Announcement
Sample: applications open for MSc and undergraduate researchersSample
Sample news item. Replace it with a real announcement — see the editing guide.
DAISY Lab Data-Centric AI Systems Lab
Data-centric AI systems, built to survive real data.
We build AI systems that treat data as a first-class citizen — learning to optimize databases and infrastructure, making data accessible through natural language, and understanding the networks that data lives in.
Department of Computer Engineering
Sharif University of Technology, Tehran
Research directions
All researchText-to-SQL and Data Interfaces
Turning questions in plain language into correct queries over real schemas, and knowing when the answer cannot be trusted.
New directionMachine Learning for Systems
Learned components inside data systems — query optimization, indexing, caching and configuration tuning.
New directionSystems for Machine Learning
The infrastructure machine learning runs on — data pipelines, training and serving efficiency, and the operational side of models in production.
New directionAI Agent Systems
Language-model agents that plan, call tools and act over real data, and the evaluation needed to trust them.
New directionComplex and Social Networks
Inferring structure from information diffusion, modelling cascades, and predicting what spreads.
5 publicationsTrustworthy AI and Misinformation
Detecting manipulated content, labelling news early, and making predictions that report their own uncertainty.
4 publications
Latest news
All newsPublication
New preprint on audio-visual deepfake detection across domains
EAV-DFD uses a teacher-student structure to adapt an ensemble audio-visual deepfake detector to unseen domains with only a small sample of target data.
Publication
Uncertainty-aware recommendation with Bayesian deep ensembles
BDECF combines Bayesian neural networks with deep ensembles so a recommender can report how certain it is, not only what it predicts.
Publication
DANI published in Scientific Reports
Our network inference method that preserves the topological structure of the recovered graph appeared in Scientific Reports.
Selected publications
11 papers across 6 research directions.
Teacher-Student Structure for Domain Adaptation in Ensemble Audio-Visual Video Deepfake Detection
arXiv preprint, 2026Preprint
arXiv
BibTeX
@misc{abolhasani2026teacher,
title = {{Teacher-Student Structure for Domain Adaptation in Ensemble Audio-Visual Video Deepfake Detection}},
author = {Elham Abolhasani and Maryam Ramezani and Hamid R. Rabiee},
year = {2026},
eprint = {2606.15117},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2606.15117}
}DANI: Fast Diffusion Aware Network Inference with Preserving Topological Structure Property
Scientific Reports, 2024
DOIPublisherarXivCode
BibTeX
@article{ramezani2024dani,
title = {{DANI: Fast Diffusion Aware Network Inference with Preserving Topological Structure Property}},
author = {Maryam Ramezani and Aryan Ahadinia and Erfan Farhadi and Hamid R. Rabiee},
journal = {Scientific Reports},
year = {2024},
doi = {10.1038/s41598-024-82286-x},
eprint = {2310.01696},
archivePrefix = {arXiv},
url = {https://www.nature.com/articles/s41598-024-82286-x}
}Joint Inference of Diffusion and Structure in Partially Observed Social Networks Using Coupled Matrix Factorization
ACM Transactions on Knowledge Discovery from Data (ACM TKDD), 2023
arXiv
BibTeX
@article{ramezani2023joint,
title = {{Joint Inference of Diffusion and Structure in Partially Observed Social Networks Using Coupled Matrix Factorization}},
author = {Maryam Ramezani and Aryan Ahadinia and Amirmohammad Ziaei Bideh and Hamid R. Rabiee},
journal = {ACM Transactions on Knowledge Discovery from Data},
year = {2023},
eprint = {2010.01400},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2010.01400}
}FNR: A Similarity and Transformer-Based Approach to Detect Multi-Modal Fake News in Social Media
Social Network Analysis and Mining (SNAM), 2023
DOIarXiv
BibTeX
@article{ghorbanpour2023similarity,
title = {{FNR: A Similarity and Transformer-Based Approach to Detect Multi-Modal Fake News in Social Media}},
author = {Faeze Ghorbanpour and Maryam Ramezani and Mohammad Amin Fazli and Hamid R. Rabiee},
journal = {Social Network Analysis and Mining},
year = {2023},
doi = {10.1007/s13278-023-01065-0},
eprint = {2112.01131},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2112.01131}
}Software and data
Everything we release- Software
DANI
Fast, diffusion-aware network inference that preserves the topological structure of the recovered graph. Linear time, with a MapReduce version for large graphs.
- Dataset
Sample Text-to-SQL benchmarkSample
Placeholder for a dataset or benchmark released by the lab. Shows how a dataset landing page with download and citation looks.
Maryam RamezaniWe are taking students for the 2027 intake
3 open positions for students who want to work between machine learning and data systems at Sharif University of Technology.