Timezone: »

 
Workshop
Climate Change: How Can AI Help?
David Rolnick · Alexandre Lacoste · Tegan Maharaj · Jennifer Chayes · Yoshua Bengio

Fri Jun 14 08:30 AM -- 06:00 PM (PDT) @ 104 A

Many in the machine learning community wish to take action on climate change, yet feel their skills are inapplicable. This workshop aims to show that in fact the opposite is true: while no silver bullet, ML can be an invaluable tool both in reducing greenhouse gas emissions and in helping society adapt to the effects of climate change. Climate change is a complex problem, for which action takes many forms - from designing smart electrical grids to tracking deforestation in satellite imagery. Many of these actions represent high-impact opportunities for real-world change, as well as being interesting problems for ML research.

Fri 8:30 a.m. - 8:45 a.m.
Opening Remarks (Organizer's introduction) [ Video
Fri 8:45 a.m. - 9:20 a.m.

This talk will set the context around "AI for climate change". This context will include the difference between long-term energy research and shorter-term research required to mitigate or adapt to climate change. I'll illustrate the urgency of the latter research by discussing the carbon budget of the atmosphere. The talk will also highlight some examples of how AI can be applied to climate change mitigation and energy research, including ML for fusion and for flood prediction.

John Platt
Fri 9:20 a.m. - 9:45 a.m.
[ Video

It's hard to have climate impact! Lots of projects look great from a distance but fail in practice. The energy system is enormously complex, and there are many non-technical bottlenecks to having impact. In this talk, I'll describe some of these challenges, so you can try to avoid them and hence reduce emissions more rapidly! Let's say you've built a great ML algorithm and written a paper. Now what? Your paper is completely invisible to the climate. How do you get your research used by the energy system? I don’t claim to have all the answers; but I’d like to discuss some of the challenges, and some ideas for how to get round them.

Jack Kelly
Fri 9:45 a.m. - 10:10 a.m.
[ Video

The time is now for the AI community to collaborate with the climate community to help understand, mitigate, and adapt to climate change. In this talk, I will present two projects as part of interdisciplinary collaborations, one in the earth system sciences and one in the energy space, to illustrate specific use cases where AI is making an impact on climate change. I hope this talk will motivate you to contribute to tackling one of the greatest challenges of our time.

Andrew Ng
Fri 10:10 a.m. - 10:20 a.m.
[ Video

About 30-40% of food produced worldwide is wasted. This puts a severe strain on the environment and represents a $165B loss to the US economy. This paper explores how artificial intelligence can be used to automate decisions across the food supply chain in order to reduce waste and increase the quality and affordability of food. We focus our attention on supermarkets — combined with downstream consumer waste, these contribute to 40% of total US food losses — and we describe an intelligent decision support system for supermarket operators that optimizes purchasing decisions and minimizes losses. The core of our system is a model-based reinforcement learn- ing engine for perishable inventory management; in a real-world pilot with a US supermarket chain, our system reduced waste by up to 50%. We hope that this paper will bring the food waste problem to the attention of the broader machine learning research community.

Volodymyr Kuleshov
Fri 10:20 a.m. - 10:30 a.m.
[ Video

Climate change and environmental degradation are causing species extinction worldwide. Automatic wildlife sensing is an urgent requirement to track biodiversity losses on Earth. Recent improvements in machine learning can accelerate the development of large-scale monitoring systems that would help track conservation outcomes and target efforts. In this paper, we present one such system we developed. 'Tidzam' is a Deep Learning framework for wildlife detection, identification, and geolocalization, designed for the Tidmarsh Wildlife Sanctuary, the site of the largest freshwater wetland restoration in Massachusetts.

Clement DUHART
Fri 10:30 a.m. - 11:00 a.m.
Morning Coffee Break + Poster Session (Coffee Break + Poster Session)
Fri 11:00 a.m. - 12:00 p.m.
Achieving Drawdown - Chad Frischmann (Keynote talk) [ Video
Fri 12:00 p.m. - 1:30 p.m.

Catered sandwiches and snacks will be provided (including vegetarian/vegan and gluten-free options). Sponsored by Element AI.

Abraham Stanway · Alex Robson · Aneesh Rangnekar · Ashesh Chattopadhyay · Ashley Pilipiszyn · Benjamin LeRoy · Bolong Cheng · Ce Zhang · Chaopeng Shen · Christian Schroeder · Christian Clough · Clement DUHART · Clement Fung · Cozmin Ududec · Dali Wang · David Dao · di wu · Dimitrios Giannakis · Dino Sejdinovic · Doina Precup · Duncan Watson-Parris · G
Fri 1:30 p.m. - 1:55 p.m.
Personalized Visualization of the Impact of Climate Change (Invited talk) [ Video
Yoshua Bengio
Fri 1:55 p.m. - 2:30 p.m.
Advances in Climate Informatics: Machine Learning for the Study of Climate Change (Invited talk) [ Video
Claire Monteleoni
Fri 2:30 p.m. - 2:40 p.m.
[ Video

One of the most pressing questions in climate science is that of the effect of anthropogenic aerosol on the Earth's energy balance. Aerosols provide the `seeds' on which cloud droplets form, and changes in the amount of aerosol available to a cloud can change its brightness and other physical properties such as optical thickness and spatial extent. Clouds play a critical role in moderating global temperatures and small perturbations can lead to significant amounts of cooling or warming. Uncertainty in this effect is so large it is not currently known if it is negligible, or provides a large enough cooling to largely negate present-day warming by CO2. This work uses deep convolutional neural networks to look for two particular perturbations in clouds due to anthropogenic aerosol and assess their properties and prevalence, providing valuable insights into their climatic effects.

Duncan Watson-Parris
Fri 2:40 p.m. - 2:50 p.m.
[ Video

Soil moisture is an important variable that determines floods, vegetation health, agriculture productivity, and land surface feedbacks to the atmosphere, etc. Accurately modeling soil moisture has important implications in both weather and climate models. The recently available satellite-based observations give us a unique opportunity to build data-driven models to predict soil moisture instead of using land surface models, but previously there was no uncertainty estimate. We tested Monte Carlo dropout (MCD) with an aleatoric term for our long short-term memory models for this problem, and asked if the uncertainty terms behave as they were argued to. We show that the method successfully captures the predictive error after tuning a hyperparameter on a representative training dataset. We show the MCD uncertainty estimate, as previously argued, does detect dissimilarity. In this talk, several important challenges with climate modeling where machine learning may help are also introduced to open up a discussion.

Chaopeng Shen
Fri 2:50 p.m. - 3:00 p.m.
[ Video

Due to imbalanced or heavy-tailed nature of weather- and climate-related datasets, the performance of standard deep learning models significantly deviates from their expected behavior on test data. Classical methods to address these issues are mostly data or application dependent, hence burdensome to tune. Meta-learning approaches, on the other hand, aim to learn hyperparameters in the learning process using different objective functions on training and validation data. However, these methods suffer from high computational complexity and are not scalable to large datasets. In this paper, we aim to apply a novel framework named as targeted meta-learning to rectify this issue, and show its efficacy in dealing with the aforementioned biases in datasets. This framework employs a small, well-crafted target dataset that resembles the desired nature of test data in order to guide the learning process in a coupled manner. We empirically show that this framework can overcome the bias issue, common to weather-related datasets, in a bow echo detection case study.

Mohammad Mahdi Kamani · Sadegh Farhang · Mehrdad Mahdavi · James Wang
Fri 3:00 p.m. - 3:30 p.m.
Afternoon Coffee Break + Poster Session (Coffee Break + Poster Session)
Fri 3:30 p.m. - 3:45 p.m.
[ Video

This talk will outline how to make climate science datasets and models accessible for machine learning. The focus will be on climate science challenges and opportunities associated with two distinct projects, 1) EnviroNet: a project focused on bridging gaps between geoscience and machine learning research through a global data repository of ImageNet analogs and AI challenges, and 2) a Mila project on changing people's minds and behavior through visualization of future extreme climate events. The discussion related to EnviroNet will be on how datasets and climate science problems can be framed for the machine learning research community at large. The discussion related to the Mila project will include climate science forecast model prototype developments in progress for accurately visualizing future extreme climate impacts of events such as floods, that particularly impact individual's neighborhoods and households.

Surya Karthik Mukkavilli
Fri 3:45 p.m. - 4:20 p.m.
[ Video

DeepMind has proved that machine learning can help us solve challenges in the Energy sector that contribute to climate change. DeepMind Program Manager Sims Witherspoon will share how they have applied ML to reduce energy consumption in data centers as well as to increase the value of wind power by 20% (compared to a baseline of no realtime commitments to the grid). Sims will also highlight insights the team has learned in their application of ML to the real-world as well as the potential for these kinds of techniques to be applied in other areas, to help tackle climate change on an even grander scale.

Sims Witherspoon
Fri 4:20 p.m. - 4:30 p.m.
[ Video

The road freight sector is responsible for a large and growing share of greenhouse gas emissions, but reliable data on the amount of freight that is moved on roads in many parts of the world are scarce. Many low- and middle-income countries have limited ground-based traffic monitoring and freight surveying activities. In this proof of concept, we show that we can use an object detection network to count trucks in satellite images and predict average annual daily truck traffic from those counts. We describe a complete model, test the uncertainty of the estimation, and discuss the transfer to developing countries.

Lynn Kaack · George Chen
Fri 4:30 p.m. - 4:40 p.m.
[ Video

We explore machine learning methods for AC Optimal Powerflow (ACOPF) - the task of optimizing power generation in a transmission network according while respecting physical and engineering constraints. We present two formulations of ACOPF as a machine learning problem: 1) an end-to-end prediction task where we directly predict the optimal generator settings, and 2) a constraint prediction task where we predict the set of active constraints in the optimal solution. We validate these approaches on two benchmark grids.

Neel Guha · Zhecheng Wang
Fri 4:40 p.m. - 4:50 p.m.
[ Video

Climate change is increasing the incidence of flooding. Many areas in the developing world are experiencing strong population growth but lack adequate urban planning. This represents a significant humanitarian risk. We explore the use of high-cadence satellite imagery provided by Planet, whose flock of over one hundred ’Dove’ satellites image the entire earth’s landmass everyday at 3-5m resolution. We use a deep learning-based computer vision approach to measure flood-related humanitarian risk in 5 cities in Africa.

Christian Clough · Ramesh Nair · Gopal Erinjippurath
Fri 4:50 p.m. - 5:15 p.m.
"Ideas" mini-spotlights (Spotlight talk)
Kevin McCloskey · Nikola Milojevic-Dupont · Jonathan Binas · Christian Schroeder · Sasha Luccioni
Fri 5:15 p.m. - 6:00 p.m.
Panel Discussion [ Video
Yoshua Bengio · Andrew Ng · Raia Hadsell · John Platt · Claire Monteleoni · Jennifer Chayes

Author Information

David Rolnick (University of Pennsylvania)
Alexandre Lacoste (Element AI)
Tegan Maharaj (Montreal Institute for Learning Algorithms)
Jennifer Chayes (Microsoft Research)
Yoshua Bengio (Montreal Institute for Learning Algorithms)

More from the Same Authors