Keynote Speakers
Damian Borth
University of St.Gallen
GeoSANE: Learning Geospatial Representations from Models, Not Data
Recent advances in remote sensing have led to an increase in the number of available foundation models, each trained on different modalities, datasets, and objectives, yet capturing only part of the vast geospatial knowledge landscape. While these models show strong results within their respective domains, their capabilities remain complementary rather than unified. Therefore, instead of choosing one model over another, we aim to combine their strengths into a single shared representation. We introduce GeoSANE, a geospatial model foundry that learns a unified neural representation from the weights of existing foundation models and task-specific models, able to generate novel neural network weights on demand. Given a target architecture, GeoSANE generates weights ready for finetuning for classification, segmentation, and detection tasks across multiple modalities. Models generated by GeoSANE consistently outperform their counterparts trained from scratch, match or surpass state-of-the-art remote sensing foundation models, and outperform models obtained through pruning or knowledge distillation when generating lightweight networks. Evaluations across ten diverse datasets and on GEO-Bench confirm its strong generalization capabilities. By shifting from pre-training to weight generation, GeoSANE introduces a new framework for unifying and transferring geospatial knowledge across models and tasks.
Dragi Kocev
Bias Variance Labs & Jožef Stefan Institute
Dragi Kocev is a Senior Research Fellow at the Department of Knowledge Technologies, Jožef Stefan Institute, Ljubljana, and co-founder and CEO at Bias Variance Labs. His research spans machine learning for structured output prediction, multi-label and multi-target learning, and the development of FAIR, trustworthy and reproducible resources for applied AI, with a particular focus on Earth observation and space applications.
He is PI of FAIR-EO (FAIR, Open and AI-Ready Earth Observation Resources, funded through OSCARS) and of the ESA projects AiTLAS and AiSTRA, WP leader in the Slovenian AI Factory, and PI of the Horizon Europe project TRUSTroke on trustworthy AI for stroke care. He leads the development of the AiTLAS toolbox and Benchmark Arena, and of the FAIR-EO Hub, an integrated catalogue of AI-ready EO datasets, models and pipelines. He is Program Committee co-chair for ECML PKDD 2026, an Action Editor for Machine Learning and Data Mining and Knowledge Discovery, and a member of the Young Academy of Europe.
From Benchmarks to Knowledge Bases:
FAIR and Trustworthy Resources for AI4EO
Earth observation has no shortage of models or datasets, but it has a shortage of comparable results. New architectures are routinely evaluated under inconsistent splits, loss functions and training regimes, and when the same models are re-run under a single controlled protocol, much of the reported progress does not survive. This is a problem for the field generally, and a blocking one for AutoML: meta-learning, performance modelling and automated pipeline design all depend on an empirical record that can be trusted and queried.
In this talk I will argue that benchmarking and open science infrastructure are two halves of the same problem, and describe our work on both. On the benchmarking side, I will present results from the AiTLAS Benchmark Arena (500+ models across 22 classification datasets) and a recent change detection benchmark (10 architectures across 10 heterogeneous datasets), where well-optimized Siamese U-Nets remain competitive with transformers and state-space models once computational cost is accounted for. On the infrastructure side, I will describe FAIR-EO and the FAIR-EO Hub: an ontology-based schema that describes datasets, methods, runs and evaluation results as linked data, making benchmark outcomes machine-actionable rather than merely human-readable.
Mitra Baratchi
Leiden University
Mitra Baratchi is an associate professor at the Leiden Institute of Advanced Computer Science (LIACS), Leiden University, where she leads the Spatio-temporal data Analysis and Reasoning (STAR) research group and co-leads the Automated Design of Algorithms (ADA) group.
Her research focuses on automated machine learning for spatio-temporal, time-series, and mobility data, with applications across urban, environmental, and industrial domains. She has led ESA- and NWO-funded projects on physics-aware machine learning and AutoML for Earth observation.
AutoML to Advance Earth Observation Research
Key Automated Machine Learning (AutoML) techniques, such as neural architecture search and hyperparameter optimisation, are increasingly applied to Earth Observation (EO)-related problems. However, developing AutoML for processing EO data is challenging, as automation removes human intervention and domain knowledge that help ensure robust machine learning for new EO applications. In this talk, I will present AutoML4EO through both data-centric and model-centric perspectives, covering the application of existing AutoML systems to EO problems and the design of entirely new systems. I will highlight specific characteristics of EO data relevant to the development and evaluation of AutoML4EO systems and offer a perspective on the challenges that must be addressed to create the next generation of AutoML4EO systems.