AMLD 2024 Track: Accelerating Climate Change Action through Machine Learning


Machine learning (ML) techniques are increasingly used to address climate change, but there are many under-explored possibilities. Broadly speaking, the climate change community has interesting data, but is increasingly unable to handle this data due to its volume and variety. Conversely, ML researchers are interested in helping to address climate chance but are held back by a lack of connections and subject expertise. Climate change applications of ML are therefore mutually beneficial: they provide a challenging test-case for novel methods while also helping to solve an urgent global crisis. In recent years, this idea has been put to the test in a first generation of interdisciplinary work. To highlight this work and advance the field further, we believe interaction and collaboration between the two communities is essential.

To this end, we are organising a conference track during the Applied Machine Learning Days (AMLD). For our track, we are inviting experts working at the intersections of climate change with artificial intelligence, ML, data science and adjacent fields. Because AMLD attracts a large audience of technical experts in ML in particular, we hope that our track will foster in-depth discussions and lead to fresh ideas on how to use state-of-the-art ML techniques for the benefit of people and the planet.

About AMLD

This track is part of the Applied Machine Learning Days (AMLD). For information on how to attend the AMLD conference, please see here

Practical details


Time (Workshop) Time (Local) Event
Panel - NLP for climate solutions: retrieval augmented generation and other strategies for mananging climate information
Details: (click to expand) Panelists:
  • Tobias Schimanski, University of Zürich
  • Erik Lehman, GIZ
  • David Thulke, RWTH Aachen University
Moderator: Maria João Sousa, Climate Change AI
Open-source LLM Alignment for Promoting Sustainability
Details: (click to expand) Speakers:
  • Cesar Ilharco Magalhaes, Google
  • Dominik Stammbach, ETH Zürich
Multilingual Topic modelling of United Nations environmental initiatives: exploring transparency on climate, food and water
Details: (click to expand) Speaker:
  • Anne Sietsma, Wageningen University
Panel - Computer vision and multimodal AI for climate solutions: combining datastreams for wildfires, finance and disaster recovery
Details: (click to expand) Speakers:
  • Alok Singh, University of Oxford
  • Isabelle Tingzon, World Bank/GFDRR
  • Julia Gottfriedsen, OroraTech
Moderator: Maria João Sousa, Climate Change AI
The Earth-2 AI Tools for Climate Risk Assessment: Latest Developments and Applications
Details: (click to expand) Speakers:
  • Jussi Leinonen, NVIDIA
Global Flood Prediction: A Multimodal Machine Learning Approach
Details: (click to expand) Speaker:
  • Cynthia Zeng, MIT
Machine Learning for Climate Change: 3 interactive case studies on climate risk, supporting global youth and climate policy
Details: (click to expand) Speaker:
  • Kalyan Dutia, Climate Policy Radar
  • Harrison Pim, Climate Policy Radar
  • David Dao, GainForest

About the track

Overview of the track

The track will be comprised of three main components of one hour long each:

  1. Natural Language Processing (NLP) applications for climate change, including Large Language Model (LLMs) applications such as information retrieval, retrieval augmented generation and multi-LLM reasoning. This session will start with short presentations, followed by a panel discussion which includes audience questions.
  2. Computer vision applications for climate change, including remote sensing applications for domains like sustainable finance and climate risks. The format is the same as the first component.
  3. Case studies where domain experts will get a chance to contribute their knowledge to real world examples in moderated short group discussions.

Call for Submissions

There are two types of submissions currently open: presentations and posters. In both cases, the submission should relate to the content of the track – i.e. it should be relevant to both climate change and machine learning. In our selection, we will focus on the novelty and relevance of the findings. Note that this does not mean that only novel methods will be considered; if established methods are applied in a novel way, the contribution can still be valuable. As such, we encourage submissions from those in the climate change community who have limited experience with machine learning, as well as submissions from machine learning experts who have limited knowledge of climate change.


For posters, the submission process is organised centrally by AMLD. Posters will be paper printed posters that will be displayed on poster boards in the exhibition hall during the 2 days of the conference (March 25 and 26). Poster authors will be required to be present next to their poster during the lunch breaks, from 13h15 to 14h00 on Monday the 25th (and possibly the 26th as well, depending on number of posters received). Submissions have a 2500 character maximum and should be made before 31 January through this Google form.


For presentations, we are inviting submissions for presentations for components 1 and 2 (NLP and Computer vision, including multi-modal approaches). Interested parties can submit an extended abstract before 31st of January through this portal.

Note that since submission deadlines are the same for posters and presentations, you need to submit your work to both if you want to be considered for both. Because poster submissions are managed by AMDL centrally, it is sadly not possible for us to offer poster presentations to unsuccessful presentation submissions, unless that submission is also in the presentations submission system already. To minimise double work on your part, you can simply write “see presentation submission” in the poster form. Apologies for the inconvenience.


Anne Sietsma (Wageningen University)
Maria João Sousa (Climate Change AI; Cornell Tech)
Tobias Schimanski (University of Zürich)
Dominik Stammbach (ETH Zürich)