|
Aydin Javadov I'm a PhD student in Data Science at ETH Zurich supervised by Prof. Dr. Florian von Wangenheim and Prof. Dr. Bjoern Schuller (Imperial College London & TUM) I'm associated PhD Student at the ETH AI Center I'm also associated researcher at Chair of Health Informatics in TUM and at Group on Language, Audio and Music in Imperial College London Before that I was a Research Student for 1.5 years at BMW working on Explainable and Multimodal AI - where I also completed my Master's Thesis. I also have a First Class honours bachelor's degree in Computer Engineering from ADA University , Azerbaijan. |
|
Interested in a Master’s thesis or collaboration?
Topics of interest:
ajavadov [at] ethz . ch
Up-to-date CV: HERE
News:
My work "RAxSS: Retrieval-Augmented Sparse Sampling for Explainable Variable-Length Medical Time Series Classification" accepted at the Learning from Time Series for Health @ NeurIPS 2025. See you in San Diego, US.
-- 5th Dec 2025
New preprint of our work "Compose and Fuse: Revisiting the Foundational Bottlenecks in Multimodal Reasoning"
is available now.
-- 29th September 2025
-- 9th May 2025 I am at CHI'25! I will present our another accepted Position Paper and Poster "BioSyncHRI: Synchronizing Human Robot Interaction via Real-Time Biosignal Adaptation" CHI 25: Envisioning the Future of Interactive Health, in Yokohama, Japan!
-- April 27th 2025 Our work "Generative AI for Wellness Applications via User-Generated Immersive Virtual Environments" is at CHI 25: Generative AI and HCI Workshop, in Yokohama, Japan!
-- April 27th 2025 I started my PhD in Data Science at ETH Zurich!
-- 1 November 2024 I got my M.Sc. degree with Distinction in Data Engineering and Analytics from Technical University of Munich!
-- 23 July 2024 A new Mathematical details of Sensitivity analysis and Explainable AI is added to Scientific Notes corner below!
-- 10 May 2024 A new Insights Note on Machine Learning for Crowd Modelling and Simulation from the course (Physics-Enhanced Machine Learning Chair at TUM) is added to Scientific Notes corner below!
-- 1 February 2024 I defended my Master's Thesis to Prof. Hans-Arno Jacobsen's Jeffrey Skoll Chair in Computer Networks and Innovation at the University of Toronto!
-- 10 August 2023 I had a speech about Explainable AI in Medicine in Kaggle Munich held at Reply!
-- 19 May 2023 ...life too busy at TUM...
-- 2022-2023 Won the "HackaTUM" hackaton!
-- 25 November 2021
ResearchI am actively engaged in research spanning Explainable AI, (probabilistic) deep learning, computer vision, time series analysis, and their interdisciplinary (especially, physics-enhanced and medical) applications. Furthermore, my focus extends to exploring the mathematical foundations of learning and deep learning theory. |
|
Associated with: ETH Applied Mathemetics and ETH AI Center Poster Aydin Javadov, Anna Kravchenko, Fanny Lehmann, In this work, we explore the effectiveness of transformer-based architectures for seismic forecasting using a recent large-scale dataset (HEMEW-3D). Specifically, we implement and evaluate the Swin-UNETR model[2], which leverages hierarchical attention and convolutional decoding to learn future wavefield dynamics from historical displacement data. |
|
Associated with: Imperial College London Published at: 11th IEEE International Conference on Intelligent Data and Security Special Track: Financial Reinforcement Learning and Foundation Models (FinRLFM) Poster Emran Alturki, Aydin Javadov, Bjoern Schuller, We enhance FinRL-DeepSeek for the “FinRL Contest 2025 Task 1” by integrating LLM signals into RL with adaptive confidence weighting. Unlike fixed perturbations, our method scales each signal by its confidence, strengthening reliable inputs and reducing noise. Applied to PPO and CVaR-PPO, this yields more stable, risk-aware trading with improved backtest performance. |
|
Aydin Javadov, Tobias Lasser Project lab / code / Guided Research on interpretable deep learning and computer vision in medical (dermatology) context. |
|
Jamaladdin Hasanov, Samir Garibov*, Aydin Javadov* Applied Intelligence 2021 Journal / bibtex / A simple Adaptive Neuro-Fuzzy Inference System (ANFIS)-based model as an alternative to the formula-based closeness evaluation. |
|
Explainable AI for Hierarchical Learning and Clustering Algorithms Aydin Javadov, Viktor Leis Lab at TUM Sponsored and funded by BMW A comprehensive explainability framework for unsupervised, graph-based, black-box representation learning models, facilitating context-dependent explanations Completed with High-Distinction. |
|
Advanced Research in Analytics with Machine Learning and Data Visualization of Distributed Temperature Sensing Data Aydin Javadov, Samir Garibov, Samir Rustamov Sponsored by BP (British Petroleum) An exhaustive investigation encompassing data analytics and anomaly detection through the utilization of diverse machine learning techniques. This research also incorporates advanced time series analysis, interpolation techniques, and diverse 3D visualizations within a full-stack web environment. Completed with High-Distinction. |
|
Project lab Poster Aydin Javadov, Jin Luo, Milena Eisemann, Georg Groh Based on the lack of research on German Text Simplification and the additional challenges of its Leichte Sprache, following regulated guidelines, we present a new method for generating "Leichte Sprache" based on an LLM model. We explore different tools that make our models more explainable. |
Other ProjectsI actively engage in the collaborative development of impactful AI-powered applications, primarily through participation in hackathons. As a result of these experiences, I have co-founded Chasey, a startup currently in the pre-seed stage. Explore more about Chasey here. |
|
|
Project details Aydin Javadov, Toghrul Rahimli, Mikhail Erushev Sponsored by ZEISS and HackaTUM Machine Learning and Software Engineering solution, offering domain experts a tool to visualize the temperature data as well as detect and predict temperature fluctuations. Winner of HackaTUM 2021 ZEISS Challenge. |
|
Project details Aydin Javadov, Toghrul Rahimli, Murad Talibov Sponsored by TUM.AI A smartphone app and trained neural network help doctors in detecting melanomas. |
|
Project details/ code Aydin Javadov, Toghrul Rahimli, Murad Talibov Sponsored by HelloFresh and HackaTUM Smart Recommendations for Personalized, Eco-Friendly Meals |
Scientific
|
PointNet ++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space [Paper Go-Through]
But how do machines ‘learn’? Insights Note on Machine Learning for Crowd Modelling and Simulation Mathematical details of Sensitivity analysis and Explainable AI |
| | >
|
Map of You! :) |
|