Nader Nemati

About Me

Welcome to my personal webpage! I am Nader, a Machine Learning researcher with 4+ years of combined academic and industry R&D experience. My work spans generative AI, agentic systems, and deep learning, grounded in a strong background in statistics, optimization, and stochastic processes. Currently, I am building a fully automated ARUNet-based pipeline for ECG print-image digitization with advanced preprocessing, segmentation, and calibrated signal reconstruction, and I am extending it with a transformer refinement module and a privacy-aware federated learning framework.

At the University of Turku, I have worked on two parallel research lines: (1) maritime perception, where I developed an RT-DETR–based real-time detector with multi-scale fusion and synthetic domain augmentation for small-object detection; and (2) ECG image digitization, where I designed an end-to-end pipeline from image preprocessing to waveform reconstruction. Earlier, as a Project Researcher, I built an arrhythmia classifier and an ECG+BCG fusion approach for A-Fib detection using timing alignment and HRV/mechanical features.

I hold an M.Sc. in Computer Science from Shahid Beheshti University, where my thesis combined CNNs across multiple time frames for stock-market prediction, and a B.Sc. in Mathematics from Isfahan University of Technology on exponential matrix computation via the Magnus expansion. My current interests include medical image reconstruction (PET/CT/MRI), kinetic modeling and dynamic PET, signal and image processing, federated learning, and probabilistic modeling.

News

• 2025/10: My new research paper "Privacy-Aware Federated nnU-Net for ECG Page Digitization" is published on medRxiv. This study provides an end-to-end federated pipeline to prevent health centers from sharing the ECG images with the central server!
• 2025/09: My new research paper "Enhancing Maritime Object Detection in Real-Time with RT-DETR and Data Augmentation" is published on medRxiv.
• 2024/06: I started a new research internship with Dr. Farahnakian. This study centered on jamming, spoofing, and multi-sensor fusion concepts, which belong to maritime object detection from RGB camera data
• 2024/06: I started a new research internship with Dr. JafariTadi in ECGs digitalization, involved in both image and signal processing.
• 2023/06: I had a presentation at the 5th International Congress on Human-Computer Interaction, Optimization, and Robotic Applications (HORA), Istanbul, Turkey.
• 2023/05: "Stock market prediction by combining CNNs trained on multiple time frames" was accepted at the 2023 5th International Congress on Human-Computer Interaction, Optimization, and Robotic Applications (HORA).
• 2022/09: I began developing a deep learning model based on ResNet 18 for heart abnormality detection from ECG signals and signal processing.
• 2022/02: I successfully defended my Master's dissertation.
• 2022/01: I joined the digital health technology lab at the University of Turku.

Publications

• N. Nemati, H. Farahani, SR. Kheradpisheh. (2023). "Stock market prediction by combining CNNs trained on multiple time frames." IEEE. DOI: 10.1109/HORA58378.2023.10156742
• N. Nemati. (2025). "Enhancing Maritime Object Detection in Real-Time with RT-DETR and Data Augmentation." Arxiv. DOI: https://arxiv.org/abs/2510.07346
• N. Nemati. (2025). "Comparative analysis of data augmentation for clinical ECG classification with STAR." medRxiv. DOI: https://doi.org/10.1101/2025.10.23.25338628
• N. Nemati. (2025). "Privacy-Aware Federated nnU-Net for ECG Page Digitization." medRxiv. DOI: https://doi.org/10.1101/2025.10.25.25338789
• N. Nemati, M. JafariTadi. (2025). "A Fully-Automated ARUNet Pipeline for ECG Print-Image Digitization." (To be submitted to physiological measurement journal)

Skills

Machine Learning and Data Science

• Machine Learning: Python (Scikit-learn, XGBoost, SciPy)
• Deep Learning: TensorFlow, PyTorch, Keras
• Generative Neural Networks: Diffusion models for pix-to-pix image analysis
• Data Analytics: Exploratory data analysis, data quality management
• Data Visualization: Python (Matplotlib, Seaborn)
• Data Cleaning: Python (Pandas, NumPy)
• Statistical Analysis: R, SPSS

Computing and Programming

• Parallel Processing: openMPI, CUDA
• Cluster Computing: CSC (PUHTI, MAHTI)
• Programming Languages: Python, SQL, R, MATLAB, Java, C/C++, SAGE, JavaScript, HTML/CSS
• Cloud Services: Azure

Languages

• Persian: Native
• English: Advanced
• French: Basic

Projects

Education

Contact Me

Feel free to reach out to me via email at naderr.nemati@outlook.com or connect with me on LinkedIn for any professional inquiries or collaboration opportunities.