Center for Computational & Data Sciences, IUB

13/08/2026

🎉✨ 𝐀 𝐌𝐢𝐥𝐞𝐬𝐭𝐨𝐧𝐞 𝐌𝐨𝐦𝐞𝐧𝐭 𝐚𝐭 𝐂𝐂𝐃𝐒! ✨🎉

We are thrilled to announce the successful defense of the 8th MSc in Computer Science Thesis at CCDS — and what a defense it was!

Heartiest congratulations to 𝐉𝐚𝐡𝐚𝐧𝐠𝐠𝐢𝐫 𝐇𝐨𝐬𝐬𝐚𝐢𝐧 𝐒𝐞𝐭𝐮 🎓 on brilliantly defending his thesis titled:

📌 "𝘋𝘦𝘤𝘭𝘢𝘳𝘦𝘥 𝘍𝘢𝘪𝘭𝘶𝘳𝘦 𝘰𝘷𝘦𝘳 𝘚𝘪𝘭𝘦𝘯𝘵 𝘍𝘢𝘪𝘭𝘶𝘳𝘦: 𝘈𝘣𝘴𝘵𝘦𝘯𝘵𝘪𝘰𝘯, 𝘌𝘹𝘱𝘭𝘢𝘪𝘯𝘢𝘣𝘪𝘭𝘪𝘵𝘺 𝘢𝘯𝘥 𝘢𝘯 𝘈𝘣𝘴𝘵𝘦𝘯𝘵𝘪𝘰𝘯-𝘈𝘸𝘢𝘳𝘦 𝘌𝘷𝘢𝘭𝘶𝘢𝘵𝘪𝘰𝘯 𝘍𝘳𝘢𝘮𝘦𝘸𝘰𝘳𝘬"

A groundbreaking work that pushes the boundaries of AI reliability and trustworthiness — asking not just what a model predicts, but when it should dare to say "𝘐 𝘥𝘰𝘯'𝘵 𝘬𝘯𝘰𝘸." 💡

This achievement stands on the shoulders of exceptional mentorship and rigorous examination:

👨‍🏫 𝐒𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐨𝐫: 𝐃𝐫. 𝐀𝐬𝐡𝐫𝐚𝐟𝐮𝐥 𝐈𝐬𝐥𝐚𝐦 — 𝘋𝘪𝘳𝘦𝘤𝘵𝘰𝘳, 𝘊𝘊𝘋𝘚
🌍 𝐂𝐨-𝐒𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐨𝐫𝐬:
▸ 𝐏𝐫𝐨𝐟. 𝐃𝐫. 𝐌. 𝐀𝐬𝐡𝐫𝐚𝐟𝐮𝐥 𝐀𝐦𝐢𝐧 — 𝘎𝘦𝘰𝘳𝘨𝘦 𝘔𝘢𝘴𝘰𝘯 𝘜𝘯𝘪𝘷𝘦𝘳𝘴𝘪𝘵𝘺, 𝘜𝘚𝘈 & 𝘍𝘰𝘶𝘯𝘥𝘦𝘳 𝘋𝘪𝘳𝘦𝘤𝘵𝘰𝘳, 𝘊𝘊𝘋𝘚
▸ 𝐃𝐫. 𝐑𝐚𝐬𝐡𝐞𝐝𝐮𝐫 𝐑𝐚𝐡𝐦𝐚𝐧 — 𝘊𝘰-𝘋𝘪𝘳𝘦𝘤𝘵𝘰𝘳, 𝘕𝘌𝘚𝘛 𝘞𝘪𝘯𝘨, 𝘊𝘊𝘋𝘚

🔎 𝐄𝐱𝐭𝐞𝐫𝐧𝐚𝐥 𝐄𝐱𝐚𝐦𝐢𝐧𝐞𝐫: 𝐏𝐫𝐨𝐟. 𝐃𝐫. 𝐌𝐨𝐢𝐧𝐮𝐥 𝐙𝐚𝐛𝐞𝐫 — 𝘜𝘯𝘪𝘷𝘦𝘳𝘴𝘪𝘵𝘺 𝘰𝘧 𝘊𝘢𝘮𝘣𝘳𝘪𝘥𝘨𝘦, 𝘜𝘒 & 𝘐𝘯𝘴𝘵𝘪𝘵𝘶𝘵𝘦 𝘰𝘧 𝘋𝘦𝘷𝘦𝘭𝘰𝘱𝘮𝘦𝘯𝘵 𝘚𝘵𝘶𝘥𝘪𝘦𝘴, 𝘜𝘒

🔎 𝐈𝐧𝐭𝐞𝐫𝐧𝐚𝐥 𝐄𝐱𝐚𝐦𝐢𝐧𝐞𝐫: 𝐃𝐫. 𝐀𝐊𝐌 𝐌𝐚𝐡𝐛𝐮𝐛𝐮𝐫 𝐑𝐚𝐡𝐦𝐚𝐧 — 𝘋𝘪𝘳𝘦𝘤𝘵𝘰𝘳, 𝘋𝘢𝘵𝘢 𝘚𝘤𝘪𝘦𝘯𝘤𝘦 𝘞𝘪𝘯𝘨, 𝘊𝘊𝘋𝘚 & 𝘏𝘦𝘢𝘥 𝘰𝘧 𝘊𝘚𝘌 𝘋𝘦𝘱𝘢𝘳𝘵𝘮𝘦𝘯𝘵, 𝘐𝘜𝘉

We are immensely proud of 𝐒𝐞𝐭𝐮's dedication, intellectual curiosity, and the quality of scholarship he has brought to CCDS. This is not just his success — it is a celebration of research, mentorship, and the spirit of 𝐂𝐂𝐃𝐒! 🚀

Onward and upward, 𝐒𝐞𝐭𝐮! The best is yet to come. 🌟

11/08/2026

We are delighted to congratulate Md. Fahim, Research Assistant (RA) / Junior Research Scientist (JRS) at CCDS, IUB, on his acceptance into a fully funded PhD program at the University of Texas at Dallas for the Fall 2026 intake. 🎓

His doctoral research will focus on:
•🤖 Large Language Model (LLM) Reasoning
•🎯 LLM Alignment
•🖼️ Multimodal AI

Since the inception of CCDS, Fahim has been one of the key contributors to our research center. He has played a pivotal role in laying the foundation of CCDS through many of our publications in leading A*/A-ranked conferences and Q1 journals, helping establish our research presence on the global stage. 📚

Beyond his research accomplishments, Fahim has been an exceptional mentor to junior researchers, generously sharing his knowledge, guiding research projects, and fostering a collaborative environment that continues to strengthen our community. His dedication to nurturing young researchers has had a lasting impact on CCDS. 🌱

He has also served as an instructor for IUB's "Coding for All" initiative, inspiring students and making programming more accessible to aspiring learners. 💻

While we will certainly miss his presence at CCDS, we are immensely proud to see him embark on this exciting new chapter. We are confident that he will continue making meaningful contributions to the field of Artificial Intelligence while representing both CCDS and IUB with distinction on the global stage. 🌍

Congratulations once again, Fahim! We wish you every success in your PhD journey at the University of Texas at Dallas. We look forward to seeing the impact of your future research and accomplishments. 🚀

26/07/2026

CCDS Weekly Paper Presentation
On Monday, 6 July 2026, the Center for Computational & Data Sciences (CCDS), Independent University, Bangladesh (IUB), hosted a paper presentation session on steering and controlling large language model behavior.
Presenter: Sumaiya Karim Katha
Research Assistant, CCDS, IUB
Papers Presented:
1. Controlling Large Language Models Through Concept Activation Vectors
2. LayerNavigator: Finding Promising Intervention Layers for Efficient Activation Steering in Large Language Models
The session featured an engaging discussion on activation-based control of LLMs, covering the GCAV framework for steering model outputs on concepts like toxicity and style without fine-tuning, and LayerNavigator, a method for efficiently identifying which layers to intervene on for multi-layer activation steering. Discussion touched on the tradeoffs of concept-vector-based control, the combinatorial challenge of layer selection, and how principled, low-overhead layer scoring can make multi-layer steering practical.
We thank all participants for their valuable contributions and insightful discussions.

Center for Computational & Data Sciences (CCDS). Exploring nature and society through data and computation.

25/07/2026

🎉 𝖶𝖾 𝖺𝗋𝖾 𝖾𝗑𝗍𝗋𝖺𝗈𝗋𝖽𝗂𝗇𝖺𝗋𝗂𝗅𝗒 𝗉𝗋𝗈𝗎𝖽 𝗍𝗈 𝖺𝗇𝗇𝗈𝗎𝗇𝖼𝖾 𝗍𝗁𝖺𝗍 𝐂𝐂𝐃𝐒'𝗌 𝖼𝗎𝗋𝗋𝖾𝗇𝗍 𝖣𝗂𝗋𝖾𝖼𝗍𝗈𝗋 𝐃𝐫. 𝐀𝐬𝐡𝐫𝐚𝐟𝐮𝐥 𝐈𝐬𝐥𝐚𝐦 𝖺𝗇𝖽 𝖥𝗈𝗎𝗇𝖽𝖾𝗋 𝖣𝗂𝗋𝖾𝖼𝗍𝗈𝗋 𝐏𝐫𝐨𝐟. 𝐌. 𝐀𝐬𝐡𝐫𝐚𝐟𝐮𝐥 𝐀𝐦𝐢𝐧 𝗁𝖺𝗏𝖾 𝖻𝖾𝖾𝗇 𝗁𝗈𝗇𝗈𝗋𝖾𝖽 𝗐𝗂𝗍𝗁 𝗋𝖾𝖼𝗈𝗀𝗇𝗂𝗍𝗂𝗈𝗇 𝖺𝗆𝗈𝗇𝗀 𝗍𝗁𝖾 𝐖𝐨𝐫𝐥𝐝'𝐬 𝐓𝐨𝐩 𝟓% 𝐒𝐜𝐢𝐞𝐧𝐭𝐢𝐬𝐭𝐬 𝖿𝗈𝗋 𝟤𝟢𝟤𝟧. 😊✨🌍

𝖳𝗁𝗂𝗌 𝗋𝖾𝖼𝗈𝗀𝗇𝗂𝗍𝗂𝗈𝗇 𝗇𝗈𝗍 𝗈𝗇𝗅𝗒 𝖼𝖾𝗅𝖾𝖻𝗋𝖺𝗍𝖾𝗌 𝗍𝗁𝖾 𝗂𝗇𝖽𝗂𝗏𝗂𝖽𝗎𝖺𝗅 𝖺𝖼𝖼𝗈𝗆𝗉𝗅𝗂𝗌𝗁𝗆𝖾𝗇𝗍𝗌 𝗈𝖿 𝖣𝗋. 𝖠𝗌𝗁𝗋𝖺𝖿𝗎𝗅 𝖨𝗌𝗅𝖺𝗆 𝖺𝗇𝖽 𝖯𝗋𝗈𝖿. 𝖬. 𝖠𝗌𝗁𝗋𝖺𝖿𝗎𝗅 𝖠𝗆𝗂𝗇 𝖻𝗎𝗍 𝖺𝗅𝗌𝗈 𝗎𝗇𝖽𝖾𝗋𝗌𝖼𝗈𝗋𝖾𝗌 𝗍𝗁𝖾 𝖾𝗑𝖼𝖾𝗅𝗅𝖾𝗇𝖼𝖾 𝖺𝗇𝖽 𝗏𝗂𝗌𝗂𝗈𝗇 𝗍𝗁𝖺𝗍 𝖽𝖾𝖿𝗂𝗇𝖾 𝖢𝖢𝖣𝖲 𝖺𝗌 𝖺 center of research excellence. 𝖶𝖾 𝖺𝗋𝖾 𝖽𝖾𝖾𝗉𝗅𝗒 𝗁𝗈𝗇𝗈𝗋𝖾𝖽 𝗍𝗈 𝗁𝖺𝗏𝖾 𝗌𝗎𝖼𝗁 𝗏𝗂𝗌𝗂𝗈𝗇𝖺𝗋𝗒 𝗅𝖾𝖺𝖽𝖾𝗋𝗌 𝗀𝗎𝗂𝖽𝗂𝗇𝗀 𝗈𝗎𝗋 𝗈𝗋𝗀𝖺𝗇𝗂𝗓𝖺𝗍𝗂𝗈𝗇 𝗍𝗈𝗐𝖺𝗋𝖽 𝖼𝗈𝗇𝗍𝗂𝗇𝗎𝖾𝖽 𝗌𝗎𝖼𝖼𝖾𝗌𝗌 𝖺𝗇𝖽 𝗂𝗆𝗉𝖺𝖼𝗍.

𝐇𝐞𝐚𝐫𝐭𝐢𝐞𝐬𝐭 𝐜𝐨𝐧𝐠𝐫𝐚𝐭𝐮𝐥𝐚𝐭𝐢𝐨𝐧𝐬 to Dr. Ashraful Islam and Prof. M. Ashraful Amin on this well-deserved accolade. Your achievements inspire our entire community to pursue excellence and push the boundaries of what is possible in science and research. 🥳🎊

19/07/2026

We are delighted to announce that our paper "RGC: a radio AGN classifier based on deep learning. I. A semi-supervised multiclass model for VLA images" has been accepted for publication in Astronomy & Astrophysics (A&A), one of the most prestigious peer-reviewed journals in astronomy and astrophysics.

Special congratulations to Mir Sazzat Hossain and Adrita Khan, Research Assistants at CCDS, on this achievement. This work was carried out in collaboration with the Center for Astronomy, Space Science and Astrophysics (CASSA), IUB.

In this work, the authors introduce FIRST-2060, a new labelled dataset of radio galaxies, and RGC 1.0, a semi-supervised deep learning model for classifying radio AGNs from VLA images. RGC 1.0 performs on par with the strongest supervised baselines while learning attention patterns that follow the true morphological structure of radio galaxies.

Congratulations to all the authors, collaborators, and supervisors on this excellent achievement! 🎉. This work hs been done under supervision of Dr. Khan Asad (CASSA) and Dr. AKM Mahbubur Rahman (CCDS).

19/07/2026

https://www.kalerkantho.com/print-edition/aI-in-higher-education/2026/07/16/1711956

16/07/2026

𝐂𝐂𝐃𝐒 𝐃𝐚𝐭𝐚 𝐒𝐜𝐢𝐞𝐧𝐜𝐞 𝐖𝐢𝐧𝐠 𝐃𝐢𝐫𝐞𝐜𝐭𝐨𝐫 𝐃𝐫. 𝐀𝐊𝐌 𝐌𝐚𝐡𝐛𝐮𝐛𝐮𝐫 𝐑𝐚𝐡𝐦𝐚𝐧’s interview on Daily Kaler Kontho today.

Full read: https://www.kalerkantho.com/epaper/2026-07-16/12

14/07/2026

Student researchers from the Center for Computational & Data Sciences (CCDS) of IUB have won the Best Paper Award at the 14th International Conference on Frontiers of Intelligent Computing: Theory and Applications (FICTA 2026). FICTA 2026 was organized by the London Metropolitan University, UK, on June 08-09, 2026. The team comprised Kazi Md. Rakibul Hassan, Komol Krisna Paul, Ayman Rahman, Siam Tahsin Bhuiyan, and Samiul Karim Mazumder. The award recognized the team's research on an artificial intelligence framework for riverine waste detection under real-world environmental conditions, conducted as part of their Final Year Design Project.

Read more: https://iub.ac.bd/news-and-events/news-archive/2026-June/6a55fcc9c9701c13403fe9f9

08/07/2026

CCDS Weekly Paper Presentation
On Monday, 29 June 2026, at 11:00 AM, the Center for Computational & Data Sciences (CCDS), Independent University, Bangladesh (IUB), hosted a paper presentation session on recent advances in "Self-Supervised Visual Representation Learning."

Presenter: Md Raqibul Islam
Research Assistant, CCDS, IUB

Paper Presented:
You Don't Need Strong Assumptions: Visual Representation Learning via Temporal Differences

The session featured an engaging discussion on a self-supervised learning paradigm called Temporal Difference in Vision (TDV), which learns visual representations directly from the natural motion in video rather than relying on hand-crafted augmentations, masking, or cropping. Discussion covered the paper's central argument that weaker inductive biases scale better with data, its jointly trained frame-and-motion encoder architecture, and its competitive performance on dense spatial tasks such as segmentation, optical flow, and stereo depth.

We thank all participants for their valuable contributions and insightful discussions.

On 29 June 2026, CCDS hosted a paper presentation on self-supervised visual learning from video motion, presented by Md Raqibul Islam.

20/06/2026

Acceptance in ECCV, WACV, ACL, EMNLP, CVPR workshop, AAAI workshop in one year!

RAs from AIML & DS wings just raised the bar