cv
A Summary of basic related-information about me
Basics
Name | Behzad Nourani |
Label | AI Researcher |
behzadnourani.ai@gmail.com | |
Phone | - |
Url | https://behzadnouranikoliji.github.io |
Summary | - |
Work
-
2021 - 2025 Machine Learning Researcher
Tübingen AI Center
Performing research in Machine Learning. Designing novel algorithms that solve real-life industrial problems and outperform state-of-the-art in various realistic scenarios.
- Bandit Algorithms on Graphs, Multi-Agent and Network Systems, AutoML, Tabular Foundation Models
-
2019 - 2021 Machine Learning Research Engineer
DIET department of Sapienza University of Rome and Signal Processing Laboratory (LTS4) of EPFL
Performing research in Signal Processing for Big Data. Designing novel algorithms for the analysis of the data of Financial Markets as well as the data of Epidemics.
- Multi-Variate Time Series Models, Causal Structures, Graph Structure Learning from Streaming Data, Time-Varying Graphs, Regularization, and Optimization
-
05.2022 - 09.2022 Lecturer
University of Tübingen
We instructed the participants on how to design a recommender system using multi-armed bandits frameworks. We managed 2 teams, with 5 students in each team. We used the Scrum framework to manage the corresponding projects.
- Title: Bandit algorithms for designing recommender systems
Education
-
2017 - 2020 Rome, Italy
Master's degree
Sapienza University of Rome, Rome, Italy
Engineering in Computer Science
- Database Management Systems (SQL)
- Knowledge Representation and Semantic Technologies (SPARQL,Protégé)
- Software Engineering (Web Services)
- Information Retrieval (MapReduce, Apache Spark)
- Visual Analytics (Tableau)
Certificates
German Language | ||
Goethe Institute |
Docker | ||
Docker Certified Associate |
Amazon Web Services | ||
Amazon Web Services |
Publications
-
2024.10.10 Clusters Agnostic Network Lasso Bandits
ICLR-2025
This is a novel framework that can be used in the design and analysis of large-scale online recommender systems. The framework identifies clusters of nodes in a network of users/items and finds a solution for each cluster separately. It does not require any prior knowledge of the underlying graph that connects the users/items.
-
2024.09.30 Online Influence Maximization with Semi-Bandit Feedback under Corruptions
Arxiv
This is a novel framework that can be used in online marketing, influence maximization, and online recommender system design and analysis. The framework identifies the most influential nodes in a network to spread a signal while considering that some nodes in the network might be contributing to the generation of a certain amount of adversarial data.
-
2023.09.30 Piecewise-Stationary Combinatorial Semi-Bandit with Causally Related Rewards
European Conference on Artificial Intelligence (ECAI) 2023
This is a novel framework that can be used for the design and analysis of recommender systems in nonstationary environments as well as other real-world applications where the causal system in the data generation process changes in time. The framework models a time-varying causal system. In the design of the framework, we used Change Point Detectors to track nonstationarity of the data.
-
2022.07.23 Linear Combinatorial Semi-Bandit with Causally Related Rewards
International Joint Conferences on Artificial Intelligence (IJCAI) 2022
This is a novel framework that can be used for the analysis and design of online recommender systems as well as the analysis of data of epidemics. The framework relies on Structural Equation Modelling to model a causal system. The goal is to find places in the system that make the utiliy signal maximized.
Skills
Machine Learning | |
Reinforcement Learning | |
AutoML | |
Foundation Models | |
Tabular Data | |
Statistical Inference | |
Algorithm Design | |
Graph Signal Processing | |
Python (PyTorch, TensorFlow, scikit-learn, Numpy, Pandas, Matplotlib, CVXPY, NetworkX), C++, MATLAB |
Languages
German | |
B2 |
English | |
C2 |
Interests
Machine Learning/Deep Learning | |
- | |
- |
References
Doctor Claire Vernade | |
Professor Sergio Barbarossa | |
Projects
- 2024.10 - Present
Foundation Models for Tabular Data in AutoML
(Please click on the link to go to the github repository of the project)
- Foundation Models
- AutoML
- Tabular Data