Agenda

Predictive Analytics World Climate
1-2 June, 2022

Amazon’s vision is to be earth’s most customer-centric company. This talk explores how the Alexa Hybrid Science team in Pittsburgh, PA applies a customer-centric lens to cutting-edge machine learning research. The team is responsible for developing on-device Alexa automatic speech recognition models to provide a faster, more reliable Alexa experience. Our research includes neural network compression techniques, end-to-end spoken language understanding and optimizing machine learning for edge devices

The quality of any machine learning or deep learning model depends on the values that define the model structure and corresponding hyperparameters. Many practitioners may find themselves investing countless hours manually searching for the right model and related hyperparameter values. Some use highly inefficient grid search methods. Others will use simple random sampling, which actually works fairly well. But alone, this method only offers a globalized search, and other sampling methods may be better suited to the job. 

Why not use machine learning to automate the search for the best model? 

This presentation details an advanced approach that uses both global and local search strategies that can be evaluated in parallel to ensure a quick and efficient exploration of the decision space. In the case of this presentation, a genetic algorithm (GA) will be examined for the global search because the selection and crossover aspects of a GA distinguish it from a purely random search. A generating set search (GSS) will be used to greedily search the local decision space. 

Accurate, timely estimation of carbon emissions is critical for businesses and governments to take action on climate change. The Climate TRACE coalition aims to furnish these estimates for all major sources of emissions globally, on a near real-time basis. In this session we’ll give an overview of our work on estimating emissions from one of the most important sources of greenhouse gases, power generation from coal. The approach focuses upon detecting plumes using a variety of approaches, including multi instance deep learning.

The energy landscape is going through a drastic transformation. We are moving away from centralized power plants to more distributed energy resources such as solar, electric vehicles. This transformation makes it very difficult for the grid to handle and manage. Through this session we would explore how using data science and artificial intelligence, industries can optimize their energy usage while aiming to reduce costs and meet their sustainability goals. We’d be delving deeper into some of the forecasting and predictive analytics techniques we have created and measuring their impact. Also, we would share insights from real-life use cases of our customers and how AI has helped them monitor, control, and optimize their energy assets.

QuantumScape is developing solid-state lithium metal anode batteries to make long-range, mass market electric vehicles faster charging, longer range, and more affordable. Software has been a key enabler of rapid learning that has spanned new materials discovery through production quality control. QuantumScape has deployed many different models of varying complexity up to neural networks for image segmentation. Working under the constraints of smaller data (10s to 1000s of labels) in a small team of developers has taught the value of empowering expert users to label data and construct their own models.

In applying AI to fight climate change, we appear to have the irresistible force meeting the immovable object. In this talk, I’ll present lessons that I’ve learned from work at Google where both AI and large-scale computation can be used for both climate mitigation and climate adaptation. The talk will include discussing carbon-aware computing, flood forecasting, and using machine learning to accelerate fluid modeling.

“Can I drive more efficiently?” was the question to be answered at the beginning of all data science efforts at Tracks. How can we accurately measure efficient driving of the thousands of truck drivers on the roads? This is a typical real-life question that is difficult to solve with ML methods. The answer to this question has sparked a number of follow up business questions that we are tackling with Tracks’ complex AI system. In this session you will learn how Tracks’ solution looks like, which ML methods are employed and which business questions are being answered. In a nutshell, how to turn ML into saved CO2. 

As demand for carbon credits accelerates, there is an immense challenge in scaling the supply of carbon offsets. It’s hard to create credits that are additional, non-leaky, and durable, and it’s impossible for all but the largest landowners to participate in carbon programs. Over the last 10 years, SilviaTerra has built technology that generates comprehensive forest inventories of unprecedented resolution and scale, enabling measurement and payment for a comprehensive set of beneficial outcomes across the landscape. This new market is making carbon and other types of natural capital work for all landowners – for every acre, every value, every year.

Models generalize best when their complexity matches the problem.  To avoid overfit, practitioners usually trade off accuracy with complexity, measured by the count of parameters.  But this is surprisingly flawed.  For example, a parameter is equivalent to one “degree of freedom” only for regression; it can be > 4 for decision trees, and < 1 for neural networks.   Worse, a major source of complexity — over-search — remains hidden.  The vast exploration of potential model structures leaves no trace on the final (perhaps simple-looking) model, but has outsized influence over whether it is trustworthy.

I’ll show how Generalized Degrees of Freedom (GDF, by Ye) can be used to measure the full complexity of algorithmic modeling.  This allows one to fairly compare very different models and be more confident about out-of-sample accuracy.  GDF also makes clear how seemingly complex ensemble models avoid overfit, and lastly, reveals a new type of outlier — cases having high model influence.

Deep learning models for forecasting and planning have shown significant promise for handling multiple variables, uncovering hidden patterns, and producing accurate forecasts. However, as one might expect, deep learning models are also complex and rife with pitfalls. Since these techniques often seem like a ‘black box,’ managers — both technical and nontechnical backgrounds — can find them hard to master.
In this session, Senior Data Scientist, Javed Ahmed will focus on the intuition behind various deep learning approaches, explore how managers can tackle highly complex models by asking the right questions, and evaluating the models with familiar tools.
Attendees at the Metis session will leave with the tools to:

●      Identify types of forecasting applications that can benefit from deep learning 
●      Broadly understand deep learning approaches relevant to forecasting 
●      Understand pitfalls related to deep learning approaches, and why simpler models may work better
●      Evaluate the results of a forecasting program

Climate change is increasing the frequency and severity of natural disasters. Natural catastrophes impact all critical infrastructures, and their resilience is essential for businesses and cities to operate effectively and safely. At One Concern, we combine machine learning and hazard modeling along with ML operational tools to better model the impacts of natural disasters on these critical infrastructures. By taking advantage of modeling, we can understand these potential impacts sooner to plan for and mitigate them. This helps to make our communities more resilient. This session will cover how One Concern applies Machine Learning algorithms to Natural Disaster Modeling. 

Carbon Re is a micro AI and Climate Tech startup developing solutions to help Foundation Industry manufacturers, such as cement, steel, chemicals and glass, transition to net-zero. These industries are vital to the global economy, producing 75% of all the material for manufacturing and construction sectors however, today, they represent 21% of global GHG emissions. Carbon Re’s first product, the Foundation Platform, is a software platform based on state-of-the-art process improvement techniques for foundation industries developed at the Institute for Manufacturing at Cambridge University (by our co-founder Daniel Summerbell) and built on cutting edge deep reinforcement learning expertise (by our co-founder Aidan O’Sullivan at UCL). Carbon Re’s solution means that manufacturers don’t have to choose between profitability and sustainability: they can cut their emissions today and improve their finances. This session will cover how Carbon Re combines Machine Learning and Process Improvement techniques to exploit the efficiency opportunities in manufacturing – which is the main path to decarbonisation in the short term.

About 30-40% of food produced worldwide is wasted. This represents a $165B loss to the US economy and poses major environmental problems: it is estimated that food waste contributes to up to 25% of all greenhouse gas emissions. This session 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 learning engine for perishable inventory management. Our system is currently deployed across 220 supermarkets in the US (handling ~2% of US produce volume) and has led to waste reductions of up to 50%. We hope that this talk will bring the food waste problem to the attention of the machine learning community. 

Discover how early-stage climate tech companies are using machine learning to help meet their challenges.

Methane, the primary component of natural gas, is responsible for 15% of global warming. Our mission of finding and stopping greenhouse gas emissions at huge scale is a critical step towards controlling climate change, but it also presents unique challenges. And, as a small startup, navigating the trade-offs between speed, accuracy, and cost in our data pipeline can often be the difference between survival and failure. In this talk, we will examine the difficulties ML pipeline design in cases where information, time, and money are constrained, and how to do so while hiding the sausage-making from our customers, who just want to know where their equipment is leaking, and want to know fast. By using a lean, iterative approach that involves input from every department, including engineering, operations, and business development, we stay focused on creating analytics that maximize value while reducing risk to the company. 

Globally, more than $200 billion worth of recyclable materials goes unrecovered annually. The economics and efficiency of identifying and sorting paper, plastics, metals, and other recyclables from the waste stream creates a major challenge for material recovery. In recent years, the waste industry has also faced stricter international quality standards for contamination-free imports of recycled materials, leaving the industry in search of cost-effective alternatives to meet these requirements. COVID-19 then forced many businesses to suspend recycling operations due to concerns for worker safety. Simultaneously, the pandemic increased demand for high-quality recycled feedstock to overcome supply chain interruptions and shifts in raw material availability.

Learn how AMP Robotics’ technology, which applies computer vision and deep learning to identify and differentiate recyclables found in complex, mixed material streams, is helping the waste industry meet these challenges by modernizing recycling—improving material quality, ensuring worker safety, increasing productivity, lowering costs, diverting waste from landfill, and reducing greenhouse gas emissions—while increasing overall rates of recycling and resource recovery.

Join our panel to find out how established industrial companies are using machine learning to address climate challenges.  

Financial institutions are playing an increasing role in the low-carbon transition by taking steps to accurately estimate, price, and disclose future climate risk. By quantifying their exposure to climate risks, financial institutions can more effectively allocate investments, avoid ‘stranded’ assets, and track adherence to Paris Agreement goals and shareholder commitments. However, it remains difficult for these institutions to assess climate related risks across a portfolio of assets and across different benchmark warming scenarios.I will cover large scale data transformation approaches as part of an end-to-end framework for quantifying annual, asset-level climate risk over multiple climate hazards including wildfires, inland flooding, and heat waves using simulations from global climate models participating in the Coupled Model Inter-comparison Project Phase 6 (CMIP6).We will be discussing techniques to quantify forward looking climate risk from 2020 to 2050 under multiple climate scenarios such as high-emissions (SSP5-8.5) and medium-emissions (SSP2-4.5) warming scenarios. I will also showcase intermediate steps to make the climate simulations and spatiotemporal data interpretable and actionable. We will cover ways to harmonize near real time observations from ground measurements and satellite derived data with forward looking climate risk projections for acute physical hazards for high accuracy predictive modeling.

We give an overview of recent developments in physics-informed AI and big data that are transforming the prediction of climate and weather in applications ranging from climate risk modeling for insurance to real-time forecasting for energy. Traditional climate and weather models require computationally expensive simulation of physical laws on supercomputers with hours to days of processing time and have limited capacity to incorporate ground-truth data sources. The development of cloud-based AI workflows based on deep neural networks provides an alternative approach to develop physical emulators of climate and weather processes that are highly scalable and natively tuned to utilize the petabytes of remote-sensing, ground-based and numerical simulation data from Earth observation that are generated daily. We present work that we are doing at Terrafuse AI, a startup out of Berkeley National Lab, to develop an AI-native climate risk and forecasting platform for problems ranging from high-resolution mapping of wildfire risk in California to real-time wind forecasting for aviation and renewable energy. 

Coffee Roundtables – grab your real coffee and share experiences virtually with your peers to explore the new and old challenges. Just like pre-show breakfast in a regular conference you’ll join a “round table” with fellow attendees and see where the conversation takes you.

Kick-starter: Share the impact the pandemic has had on your working environment, interaction with colleagues, management of projects and processes. Do you see digital transformation in your organization being accelerated as a result and what lasting effects do you think it will have on your career and working environment once the pandemic is over?

1. Introduction to Climate/Sustainable Finance & Its Link with AI (Pedro Baiz)
   

   The Financial Stability Board (FSB) after working on regulations to address the 2008/9 financial crisis, moved to the next biggest risk: Climate Change. The talk will provide a comprehensive overview of what constitutes Climate/Sustainable Finance. Standards and potential upcoming regulations, such as TCFD (Task Force on Climate related Financial Disclosures) among many others (e.g. SASB, GRI, IRFS, etc) will be covered. Finally, the role of AI and on this emerging field will be discussed.
   

2. Predictive Analytics for Climate Risk Assessment (Christian Spindler)
   

   Climate change is a systemic risk which impacts on all business sectors. It increases uncertainty and investment risk and endangers entire business models. Professional investors and asset managers are taking climate change more and more into account. Corporates are starting to quantify climate risk in their mid- and long-term strategies. Predictive analytics turns out to be key in making quantified assessments in mostly unexplored terrain: How are extreme weather risks impacting on production sites and physical assets of the firm? How is the upcoming carbon taxation impacting on the company now and in future? And how vulnerable is the global supply chain of the company against business interruption risks? This presentation explores methods and tools for climate risk quantification.
   

3. Sustainable Federated Learning for Predicting Climate Changes (Sharmistha Chatterjee) 
   

   The talk unveils the art of building sustainable federated machine learning models by considering different aspects of ethical AI. The talk will highlight various techniques of incorporating fairness in private federated learning while ensuring the sustainability of future smart ecosystems. By the end of the talk, the audience will get the know-how of sustainable federated learning, how to build a scalable distributed architecture, and important KPIs to focus on when designing such a system.

1. Turning Data Science into CO2 savings (Daniel Rohr)

“Can I drive more efficiently?” was the question to be answered at the beginning of all data science efforts at Tracks. This is a typical real-life question that is difficult to solve with ML methods. The answer to this question has sparked a number of follow up business questions that we are tackling with Tracks’ complex AI system. In this session you will learn how to turn ML into saved CO2.

2. Finding Sensitive Intervention Points by Mapping the Global Fossil Fuel Supply Chain (Jannes Klaas)

Over 84% of the worlds energy comes from fossil fuels. But oil, gas and coal are also a major transport good. By identifying the networked structure of the global fossil fuel supply chain we can identify sensitive intervention points. We present our work in creating an asset level network of the global fossil fuel supply chain. We will highlight some early research avenues enabled by this dataset, including the use of new modelling techniques.

3. Applying Machine Learning to Solve the Ocean-Bound Plastic Crisis (Bosse Rothe)

11 million tonnes of plastic are estimated to end up in our oceans every year. Cleanhub developed a digital solution that can scale plastic collection in high-impact countries. To deliver digital evidence about how much plastic was collected, all collection partners use CLEANHUB’s software to track and trace the entire recovery process. Detecting anomalies is key to mitigate fraudulent behaviours. We will discuss the problem, opportunities and the role of machine learning.

1. Data Can Feed the World. But Do We Have the Right Data? (Sharavani Basu and Sébastien Foucaud)

Amid the current health and economic crisis, another chronic but fast deteriorating one is the food crisis! Not to sound alarmist, but shifting climatic patterns, trade & geopolitical instability, rising population among others has brought millions to the brink of poverty and starvation. As the productivity gains through mechanization and chemistry-based solutions in an intensive mode is fast plateauing or even reversing due to resistance build-ups, Digital Agriculture & Precision Farming, powered with advanced Analytics and Data Science seems to hold much promise in building sustainable systems with reduced environmental impact. Challenges to use data in agriculture is an experience shared across industries: aggregating data at large scales can be difficult, but identifying the right data is where the real challenge lies! We will present different use cases of novel pest outbreak forecasts, where despite abundant data availability, new approach for data acquisition is urgently needed.

2. Challenges & Best Practices: How to Handle Weather Data in Forecasting Models with Success (Christian Schneider)

Weather happens all the time with an undeniable influence on behavior, such as consumption, and therefore also on commercial success. Though one can´t change the weather, there are many options to utilize weather for one’s advantage. However, weather is not just weather. What are the special characteristics of weather compared to other data? Which machine learning models are most suitable? Christian Schneider talks about challenges and opportunities using weather in scalable forecast solutions.

 3. ESG in the News: Text Mining for Sustainability Signals (Gerhard Rolletschek)

Many current ESG rankings lead to absurd results. Greenwashing occurs when companies can rank high in ESG scores by streamlining their reports and figures instead of truly reducing their ecological footprint. Using text mining based on predicate-argument-structures and using a few-shot learning             approach, we show how ESG-relevant signals in multiple dimensions can be extracted from unstructured news sources and how this can lead to a more balanced and fair assessment of ESG activities.

Meet the speakers, fellow attendees, sponsors, moderators – randomly for a quick chat, just like in real life. If you are a match you can exchange contact details with one click. If not, you simply move on to the next contact.

1. ESA Earth Observation Data Generation, Management and Access (Nicolaus Hanowski)
ESA is operating the most productive and sophisticated fleet of Earth Observation satellites in the world. Understanding the data content, the way it is processed and made accessible to everyone should facilitate new applications. New missions are relied for generating entirely new Earth Observation data sets and provide new opportunities for new applications. The presentation will illustrate easy data access options. The potential of “predictive” Earth Observation will be explained.

2. Why Open EO Data Should be Accessible for Everyone and How They Could Help Solve Our Global Issues (Niklas Jordan)
Earth observation data from public space agencies, such as the ESA, is available to everyone free of charge, but not everyone can access it. Technical and professional requirements make it challenging to access this treasure trove of data. In my presentation, I will discuss how, in addition to science and professional users, civil society, such as journalists, teachers, NGOs, and almost all citizens, can benefit from this data to solve our time’s major global problems.

3. Convolutional Neural Networks for Damage Assessment and Disaster Relief (Thomas Chen)
Natural disasters ravage the world’s cities, valleys, and shores on a monthly basis. Having precise and efficient mechanisms for assessing infrastructure damage is essential to channel resources and minimize the loss of life. Using a dataset that includes labeled pre- and post- disaster satellite imagery, we have conducted research training multiple convolutional neural networks to assess building damage on a per-building basis. In order to investigate how to best classify building damage, we present a highly interpretable deep-learning methodology that seeks to explicitly convey the most useful information required to train an accurate classification model. Participants in this session will learn about why interpretability is important and the ramifications of AI for disaster management. Our research seeks to computationally contribute to aiding in this ongoing and growing humanitarian crisis, heightened by climate change.

Europe mastered the corona crisis, but a bigger one is already on the list: the climate crisis. The EU set ambitious goals for their climate change programme, but still the question remains how to reach them. Hopes in ClimateTech are thus huge and AI is to expected to play a key role. However, AI is also a driver to CO2 emissions: it consumes a lot of energy and it is also used e.g. to detect new oil deposits. How helpful is AI really and is Europe prepared to lead in and with AI the climate movement? Finally, what political and legal changes, technical infrastructure, economic conditions and so on are required to strengthen the climate tech community?

Keynote 1: Why It’s Time to Quit Your Big Tech Job and Work on Climate (Eugene Kirpichov)

Most climate-concerned engineers are still unaware that they can directly work on climate solutions, because of the outdated idea that climate is about activism and non-profits. This may have been true a few years ago, but could not be further from the truth today. I quit Google AI last year to work on climate, and I will try to convince you, too. I’ll give an overview of the commercial climate tech ecosystem, the role ML plays in it, and how to find a climate job as an ML practitioner.

Keynote 2: Tackling Climate Change with Machine Learning (Nikola Milojevic-Dupont) 

Collaborations between researchers and academics are essential to scale up solutions and enable GHG emissions reduction at scale. I will give an overview of existing academic research at the intersection of climate change and machine learning. I will also present Climate Change AI, an organization of volunteers facilitating impactful applications of machine learning to climate change, highlighting resources that can be helpful for companies aiming to work in this space.

Join the moderators of the day for a wrap up of content, discussions, findings and an outlook.

For decades, climate activists would tell you that profiting from helping the planet is an oxymoron – and this used to be largely true, but has quickly become an outdated stereotype in the past five years. As consumer preferences, regulatory pressure, and public markets place increasing premiums on lowering emissions, climate tech entrepreneurs are uniquely well positioned to generate tremendous shareholder value. In this session I will discuss these recent market forces and highlight the lessons from climate tech companies who have achieved massive commercial success and done well by doing good, such as Apeel, Aurora Solar, and Farmers Business Network, and from building my own company, Therma.

A fast-moving video montage of professionals and students who are operating at the intersection of machine learning and climate & sustainability.

Today, there’s still a crude understanding of the operational efficiency of materials recovery facilities, where recyclables are processed—weigh trucks on the way in and weigh commodities on the way out. Facilities cannot report the purity of the products with certainty without a tractable way to measure it aside from breaking open bales and auditing quality periodically. An accurate observer could deduce this by counting contamination prior to its deposit into a bunker, but this is financially intractable and operationally burdensome.

However, a computer vision system that can accurately detect everything it sees offers a solution, and it’s gaining traction. Operators are installing AI-powered computer vision systems throughout their facilities to analyze the inputs and outputs of relevant parts of the process to create a holistic view of where material is flowing, and most importantly, what’s in it. This development is making it possible to observe with individual scrutiny the inputs and outputs of a plant to inform actions that influence these quantities. As an industry, we’re still in the early stages of unlocking the power of AI, but better observability will lead to insights and closed-loop actions that continue to make current recycling operations more efficient.

Estimating CO2 emissions from power plants is no easy task. And as with any challenging project, data quality is key to its success. No matter how complex and state-of-the-art a model is, it won’t be very good unless there is data in sufficient quantity and with adequate processing. Thus, this talk aims to demonstrate how we at TransitionZero and Climate TRACE rely on thorough analysis, dataset iteration and other data-centric tools to detect CO2 emissions of power plants from satellite imagery.

At Regrow we are building the products and tech which enable farmers to receive payments for climate-positive food production. However, these carbon markets, including the products and tech which enable them, are just the beginning. Additional ecosystem service markets have the potential to transform agriculture by incentivising sustainable practices. These markets can be enabled by reliable and high resolution information on agricultural practices at continental scales, which can only be achieved by applying the latest machine learning techniques to large satellite imagery datasets. Join this session to learn how cutting-edge machine learning and satellite data can support carbon markets, and how this tech will enable new products and markets in the future.

Every 5 minutes consumers and generators of electricity in Australia’s wholesale National Electricity Market must decide how much power to buy or sell at a given price. When there is an oversupply of electricity the market price becomes negative, costing generators millions of dollars per year. Until recently renewable generators assumed these negative price periods as a cost of doing business, but the increased frequency of negative price periods and the deployment of utility scale batteries has made the application of highly adaptive time series forecasting essential to integrating renewable energy and battery storage into the NEM. We will describe how Fluence uses Machine Learning to efficiently operate these renewable energy assets.

NCX maintains a high-resolution map of the forests of the United States to underpin our forest carbon marketplace. Recent work has focused on transitioning this from an annual data product to a quarterly updated dataset, testing the integration and combination of various machine learning techniques. We’ll present successes and failures, and relative impacts of processing remote sensing data using Deep Markov models with Pyro, deep sequence models like RNN’s, CNN’s, and transformers, and varying forms of image representation using VAE’s. We will present our evaluation of time, cost, and impact on company goals of precise, robust carbon accounting.

Comprehensive electrification of transportation is a crucial element of decarbonizing the world economy.  But the ability to design, test, and manufacture new batteries is already becoming a bottleneckin the race to carbon-neutral transportation. Battery development and manufacturing is a complex process that requires considerations of power, energy density, cost, form-factor, and safety. We will discuss the ways in which Toyota Research Institute has used ML to drive improvements across the lifecycle of this process, including basic materials science research, battery design, and ML-driven manufacturing and quality control, in accordance with kaizen, the Toyota principle of “continuous improvement”.

New materials are critical to unlocking technological innovations to fight climate change. However, the status quo for bringing these materials to market is often a decades-long process.  To meet global sustainability needs and sharply curb global emissions, we must bring breakthrough products to market faster and more sustainably.  At Citrine Informatics, we are working toward this mission by deploying ML for materials discovery at some of the world’s leading materials and chemicals companies. In this talk, I’ll outline Citrine’s approach to ML-driven materials discovery and share some case studies of its application to sustainability-oriented use-cases.

Climate change is increasing the frequency and severity of extreme weather events. In order to quantify the impacts of climate on financial systems, improved climate models are needed with much higher spatial resolution than is currently possible. Moreover the models must be able to accurately quantify uncertainty in the frequency and severity of impacts under different climate forcing scenarios. Terrafuse AI is developing deep learning-based climate models trained on large Earth observation and simulation data sets, deployed on cloud computing infrastructure. With this approach our aim is to develop a “digital twin” of the climate system and its societal impacts so that appropriate adaptation and resilience strategies can be developed.

Batteries enable portable energy and as such they are ubiquitous in modern daily life and a key component for societal adaptation to climate change.  Voltaiq is committed to empowering this transition to electrification. We will cover three case studies overviewing machine learning uses in commercial battery development. We start with an illustrative case study on data form, volume, and collection method at each step throughout battery development. Next we will deep-dive and provide a case study on the use of machine learning to predict key performance metrics for a battery. We end with a study of machine learning to identify features in battery data.

Join us for this deeply technical conversation on geospatial machine learning in climate: the relevant ML methods, challenges, data, and applications to planning and monitoring of nature-based solutions, climate risk, and more.

AI and Robotics are both dramatically shifting our industrial capabilities and opening new doors to our functional understanding and ways of supporting the natural world. Together these advances can enable something far beyond simply limiting our damage to the planet – they create the possibility of building a new relationship with nature, wherein our industrial footprint can be radically reduced and nature’s capability to support itself and all life on Earth (including us!) can be amplified.

Join us for this deeply technical conversation on machine learning for green energy: the relevant ML methods, challenges, data, and applications to energy supply/demand forecasting, grid management, energy efficiency, and more.

Cement is responsible for 8% of worldwide CO2 emissions. Fortunately, the footprint can be reduced by 60% if burnt limestone, the main ingredient, is replaced partly by limestone powder. However, such low-carbon recipes react much more sensitive to changes in the chemical and mineralogical composition, limiting reliable production to laboratory conditions.

Alcemy is changing this. Robert presents a Machine Learning control and data analytics case study to optimize production processes of multiple plants such that low-carbon cement and concrete can be manufactured in real plants and at scale.

Myst AI has over three years of experience delivering highly accurate forecasts to organizations in clean power like climate-conscious load-serving entities, renewable power providers, and more. This talk will discuss some of the key ingredients we’ve identified as critical to delivering ongoing accuracy in our deployed load forecasting models. We’ll speak in particular to solutions to real-world data complexities: cross-validation in a domain where upstream forecasts are themselves critical features and historical data update over time, as well as ensembling approaches that solve unique challenges around data availability.

Transportation industry is responsible for 29% of the carbon emissions in the United States. In Convoy we use cutting edge technology to play a part in reducing waste in the trucking industry. Routing in particular, can be optimized to reduce the empty miles that trucks drive. In the network efficiency team we use Machine Learning to learn from driver’s routing behavior and Operations Research to optimally plan their work.

Better, cheaper and more abundant batteries will be a key driver of a faster transition to sustainability.  Improving batteries requires pushing the boundaries and empirically comparing many variations on design.  Therefore, quantifying and comparing important issues and performance metrics enables faster improvements. In this session, we use neural nets and various data-oriented processes to extract such a metric  from the raw timeseries cycling data from scratch,  and present our learnings integrating various ideas such as structured labels, ensembles, distillation, various active learning techniques as well as model-assisted quality control of data to tame this challenging data modality.

Soil-based carbon sequestration has attracted substantial attention due to its massive potential capacity and relative ease of implementation. However, scalable and accurate soil measurement methods have been elusive, making verification of sequestration a notable barrier. We present an ML-based approach relying on multispectral satellite imagery as a scalable, robust measurement methodology for soil carbon on farmland. The solution performs above existing standards under testing in the USA and Australia, and at a much lower cost. The audience will learn the background on carbon credit markets and soil-based sequestration, as well as select model implementation details and real-world test results.

At Open Climate Fix we aim to use ML to address climate change. We do this by connecting researchers using the most recent ML learning models with industry practitioners. The first challenge we are addressing is Solar nowcasting. Solar energy is predicted to be the largest form of generation globally by 2040. and having accurate real-time forecasts is hugely important to balance the energy system in a carbon effective way. We are delivering this service for the grid operator in the UK, National Grid. This talk shows the ML techniques we used to tackle this problem.

Recycling is one of the primary levers we can use to fight our society’s impact on our climate, and the industry has tremendous potential to improve its operations and drive real impact in our fight against climate change.

Glacier is deploying cutting-edge ML to perform classification and detection work that is revolutionizing the circular economy. Our challenges are unique and complex – distinguishing plastic resins, characterizing truckloads of trash, automatically detecting contamination. We’re using ML to extract value out of the products we consume and throw away, to ensure that businesses in the space have the technology they need to thrive, and to make an immediate impact on our climate.

Granular, multi-year time series of building-level electricity consumption are fundamental to distribution planning processes to support the adoption of solar panels, battery storage, and electric vehicles. These data are often measured using smart meters. However, some electric utilities do not have the hardware or data transmission processes in place to collect this key dataset. The absence of time series data can slow the adoption of distributed energy resources by making it difficult to prepare for their impacts on distribution grid infrastructure. This session covers Kevala’s load simulation tool, a modeling system used to synthesize granular electricity consumption time series data at the building level. The load simulation tool relies on static features, such as parcel attributes, and time-varying features, such as weather, to simulate hourly demand values over arbitrary time spans. Kevala’s load simulation tool learns relationships between these features and the scale and temporal variability of measured energy consumption values from smart meter readings. The results of Kevala’s load simulation tool have been used as a basis for modeling how the proliferation of solar panels, battery storage, and electric vehicles impacts electricity consumption.

At One Concern, we develop models for digital twins and their resilience by employing machine learning and advanced statistical methods to build a platform where organizations, communities, and private and public sectors understand, forecast, and mitigate climate risk. This session will cover how One Concern develops digital twins and resilience models by applying machine learning algorithms.

The quality and disclosure of corporate carbon emissions is critical in benchmarking progress towards net-zero emission targets for regulators and the wider market. in addressing climate change. We investigate the use of machine-learning algorithms to predict and forecast corporate carbon footprints, in particular we compare ML models for Scope 1, 2 and 3 emissions, comparing their performance against traditional OLS linear models. We forecast carbon footprints based on the availability of predictors in a three-step framework (Historical Model, Energy Model and Financial Model).  We find that machine learning algorithms improve prediction accuracy for firms without historical emission disclosure data. The largest gain comes from Linear Random Forest model. In contrast, past emission data is the best predictor for disclosing firms, and machine learning algorithms under-perform the traditional OLS estimator for these firms.

In recent years, we have seen an increase in power outages resulting in brownouts, blackouts and deadly wildfires, which will only increase if we don’t change how we use our energy resources. Manik will discuss the opportunity businesses have to leverage ML paired with sensors & smart plugs to use cooling – refrigeration and air conditioning – dynamically to reduce energy consumption, carbon emissions and operating costs. Manik will share how companies like 7/11, McDonald’s, Marriott Hotels & Vail Resorts are using this technology to shift energy usage to cleaner times of day, relieving grid stress while reducing energy costs without affecting the quality of food and guest experience.

In any major climate-linked crisis, access to geo-spatial-temporal datasets, mapping, modeling, and analytical tools are critical to aid recovery efforts and protect communities. Many expert researchers, especially scientists and institutions in low-income countries, lack the tools to access and analyze relevant data, to inform local decision makers on how to act rapidly and effectively. Currently, there is no mechanism to harmonize global access to urgently needed datasets and high-performance computing tools. This table discussion, led by the International Science Reserve, will bring together climate tech and corporate leaders to discuss how to bridge the data analytics divide.

KoBold is accelerating the clean energy future by finding new battery metal deposits. This requires effectively leveraging various types of geoscience data and applying ML and physical computational techniques to guide exploration decisions.

We will walk through challenges we faced and lessons learned from developing and scaling scientific computation and ML techniques (FFT, derivative raster products, tree based ML models) to rasters, which requires a different set of considerations from scaling computation on tabular data. We will discuss key differences between computation of geospatial rasters vs. tabular data, and generalize our learnings to other climate change problem domains.

Recycling is one of the primary levers we can use to fight our society’s impact on our climate, and the industry has tremendous potential to improve its operations and drive real impact in our fight against climate change.

Glacier is deploying cutting-edge ML to perform classification and detection work that is revolutionizing the circular economy. Our challenges are unique and complex – distinguishing plastic resins, characterizing truckloads of trash, automatically detecting contamination. We’re using ML to extract value out of the products we consume and throw away, to ensure that businesses in the space have the technology they need to thrive, and to make an immediate impact on our climate.

The present era is marked by the imminent reality of climate change, a phenomenon that carries undeniable scientific evidence regarding its profound impact on society. Nevertheless, we possess the means to confront this challenge through a multi-faceted approach. This deep dive delves into various strategies tailored to data scientists, environmental researchers, and corporations seeking to analyze corporate climate mitigation initiatives. The discourse entails a comparative analysis between corporate environmental and sustainability disclosures and the Intergovernmental Panel on Climate Change (IPCC) mitigation report, leveraging the power of Natural Language Processing (NLP) and Topic Modeling. By delving deep into the subject matter, the talk elucidates the capacity of the topic model to extract latent themes from the IPCC report and subsequently rank climate investments in relation to the corporate disclosures.

Electricity Maps maintains a comprehensive hourly dataset for electricity generation and exchanges. Our most recent work leveraged this dataset to predict the best hour to use electricity to minimize carbon footprint. We will present how we broke down electricity generation into granular time series production events which can be forecasted with simple Lasso models and the infrastructure we built to support it. We will discuss challenges we encountered along the way and how we used those predictions to deploy a carbon aware scheduler for running our computational workloads. Finally, we will present our evaluation of avoided emissions using this scheduler.

How will we meet the nutritional needs of a growing population using fewer hectares per ton of food, with lower labor availability? Conventional agriculture will not get us there. With AI and machine-learning integrated into modern agricultural solutions, we can leapfrog to precision agriculture and encourage soil regenerative practices, driven by actionable data. For example, sensor-supported precision agriculture for efficient water (80-90% reduction) and nutrient delivery (80% less fertilizer use) can cap input costs while significantly improving production output through data-driven harvesting and planting schedules.

In this session, we will explore the transformative power of machine learning in optimizing sustainable energy production from agricultural waste. Specifically, we will examine how predictive analytics can be employed to forecast spatial biomass distribution, efficiently allocate resources, and design a cost-effective bioenergy supply chain. Using real-world examples, we’ll demonstrate how these technologies are vital in transitioning to a lower-carbon future, benefitting both the environment and farmers. This discussion promises to provide insightful, practical knowledge on employing machine learning for sustainability challenges in energy sector.

In this session we will look at several examples of use of AI and predictive modeling applied to the natural world – and enabling trust and collaboration within the value chain that supports our food industry. We’ll look at how TheoryMesh is enabling data analysis from crop farming through food processing, and on to consumers, and at other use cases of the use of AI in anticipating climate change for woodland creation and conservation, and soil protection, leveraging LIDAR and other public data.

The end goal is to find the balance between current insights and future trends – both social and natural – to increase and preserve quality and yields, drive circularity and better serve our changing needs. Soils and forests evolve slowly, which is in itself a challenge when it comes to future predictions, as we are building on data lacking long enough time series. However, the analysis can allow short term decision making in terms of sourcing, use and optimization of resources, land management and stakeholder engagement.

The impact on future generations as well as on short term business cases is well worth the work.

Air pollution in cities has to stay below certain limits for health reasons. Reliable forecasts of air quality offer the opportunity to control air pollution, e.g. by managing peak traffic flows.  This talk shows an approach to forecast air pollution within the city of Berlin for the next few days using xgboost. The talk will focus on the model and the challenges we faced during the modeling process. We will also have a glance at the tech stack we use (e.g., Clickhouse, kubernetes, Docker, Redash, FastAPI).

Harnessing the potential of predictive analytics, we revolutionize supply chains for a greener tomorrow. By leveraging data-driven insights, we optimize inventor, and distribution processes, minimizing waste and reducing carbon emissions. Predictive models empower us to foresee demand patterns, enhance resource allocation, and lowering energy consumption. This strategic approach not only enhances efficiency but also significantly mitigates the supply chain’s environmental footprint. The synergy of predictive analytics and sustainability transforms industries, showcasing how innovation aligns with responsible practices, ultimately fostering a positive impact on the climate.

In an era of heightened accountability for environmental and social impacts, comprehensive sustainability reporting is crucial. Currently, existing reporting methods are unreliable and a hindering process. Discover how our framework transforms the document querying, providing a quantifiable process with natural language processing and dynamic search term generation. Gain insights into the step-by-step process and uncover the benefits of enhanced document querying in driving meaningful change. Don’t miss this chance to shape a data-driven, sustainable future.