technical reportAIAC Technical Report Series · 2026-02-15

AIAC-TR-2026-01: Baseline Evaluation of Tokenization and Morphological Parsing for Kannada Corpora

Comprehensive empirical analysis comparing byte-pair encoding (BPE) vs sentencepiece subword segmentation on low-resource regional text with vocabulary boundary metrics. Evaluates agglutinative morphology preservation across 10,000 sentences from regional public administrative records.

Authors: Darshan K., Aditi S., Bhavana V.CC BY 4.0 / Apache-2.0
preprintDraft Under Internal Faculty Review (Targeting Q2 2026) · 2026-04-01

Cross-Lingual Pseudo-Labeling with Calibrated Confidence Bounds for Low-Resource Indic Sentiment

Investigation into teacher-student knowledge distillation when transferring English sentiment priors into Kannada regional dialects with temperature calibration. Outlines empirical bounds where pseudo-label confidence filters prevent negative transfer in low-resource target domains.

Authors: Darshan K., Bhavana V., et al.Preprint / CC BY-NC 4.0