Taki Hasan Rafi
Profile-photo placeholder for Taki Hasan Rafi

Taki Hasan Rafi

Ph.D. Candidate

Computer Science · Hanyang University

Seoul, Republic of Korea

Email Google Scholar GitHub

Robust & Trustworthy Machine Learning

Short Bio

I'm a Ph.D. candidate in Computer Science at Hanyang University, where I'm a research member of the Data Intelligence Lab under Prof. Dong-Kyu Chae. I'm also co-advised by Prof. Junegak Joung. My Ph.D. is jointly supported by the National Research Foundation of Korea (NRF) and Brain Korea 21. Previously, my Ph.D. was partially supported by Samsung Electronics. I was a research collaborator with Oracle, USA. Before joining Hanyang, I received a bachelor's degree in Electrical Engineering from Ahsanullah University of Science and Technology in 2021.

My research broadly solves challenges in trustworthy ML. Particularly, robustness (e.g., test-time adaptation), safety & evaluation, and privacy-preserving. Specifically, my research answers the following questions: (1) How does the model understand a dynamic test stream? [CIKM 25', TMLR(a), TMLR (b)] (2) How robust are the model dynamics under open-world settings? [WACV 26'] (3) How do models handle corrupted and noisy data during testing under diverse settings? [DASFAA 25'] (4) How can the production industry effectively utilize reliable ML frameworks? (5) Do we trust LLMs/VLMs for specific tasks? [ACL 25', NAACL 25', FAccT 26'] (6) How can ML models preserve user data privacy? [EAAI, IF] (7) How robust are ML models in different applications? [ACCV 24', TCBB, DASFAA 25']

Research Output

Publications

2026 and Forthcoming

  1. Continual Test-Time Adaptation in Computer Vision: Methods, Benchmarks, and Future Directions

    Sarthak Kumar Maharana, Shambhavi Mishra, Yunbei Zhang, Shuaicheng Niu, Taki Hasan Rafi, Jihun Hamm, Marco Pedersoli, Jose Dolz, and Yunhui Guo

    TMLR

  2. Family Matters: A Systematic Study of Spatial vs. Frequency Masking for Continual Test-Time Adaptation

    Chandler Timm C Doloriel, Yunbei Zhang, Yu Yeonguk, Taki Hasan Rafi, Muhammad Salman Siddiqui, Tor Kristian Stevik, Habib Ullah, Al Fadi Machot, and Kristian Hovde Liland

    TMLR

  3. Learning from Unknown for Open-Set Test-Time Adaptation

    Taki Hasan Rafi, Amit Agarwal, Hitesh L Patel, and Dong-Kyu Chae

    🏆 WACV 2026 · Oral (A/R: 3.3%)

  4. AI Governance Needs ISO-like Interoperability Protocols, Not Just Laws

    Azmine Toushik Wasi, Mst Rafia Islam, Mahfuz Ahmed Anik, Taki Hasan Rafi, Md Manjurul Ahsan, and Dong-Kyu Chae

    🏆 ICML 2026 (Position)· Highlight (A/R: 5%)

  5. BengaliMoralBench: A Benchmark for Auditing Moral Reasoning in Large Language Models within Bengali Language and Culture

    Shahriyar Zaman Ridoy, Azmine Toushik Wasi, Koushik Ahamed Tonmoy, Taki Hasan Rafi, and Dong-Kyu Chae

    FAccT 2026 (A/R: 33%)

2025

  1. Towards Robust Continual Test-Time Adaptation via Neighbor Filtration

    Taki Hasan Rafi, Amit Agarwal, Hitesh L Patel, and Dong-Kyu Chae

    CIKM 2025 (A/R: 30%)

  2. MVTamperBench: Evaluating Robustness of Vision-Language Models

    Amit Agarwal, Srikant Panda, Angeline Charles, Hitesh Laxmichand Patel, Bhargava Kumar, Priyaranjan Pattnayak, Taki Hasan Rafi, Tejaswini Kumar, Hansa Meghwani, Karan Gupta, and Dong-Kyu Chae

    Findings of ACL 2025 (A/R: 18%)

  3. Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia

    Samuel Cahyawijaya et al. (with Taki Hasan Rafi)

    🏆 ACL 2025 · Oral (A/R: 5%)

  4. Instance-Aware Test-Time Adaptation for Domain Generalization

    Taki Hasan Rafi, Karlo Serbetar, Amit Agarwal, Hitesh L Patel, Bhargava Kumar, and Dong-Kyu Chae

    DASFAA 2025 (A/R: 32%)

  5. SweEval: Do LLMs Really Swear? A Safety Benchmark for Testing Limits for Enterprise Use

    Hitesh Laxmichand Patel, Amit Agarwal, Arion Das, Bhargava Kumar, Srikant Panda, Priyaranjan Pattnayak, Taki Hasan Rafi, Tejaswini Kumar, and Dong-Kyu Chae

    NAACL 2025 · Industry Track (A/R: 32%)

  6. Gaussian Regularization in Neural Graph Learning

    Azmine Toushik Wasi, Taki Hasan Rafi, and Dong-Kyu Chae

    🏆 DASFAA 2025 · Oral (A/R: 20%)

  7. CADGL: Context-Aware Deep Graph Learning for Predicting Drug-Drug Interactions

    Azmine Toushik Wasi, Taki Hasan Rafi, Raima Islam, Šerbetar Karlo, and Dong-Kyu Chae

    IEEE Transactions on Computational Biology and Bioinformatics (JCR < 5%)

2024

  1. GReFEL: Geometry-Aware Reliable Facial Expression Learning under Bias and Imbalanced Data Distribution

    Azmine Toushik Wasi*, Taki Hasan Rafi*, Raima Islam, Karlo Serbetar, and Dong-Kyu Chae (*=Equal first)

    ACCV 2024 (A/R: 32%)

  2. Towards Collaborative Fair Federated Distillation

    Faiza Anan Noor, Nawrin Tabassum, Tahmid Hussain, Taki Hasan Rafi, and Dong-Kyu Chae

    Engineering Applications of Artificial Intelligence (JCR < 5%)

  3. Fairness and Privacy-Preserving in Federated Learning: A Survey

    Taki Hasan Rafi, Faiza Anan Noor, Tahmid Hussain, and Dong-Kyu Chae

    Information Fusion (JCR < 2%)