I am currently a Postdoctoral Researcher at IIIT Bangalore, working in collaboration with Siemens. My research focuses on industrial AI applications, including programmable logic controller (PLC) code modeling, neural-symbolic representations, and knowledge-grounded AI systems.
Previously, I was a Postdoctoral Scholar at the Center for Research in Computer Vision (CRCV), UCF, USA, where I worked on multimodal medical AI, focusing on medical visual question answering (VQA), causal reasoning, and image-text understanding. I developed systems that integrate domain-specific knowledge graphs, logical reasoning, and vision-language models for enhanced interpretability in medical diagnosis.
My broader research vision spans interpretable AI, multimodal learning, and causal inference, with a strong emphasis on building trustworthy human-AI collaboration frameworks β particularly for high-stakes domains like healthcare and industrial automation.
Authors: Ron Campos, Ashmal Vayani, Parth Parag Kulkarni, Rohit Gupta, Aizan Zafar, Aritra Dutta, Mubarak Shah
Conference: WACV 2026
ππ― ππ―π΅π¦π³π’π€π΅πͺπ·π¦ ππ¦π°ππ°π€π’ππͺπ»π’π΅πͺπ°π― ππ©π’π΅π£π°π΅, moving beyond raw coordinates, GAEA provides rich contextual insights about locations directly from images.
[Paper] [Dataset] [Project Page]Authors: Shaina Raza, Rizwan Qureshi, Anam Zahid, Joseph Fioresi, Ferhat Sadak, Muhammad Saeed, Ranjan Sapkota, Aditya Jain, Anas Zafar, Muneeb Ul Hassan, Aizan Zafar, Hasan Maqbool, Ashmal Vayani, Jia Wu, Maged Shoman
Conference: arxiv 2025
Responsible Artificial Intelligence (RAI) challenges and opportunities in implementing ethical, transparent, and accountable AI systems in the post-ChatGPT era, an era significantly shaped by Gen AI.
[Paper]Authors: Aizan Zafar, Kshitij Mishra, Asif Ekbal
Conference: COLING 2025
MedEx integrates First-Order Logic (FOL) reasoning with knowledge graphs to model symptom-disease-treatment relationships, enhancing answer accuracy in medical QA.
[Paper]Authors: Aizan Zafar, Kshitij Mishra, Asif Ekbal
Conference: EMNLP 2024
MedLogic-AQA leverages First-Order Logic rules and neural networks to improve reasoning in medical QA, achieving state-of-the-art performance.
PaperAuthors: Aizan Zafar, Sovan Kumar Sahoo, Harsh Bhardawaj, Amitava Das, Asif Ekbal
Journal: Neurocomputing (2024)
KI-MAG enhances medical QA by integrating Knowledge Graphs (KGs) and synthetic data generation to improve model generalization. Compared to traditional QA models, our approach boosts BLEU scores by 15%, ensuring more accurate and context-aware medical responses.
PaperAuthors: Aizan Zafar, Sovan Kumar Sahoo, Deeksha Varshney, Amitava Das, Asif Ekbal
Journal: Journal of Intelligent Information Systems (2024)
KIMedQA enhances medical QA by refining Knowledge Graph selection and pruning, ensuring vector space consistency. The model outperforms ChatGPT on MASH-QA and COVID-QA, achieving higher F1 scores and human evaluation metrics.
PaperAuthors: Aizan Zafar, Deeksha Varshney, Sovan Kumar Sahoo, Amitava Das, Asif Ekbal
Journal: Applied Intelligence (2024)
Our model employs Medical Entity Scoring (MES) and Context Relevance Scoring (CRS) to rank medical entities from knowledge graphs like UMLS and PharmKG, achieving state-of-the-art results on MASH-QA and COVID-QA
PaperAuthors: Dibyanayan Bandyopadhyay, Aizan Zafar, Asif Ekbal, Mohammed Hasanuzzaman
Conference: IJCNN 2023
A transformer-based model for simultaneous Continuous Sign Language Recognition (CSLR) and text translation, significantly improving translation accuracy.
PaperAuthors: Deeksha Varshney, Aizan Zafar, Niranshu Kumar Behera, Asif Ekbal
Journal: Artificial Intelligence in Medicine (2023)
This research integrates medical knowledge graphs (UMLS) with pre-trained transformers, significantly improving medical chatbot accuracy and coherence in clinical settings.
PaperAuthors: Deeksha Varshney, Aizan Zafar, Niranshu Kumar Behera, Asif Ekbal
Journal: Scientific Reports, Nature (2023)
Propose a Masked Entity Dialogue (MED) model that augments knowledge graphs and leverages conversational history for context-aware medical chatbot responses.
PaperAuthors: Deeksha Varshney, Aizan Zafar, Niranshu Kumar Behra, Asif Ekbal
Conference: EMNLP 2022
Introduces CDialog, a multi-turn medical dialogue dataset with annotated medical entities, improving chatbot responses in COVID-related patient interactions.
PaperAuthors: Aizan Zafar, K Swarupa Rani
Conference: BigDML 2019
Proposes a density-based initialization method for K-modes clustering, enhancing clustering accuracy for categorical datasets.
paper
Multilingual chatbot supporting Hindi, Bengali, Telugu, and English with domain expertise in Railways, Judiciary, and Healthcare.
Comprehensive AI-driven Question Answering system for clinical text, integrating knowledge graphs.
Indian Institute of Technology Patna, India (July, 2019 β Feb,2025)
University of Hyderabad, Telangana, India (July, 2017 β June, 2019)
Guru Ghasidas Central University, Chhattisgarh, India (Aug, 2012 β July, 2016)
Email: aizanzafar@gmail.com; aizan_1921cs17@iitp.ac.in; aizan.zafar@ucf.edu
