Climate change is widely agreed to be one of the greatest challenges facing humanity, if not the greatest. Emissions of greenhouse gases have caused a sharp increase in global temperatures since the industrial revolution, far more abrupt than any previous climate fluctuations. Changing temperatures have already led to increased incidence and severity of storms, droughts, fires, and flooding, besides triggering significant changes to global ecosystems, including the natural resources and agriculture that humanity depends on. These changes will accelerate due to feedback loops such as the melting of ice and release of methane from permafrost. The 2018 UN report on climate change estimates that the world has only thirty years to eliminate greenhouse emissions completely if we are to avoid catastrophic consequences.
Many in the ML community wish to take action on climate change, yet feel their skills are inapplicable. Our workshop (at ICML2019) and white paper (forthcoming) aim to show that in fact the opposite is true: machine learning, while no silver bullet, can be an invaluable tool in fighting climate change, encompassing a wide array of applications and techniques. Climate change is a complex problem, for which action takes many forms - from designing smart electrical grids to tracking deforestation from satellite imagery. Many of these actions represent high impact opportunities for machine learning (ML), as well as interesting problems for ML research.
Goals. In this workshop, we aim to do the following:
- Introduce the machine learning community to meaningful ways ML can be applied to address climate change.
- Show that this work is compelling from an ML perspective, even irrespective of climate.
- Provide a platform for researchers and entrepreneurs who are working in this space already.
- Inspire broader discussion within the ML community on climate action.
- Crucially, we are not interested purely or even primarily in applying ML to climate modeling - rather we are interested in applying ML to solutions to the problem of climate change (see “Areas of focus” below).
No politics. This workshop will consider technologies for addressing climate change; political discussion is wholly inappropriate for this venue. There will be no endorsement of any political stance, nor imputation of such to ICML organizers or attendees. All speakers will be strictly held to this standard.
Organizers. David Rolnick, Alexandre Lacoste, Tegan Maharaj, Jennifer Chayes, Yoshua Bengio
We welcome submissions on any topic which approaches ML as a solution to climate change, in two forms: (1) short papers (2-4 pages +refs) may be position papers, extended abstracts, propositions for novel solutions or ways ML might be applied to address climate change; (2) full papers (6-8 pages +refs) should present technical work which uses ML to address the problem of climate change. Submissions should use the ICML latex template. A summary of anticipated topics is detailed below, however we welcome creative ideas/solutions! All accepted papers will be presented in the form of posters, as well as as many spotlights as time permits. Exceptional submissions will be considered for longer talks, and two prizes will be awarded, for best full paper and best short paper.
We have categorized the areas in which we see machine learning playing a meaningful role in the fight against climate change. For example, education of women has been ranked as one of the top ten most impactful actions on climate change. While not all of the areas listed below will be discussed in the workshop, we aim to cover as many as possible through talks and posters. A more detailed examination of all these topics will be provided in a white paper currently underway by the organizers. This will be complementary to the workshop and is intended to provide concrete steps to attendees and other members of the community who wish to be involved in climate-positive action.
Mitigation
- Energy: Smart grids, accelerated science for battery design and photovoltaic cells, wind flow modeling, variable renewable energy forecasting, turbine placement and management, emissions modeling
- Transportation: Autonomous vehicles, intelligent transportation systems, telepresence
- Architecture and cities: Smart buildings, heating & cooling, reducing food waste
- Industry: Industrial robots, supply chain management, tracking hydrofluorocarbons (HFCs), industrial demand response, natural gas leak detection
- Sequestration: Optimizing carbon-capture materials/methods, CO2 leak detection
- Agriculture, Forestry and Other Land Use (AFOLU): Farming robots, sensing/interpreting data for precision agriculture, tools for farm management, methane capture, plant-based foods, tracking deforestation, automated afforestation, carbon stock estimation
Adaptation
- Climate models: Fine-grained and multi-resolution models (higher spatial & temporal resolution), cross-domain interactions (e.g. atmospheric/coastal) and domain transfer, tracking ice
- Societal impacts: Effects on agriculture, natural resources (fisheries, forestry, etc.), climate-informed disease models, analysis of water systems (rain, rivers, deltas, dams), migration patterns, ecosystem monitoring
- Extreme events: Fire prediction and intervention, storm prediction, coastal engineering, lethal heat events
Society
- Tools for individuals: Cost/benefit analysis of personal power usage/renewables, impact calculation and recommendations, local resources (e.g. compost), voluntary carbon tax
- Tools for cooperation and coordination: Large-scale planning, aggregation of preferences, mechanism design / game theory, structural risk estimation and management
- Education: Personalized education, education of women
- Data communication: Visualization and data interpretation, news recommendation
- Finance: Cost/benefit analysis, financial instruments related to climate change, tools for climate-positive investment
- [Chad Frischmann]
- [Claire Monteleoni]
- [John Platt]
- [Karthik Mukkavalli]
- Yoshua Bengio (Mila, University of Montreal)
| Time | Event |
|---|---|
| 8:35 - 8:45 | Opening Remarks |
| 8:45 - 9:15 | Speaker: Title |
- Title
Authors
The posters are listed in order of submission.