20 Mar 2025
Sezione di Logica e Filosofia della Scienza
Machine learning in climate change mitigation scenarios: From model development to ensemble analysis
Machine learning (ML) has emerged as a critical tool in contemporary climate research. This workshop will explore prospects for ML tools to enhance the design and assessment of medium- and long-term scenarios of greenhouse gas emissions. We will cover some of the latest applications, from the use of ML in the development of integrated assessment models (IAMs) of climate change to the opportunities of ML to assist the construction and evaluation of large ensembles of emission scenarios. Due to the specific nature of ML tools, the workshop will also be an occasion to reflect on emerging issues of opacity and the potential challenges in the implementation of ML in influential exercises of climate policy evaluation, such as those of the IPCC.
Specifically, the workshop will address the following questions:
- DATA ENHANCEMENT: Can ML tools help streamline the collection and curation of data to feed the models (IAMs) that produce emission scenarios?
- MODEL ENHANCEMENT: Can neural networks and equations discovery tools integrate standard numerical modeling approaches to help re-design IAMs?
- POST-PROCESSING: Can ML emulators be used to detect bias and infill missing dimensions of uncertainty in large ensembles of emission scenarios?
- ASSESSMENT: What consequences can the use of ML tools have on the informativeness and trustworthiness of assessments of large scenario ensembles?
We look forward to the participation of engineers, mathematicians, and other researchers with an expertise in ML. One of the workshop’s aims is to build further capacity to assist climate research and explore connections across emerging fields of research.
The workshop will be held in Aula FDS, IV Floor, Department of Mathematics, Politecnico di Milano. Participation is free of charge. Registration is not required.
Contact: francesco.nappo@polimi.it
Specifically, the workshop will address the following questions:
- DATA ENHANCEMENT: Can ML tools help streamline the collection and curation of data to feed the models (IAMs) that produce emission scenarios?
- MODEL ENHANCEMENT: Can neural networks and equations discovery tools integrate standard numerical modeling approaches to help re-design IAMs?
- POST-PROCESSING: Can ML emulators be used to detect bias and infill missing dimensions of uncertainty in large ensembles of emission scenarios?
- ASSESSMENT: What consequences can the use of ML tools have on the informativeness and trustworthiness of assessments of large scenario ensembles?
We look forward to the participation of engineers, mathematicians, and other researchers with an expertise in ML. One of the workshop’s aims is to build further capacity to assist climate research and explore connections across emerging fields of research.
The workshop will be held in Aula FDS, IV Floor, Department of Mathematics, Politecnico di Milano. Participation is free of charge. Registration is not required.
Contact: francesco.nappo@polimi.it
Francesco Nappo
Aula FDS IV piano, dip. Matematica, ed. 14 (La Nave)