About
I am an Applied Scientist II at Amazon. My public research focuses on reliable multimodal AI, including uncertainty quantification, model calibration, confidence-aware decision making, and vision-language systems.
At Rochester Institute of Technology, my doctoral research is advised by Dr. Xumin Liu and Dr. Qi Yu in the Machine Learning and Data Intensive Computing Laboratory.
Recent public work includes ICML 2026 and ICLR 2026 papers on calibrated knowledge aggregation and selective LLM integration for reliable visual question answering.
I received my B.S. in Software Engineering from Sharif University of Technology and my M.S. from the University of South Florida, where I worked on online anomaly detection in high-dimensional data streams.
Research Overview
My public research studies how AI systems can make reliable decisions under uncertainty.
I focus on calibrated and confidence-aware multimodal systems, especially vision-language models, visual question answering, and LLM-assisted decision pipelines. A central theme of my work is using uncertainty not only as a diagnostic signal, but also as a control signal.
My broader work also includes efficient generative models, tensor methods, and online anomaly detection.
Research Interests
Machine Learning, Deep Learning, Computer Vision, Generative AI, Multimodal Machine Learning, Tensor Learning, Uncertainty Quantification, Reliable AI.
News
- (May 2026) Started as an Applied Scientist II at Amazon.
- (April 2026) One paper is accepted at ICML 2026.
- (Jan 2026) One paper is accepted at ICLR 2026.
- (June 2025) Returning to AWS as an Applied Science intern.
- (Dec 2024) US patent issued: (Patent number: 12174689).
- (Dec 2024) I received the student scholarship award for AAAI 2025.
- (Dec 2024) One paper is accepted at AAAI 2025.
- (Oct 2024) One paper is published in IEEE Access.
- (May 2024) Started as an Applied Science intern at Amazon.
- (Mar 2024) I am invited to deliver an advanced PhD student talk in CHAI seminar series, at RIT.
- (Feb 2024) I am visiting the Computer Science Department at Wellesley College to present my doctoral research, in Boston.
- (Dec 2023) One paper is accepted at ICASSP 2024.
- (June 2023) One paper is published in Electronics.
- (May 2023) I am accepted into the AWARE-AI NSF Research Traineeship program, funded by the National Science Foundation (NSF), where I will be working in the area of multimodal AI.
- (Oct 2022) Two papers are published in Sensors.
- (Feb 2022) I passed my doctoral qualifying exam.
- (Jan 2022) I am selected to participate in Grad Cohort for Women (CRA-WP), in April 2022.
- (July 2021) I am selected as a student scholar to participate at vGHC 2021, in September 2021.
- (July 2021) One paper is accepted at ASILOMAR 2021.
- (June 2021) I’m selected as a speaker in vGHC-21, to present a poster “Robust Multilinear Subspace Estimation”.
- (May 2021) One paper is accepted at EUSIPCO 2021.
- (Aug 2020) I joined Machine Learning Optimization and Signal Processing Laboratory, at Rochester Institute of Technology.
- (July 2019) One paper is accepted at MLSP 2019.
Publications
- Calibrated Knowledge Aggregation in Bayesian Mixture-of-Experts for Continual VQA
M. Mozaffari, H. Sapkota, Y. Kong, X. Liu, Q. Yu
ICML 2026 [PDF] - Knowledge Exchange with Confidence: Cost-Effective LLM Integration for Reliable and Efficient Visual Question Answering
M. Mozaffari, H. Sapkota, X. Liu, Q. Yu
ICLR 2026 [PDF] - GLEN: Generalized Focal Loss Ensemble of Low-Rank Networks for Calibrated Visual Question Answering
M. Mozaffari, H. Sapkota, Q. Yu
AAAI 2025 [PDF] [Supp] - Enhancing GANs With MMD Neural Architecture Search, PMish Activation Function, and Adaptive Rank Decomposition
P. P. Pulakurthi, M. Mozaffari, S. A. Dianat, J. Heard, R. Rao, M. Rabbani
IEEE Access 2024 [PDF] - Enhancing GAN Performance Through Neural Architecture Search and Tensor Decomposition
P. P. Pulakurthi, M. Mozaffari, S. A. Dianat, M. Rabbani, J. Heard, R. Rao
IEEE ICASSP 2024 [PDF] - Self-Supervised Learning for Online Anomaly Detection in High-Dimensional Data Streams
M. Mozaffari, K. Doshi, and Y. Yilmaz
Electronics 2023 [PDF] - Online multivariate anomaly detection and localization for high-dimensional settings
M. Mozaffari, K. Doshi, and Y. Yilmaz
Sensors 2022 [PDF] - Real-Time Detection and Classification of Power Quality Disturbances
M. Mozaffari, K. Doshi, and Y. Yilmaz
Sensors 2022 [PDF] - Improved L1-Tucker via L1-Fitting
M. Mozaffari and P. P. Markopoulos and A. Prater-Bennette
EUSIPCO 2021 [PDF] - Robust Barron-Loss Tucker Tensor Decomposition
M. Mozaffari, P. P. Markopoulos
IEEE ACSSC 2021. [PDF] - Online Anomaly Detection in Multivariate Settings
M. Mozaffari and Y. Yilmaz
IEEE MLSP 2019 [PDF] - RAPID: Real-time Anomaly-based Preventive Intrusion Detection
K. Doshi, M. Mozaffari and Y. Yilmaz
ACM WiseML 2019 [PDF]
Patents
- System and method for online multivariate anomaly detection and localization
Inventors: M. Mozaffari, K. Doshi, Y. Yilmaz
US Patent: 12174689 [LINK]