{"id":1641,"date":"2025-08-22T06:49:43","date_gmt":"2025-08-22T06:49:43","guid":{"rendered":"https:\/\/ccds.ai\/?p=1641"},"modified":"2025-12-09T12:22:06","modified_gmt":"2025-12-09T12:22:06","slug":"two-of-our-research-papers-have-been-accepted-at-emnlp-2025","status":"publish","type":"post","link":"https:\/\/ccds.ai\/?p=1641","title":{"rendered":"Two of our research papers have been accepted at EMNLP 2025"},"content":{"rendered":"
\n
\n
\"\ud83d\udea8\"<\/span> Exciting News! A great achievement for CCDS!!!!<\/div>\n<\/div>\n
\n
We are thrilled to announce that two of our research papers \u2014 BanMime and DM-Codec \u2014 have been accepted at EMNLP 2025, one of the most prestigious conferences in NLP (Core Rank A*).<\/div>\n
\"\ud83d\udccc\"<\/span> BanMime has been accepted into the EMNLP 2025 Main Conference \"\ud83c\udf89\"<\/span><\/div>\n
\"\ud83d\udccc\"<\/span> DM-Codec has been accepted into the EMNLP 2025 Findings \"\ud83c\udf89\"<\/span><\/div>\n<\/div>\n
\n
\"\ud83d\uddbc\ufe0f\"<\/span> BanMime: A Novel Bangla Meme Dataset for Visual Metaphor Explanation<\/div>\n
In this work, we introduce BanMime, the first-ever dataset for Bangla memes enriched with human-annotated explanations for metaphors.<\/div>\n
\"\ud83e\udde9\"<\/span> We curated a dataset of 2,000 misogynistic memes in Bangla. Each meme was labeled, and based on those labels, explanations were annotated to uncover the underlying metaphor.<\/div>\n
\"\ud83d\udd0d\"<\/span> Our experiments reveal that existing open-source Vision-Language Models (VLMs) struggle to comprehend visual metaphors, and their performance is even worse when generating explanations.<\/div>\n
This work is a joint collaboration between CUET, DIU, and CCDS.<\/div>\n<\/div>\n
\n
\"\u270d\ufe0f\"<\/span> Authors: Md Ayon Mia*, Akm Moshiur Rahman Mazumder*, Khadiza Sultana Sayma, Md Fahim, Md Tahmid Hasan Fuad, MUHAMMAD IBRAHIM KHAN, AKM MAHBUBUR RAHMAN<\/div>\n
(*Equal Contribution)<\/div>\n<\/div>\n
\n
\"\ud83d\uddbc\ufe0f\"<\/span> DM-Codec: Distilling Multimodal Representations for Speech Tokenization<\/div>\n<\/div>\n
\n
DM-Codec is a novel speech tokenizer that integrates multimodal (acoustic, semantic, and contextual) representations via a language model (LM) and self-supervised speech model (SM) through distillation.<\/div>\n
\"\ud83d\udd0d\"<\/span> DM-Codec outperforms state-of-the-art models, reducing Word Error Rate (WER) by up to 13.46% and enhancing speech intelligibility and quality on the LibriSpeech dataset.<\/div>\n
This work is done in collaboration with AWS (GenAI), Qatar Computing Research Institute, and University of Virginia<\/div>\n<\/div>\n
\n
\"\u270d\ufe0f\"<\/span> Authors: Md Mubtasim Ahasan, Md Fahim, Tasnim Mohiuddin, A K M Mahbubur Rahman, Aman Chadha, Tariq Iqbal, M Ashraful Amin, Md Mofijul Islam, Amin Ahsan Ali<\/div>\n<\/div>\n
\n
Congratulations to the all co-authors, collaborators and supervisors \"\ud83c\udf8a\"<\/span>\"\ud83c\udf8a\"<\/span><\/div>\n<\/div>\n<\/div><\/section><\/div>