ISSAI researchers took part in the 9th International Natural Language Processing and Information Retrieval Conference (NLPIR 2025), held on December 12–14 at Kyushu University’s Research and International Exchange Plaza in Fukuoka, Japan.
During the conference, Anuar Aryngazin, Data Scientist at ISSAI, presented the paper “Multi-Head Vision Transformer for Multilingual Font and Language Identification”, co-authored with Professor Huseyin Atakan Varol, Founding Director of ISSAI. The work explores advanced vision-based approaches for multilingual font and language identification, highlighting the potential of transformer architectures in complex, real-world settings.
In addition, Professor Huseyin Atakan Varol presented the paper “Efficient Language Expansion for Machine Translation Using LoRA: A Case Study for Kazakh”, co-authored with Saida Mussakhojayeva. The study demonstrates how parameter-efficient fine-tuning techniques can accelerate the inclusion of low-resource languages—such as Kazakh—into modern machine translation systems.
Both studies were selected for oral presentations, reflecting their strong research contributions and relevance to the NLP community. The papers will be published by Springer Nature following the conference.
Participation in NLPIR 2025 underscores ISSAI’s ongoing commitment to advancing multilingual AI research and contributing to the global scientific dialogue—while ensuring that underrepresented languages are meaningfully supported in next-generation language technologies.