the intent of the user. From relevance feedback, such as eye movements or pointer clicks, and visual features of images, PinView learns a similarity metric between images which depends on the current interests of the user. It then retrieves images with a specialized online learning algorithm that balances the tradeoff between exploring new images and exploiting the already inferred interests of the user. We have integrated PinView to the content-based image retrieval system PicSOM, which enables applying PinView to real-world image databases. With the new algorithms PinView outperforms the original PicSOM, and in online experiments with real users the combination of implicit and explicit feedback gives the best results.