Avalik loeng hübriidse statistilise modelleerimise võimalustest

7. mail kell 16.15 peab Linköpingi Ülikooli analüütilise sotsioloogia instituudi kaasprofessor Adel Daoud ingliskeelse avaliku loengu „Hübriidne statistiline modelleerimine: analüüsides inimeste tervist ja elutingimusi Aafrikas satelliidiandmete ja süvaõppe kombineerimise abil“ ("The power of the hybrid-statistical-modeling culture: exemplified by combining earth-observation satellites and deep learning to analyze health and living conditions in Africa"). Loeng toimub Lossi 36 õppehoone ruumis 215 ja Zoomis.

Siin on loengu ingliskeelne tutvustus.

A fundamental distinction in the sciences, particularly the social sciences, lies between predictive and causal research goals. Yet, the introduction of Machine Learning (ML), while initially lauded for its predictive ability, has blurred this line as researchers explore its use for causal inference. My article on the Three Statistical Cultures addresses this evolving landscape, identifying a Hybrid Modeling Culture (HMC). Within HMC, predictive and causal inference form an intricate synergy, suggesting that, at its limit, the distinction itself can dissolve. This talk will delve into a compelling manifestation of this hybrid culture: Planetary Causal Inference. We explore how combining Earth Observation (EO) data – granular geographical and temporal information from satellites – with causally-oriented ML overcomes the limitations of traditional survey or census data. This allows us to analyze critical social dynamics like urbanization, conflict, climate, and crucially, poverty, at fine spatial and temporal resolutions across the globe. As a central example, I will present our work using satellite imagery within this Planetary Causal Inference framework to measure and understand poverty dynamics in Africa, illustrating the potential for scalable, high-resolution insights into pressing global challenges.

Adel Daoud is a Senior Associate Professor at the Institute for Analytical Sociology, Linköping University, and an Affiliated Associate Professor in Data Science and Artificial Intelligence for the Social Sciences, Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg, Sweden. His research has both a social scientific and methodological orientation. For the social sciences, he researches global development, focusing on global poverty and health. Daoud implements novel methodologies in machine learning and causal inference. Daoud also leads The AI and Global Development Lab. Daoud is also the creator of a new podcast called The Journeys of Scholars, dedicated to understanding how leading academics achieve success.