Selected work

Representative Papers

From Phonemes to Meaning: Evaluating Large Language Models on Tamil

Nov 15, 2025

Low-resource, morphologically rich languages like Tamil lack native linguistic evaluation benchmarks, hindering reliable assessment of large language models (LLMs). Method: We introduce ILAKKANAM—the first linguistically grounded, culturally authentic Tamil evaluation benchmark—constructed from real Sri Lankan K–12 examination items. It covers five linguistic dimensions (morphology, syntax, semantics, pragmatics, and factual knowledge) via 820 expert-annotated, native-language questions organized within a grade-based difficulty framework to avoid cultural and linguistic distortions from English translation. Contribution/Results: Our systematic evaluation of leading closed- and open-weight LLMs reveals that Gemini 2.5 achieves highest accuracy; open-source models consistently underperform. Accuracy declines markedly with increasing grade level (i.e., rising linguistic complexity), and improvements in linguistic competence show no strong correlation with language identification capability. ILAKKANAM establishes a reproducible, culturally grounded paradigm for evaluating LLMs in low-resource languages.

0 citationsRead paper

Semi-Supervised Learning with Online Knowledge Distillation for Skin Lesion Classification

Aug 15, 2025

To address the performance bottleneck of deep learning in skin lesion classification caused by scarce labeled data, this paper proposes a semi-supervised online knowledge distillation framework. The method integrates ensemble learning with online knowledge distillation, enabling multiple convolutional neural networks to collaboratively train using only a small number of labeled samples and abundant unlabeled data via real-time mutual distillation—without requiring additional manual annotations. Its core innovation lies in a self-augmented bidirectional knowledge transfer mechanism, allowing a single student model to approximate ensemble-level inference performance. On the ISIC 2018 and 2019 benchmarks, the single-model accuracy significantly surpasses that of independently trained baselines and approaches the performance of fully supervised ensembles, substantially reducing dependency on labeled data and deployment resource overhead.

0 citationsRead paper

Recent publications

Latest Papers

From Phonemes to Meaning: Evaluating Large Language Models on Tamil

Nov 15, 2025

Low-resource, morphologically rich languages like Tamil lack native linguistic evaluation benchmarks, hindering reliable assessment of large language models (LLMs). Method: We introduce ILAKKANAM—the first linguistically grounded, culturally authentic Tamil evaluation benchmark—constructed from real Sri Lankan K–12 examination items. It covers five linguistic dimensions (morphology, syntax, semantics, pragmatics, and factual knowledge) via 820 expert-annotated, native-language questions organized within a grade-based difficulty framework to avoid cultural and linguistic distortions from English translation. Contribution/Results: Our systematic evaluation of leading closed- and open-weight LLMs reveals that Gemini 2.5 achieves highest accuracy; open-source models consistently underperform. Accuracy declines markedly with increasing grade level (i.e., rising linguistic complexity), and improvements in linguistic competence show no strong correlation with language identification capability. ILAKKANAM establishes a reproducible, culturally grounded paradigm for evaluating LLMs in low-resource languages.

0 citationsRead paper

Semi-Supervised Learning with Online Knowledge Distillation for Skin Lesion Classification

Aug 15, 2025

To address the performance bottleneck of deep learning in skin lesion classification caused by scarce labeled data, this paper proposes a semi-supervised online knowledge distillation framework. The method integrates ensemble learning with online knowledge distillation, enabling multiple convolutional neural networks to collaboratively train using only a small number of labeled samples and abundant unlabeled data via real-time mutual distillation—without requiring additional manual annotations. Its core innovation lies in a self-augmented bidirectional knowledge transfer mechanism, allowing a single student model to approximate ensemble-level inference performance. On the ISIC 2018 and 2019 benchmarks, the single-model accuracy significantly surpasses that of independently trained baselines and approaches the performance of fully supervised ensembles, substantially reducing dependency on labeled data and deployment resource overhead.

0 citationsRead paper