[00:00:00] Host Create clip in June off this year, 22 Ai Ai researchers published a paper on how climate change can be tackled with machine learning. They collected hundreds of research works and projects that offer climate change solutions. They all aim to reduce greenhouse gas emissions either in electricity systems, transportation, buildings and cities, farms and forests or in industry. In this video, I would like to give a quick overview by outlining promising solutions from each category to learn more. Have a look at the block post serious links in the description below where I created the list with solutions. The respective machine learning technique used and links through relevant papers and projects. So let's get started with electricity systems. They are responsible for 1/4 off human costs. Greenhouse gas emissions Machine learning can help in many ways to reduce their carbon footprint. Let's have a look at three solutions regarding energy forecasts. Methane leaks and the discovery off new materials energy for cars since renewable energy production, various depending on conditions and solar radiation. Energy grid operators use polluting sent by plants which can feed electricity into the grit when unexpected production shorter's occur by applying machine learning to better forecast how much power will be generated by renewable sources and also how much power demand there is. Their lines on polluting standby plants can be reduced. Also, long term forecasts help operators understand where and how much new renewable plants should be built. In the case off solar energy grid operators can make better predictions if they know how much solar panel capacity exists in their operating region. Using satellite data, machine learning can detect rooftop solar panels and inform them about estimated solar capacity. Furthermore, using historical data about power generation off panels, machine learning can predict false, such as the defect for solar module and inform affected customers.

[00:01:56] Host Create clip Methane leaks, some off the fossil fuel gets lost during transport, for example, due to leaks and pipelines. In the case off methane it's released into the atmosphere should be avoided since it's a much more potent greenhouse gas than seo, too. On the basis off data from on site sensors and satellite sensors, methane leaks in natural gas pipelines can be detected and for the more pipeline maintenance can be predicted. Machine learning can use the data from these sensors to learn the spectral signature off methane in sunlight that is reflected off the crowd. Their discovery off new materials process off. Discovering new materials can be slow and imprecise. Machine learning can accelerate the discovery by automating parts off the process as an example, The research off better battery storage technologies can be accelerated by predicting the connectivity of various compositions based on experiments with data. So what about transportation? Transportation is responsible for around 1/4 off. Global Energy related Co. Two Emissions passenger and freight transportation are each responsible for about half transport greenhouse gas emissions. 2/3 of emissions are cost by road travel, but emissions off air travel are on the rise. Machine running can help in numerous ways. Let's have a look at five solutions regarding transportation demand modeling, shared mobility concepts, bike sharing, autonomous vehicles and electric vehicles.

[00:03:27] Host Create clip Transportation demand By understanding transportation patterns and modeling transportation demand, city planners can make smarter decisions on how to encourage low carbon modes of transport in the case off public transport. But she's learning can make use of smart card data or online booking data to discover behavior patterns and determined demand. Shared mobility. Shared mobility concepts are on the rice, but we don't yet know if shed mobility will actually lead to lower greenhouse gas emissions. In the long run, for example, somebody starts to use car sharing instead of using public transport. The energy impact in this scenario is negative. Machine learning can help to predict switching behavior between transport, Moz bike sharing, low carbon modes of transport. Such a spike sharing often suffer from a re balancing problem. This happens when, over the course of a period by cycles, accumulate in certain areas and completely leg in others. Making three service is less attractive. ML can predict demand an inventory enabling smart pricing mechanisms to set incentives for a more balanced bike distribution autonomous vehicles. These days, there's a big hype around autonomous vehicles, though it is yet unclear if they will lead to less or more traffic. There's a few scenarios where there is potential for positive impact. ML could, for example, improve track platooning where multiple trucks can drive very close to each other since they can break and accelerate simultaneously. This would lead to less air resistance between the trucks and therefore lower fuel consumption electric vehicles. Electric vehicles are considered to be the main solution for Dick Organizing transport batteries, hydrogen fuel cells and electrified railways and roads are all considered electric vehicle technology. These technologies can have very low greenhouse gas emissions, of course, only if the electricity they consume is produced from mostly renewables. ML can help predict a battery st degradation and remaining lifetime. Furthermore, it can also forecast Van Electric vehicles will be charged and informed grid operators about the electric load they should expect.

[00:05:41] Host Create clip So let's talk about buildings and cities. They are responsible for about 1/4 off global energy related greenhouse gas emissions, and there's huge potential in reducing emissions in existing and new buildings. Let's have a look at three solutions regarding hot water and cooling systems, the energy impact off cities and smart city projects. Hot water, including systems the majority off energy consumed by buildings was caused by heating, ventilation and air conditioning. ML could enable buildings to reduce the energy consumption of hot water and cooling systems. For the more it could recognize leakage off refrigerants from devices such as air conditioners, which can have a big impact. Refrigerants often contain a gas called HFC, which one released into the atmosphere has up to a 9000 times greater capacity to warm the atmosphere compared to carbon dioxide. Although many countries plan to face out H of sea within the next decade and replace it with less damaging substitutes, devices containing it might be around for longer. Applying fall detection techniques can help better detect each of C leeks, the energy impact off cities. Local governments hook in shape cities and past the mission. Reducing regulation often create policies without sufficient data about their city. They could benefit from a mission relevant data about their city and buildings in order to make smarter decisions. ML can help to estimate the energy impact off cities where there is little information. Models can use satellite imagery to classify all buildings regarding their estimated energy use.

[00:07:14] Host Create clip Smart city Projects Smart City projects which proposed climate friendly solutions in areas such as renovation, heating, waste management and mobility, are mostly conducted in the global North. However, the highest potential off climate change mitigation is in the global South American cluster cities based on climate relevant factors. This makes it easier to identify cities in the global self that have similar characteristics and therefore might be ideal candidates for the replication off a proven climate solution. Let's now have a look at farms and forests. Overall worldwide land use is estimated to be responsible for about 1/4 of greenhouse gas emissions. Melting permafrost in the Arctic is expected to add talk to 17% of global emissions within the next decades. For the more an increase of forest fires releases sequester carbon as well. According to Project Drawdown, a research organization that reviews, analyzes and identifies the most viable global climate solutions, better land management and more efficient agriculture could achieve 1/3 off all potential greenhouse gas reductions. Although plants take up sea or two from the air, there are three major emission causes. First, cutting off trees, second tilling off soil and third fertilizers. Leasing nitrous Excite a gas, which is 300 times more potent than seo, too. Machine learning can help in a few ways. Let's have a look. A Three solutions regarding agriculture, forests and peatlands.

[00:08:44] Host Create clip Agriculture Agriculture alone is responsible for about 14% off greenhouse gas emissions. Machine learning can predict crop yield and detect plant diseases and thereby make agriculture more efficient. As an example, there's a project where a robot on wheels with a camera automatically scans letters and outputs estimates about current crop yield and plant ceases. The robot can also reduce the amount of fertilizers needed by only spraying chemicals on appropriate spots instead off the whole field, reducing the release off nitrous excite forests. Disappearing forests are a problem, and there are two major causes. Forest fires and deforestation through logging. Forest fires obviously released current into the atmosphere and reduce the number of trees that can sequester carbon. In the future, ML can help forecast droughts, which helps locate forests at risk and also predict in which directions and speed corn fires are spreading. The legal logging, on the other hand, can also be checkered with ML. Analyzing satellite imagery is one way it can be detected, but there are other solutions. For example, the Rainforest Connexion project attach is old phones to trees and rainforests. Once attached, they constantly listen for chainsaw sounds. When chainsaws are detected, push notifications are sent to local officials. Um, l can improve the detection off chainsaw sounds, increasing the chances off catching illegal loggers peatlands. When you think about climate protection, you probably don't think about peatlands who cover only 3% off Earth's land area. But peatlands contained tries as much carbon as all the world's forests combined. A single peat fire in Indonesia in 1997 costs between 20% and 50% off all the missions off that same year.

[00:10:31] Host Create clip ML can help protect peatlands by estimating their thickness and deriving how much carbon certain peatlands contained. Also risks off fire in peatland regions can be better predicted using him out. Let's have a look it in this tree. Greenhouse gas emissions caused by industry often hard to eliminate. However, industries collect more data than ever, and cloud storage and computing is becoming more affordable. It is estimated that 60 to 70% of industry data it's not used, though a challenge is to get access to it. Since most of it is proprietary. Let's have a look at two solutions regarding cement and sea of production and data centers, cement and steel production, cement and steel production cost staggering 9% off all global greenhouse gas emissions. If the cement industry would be a country, it would emit more greenhouse gases than every other country except China and the U. S. Three d printing allows for unusual shapes that use less material but may be impossible to produce throughput concrete or traditional metal casting ML can have advanced generative designed algorithms, which reduces the cement and steel acquired to form a product while maintaining the products. Stability Data Center's data centers need to be cooled constantly, and since there are an increasing amount of data centers worldwide, their energy impact is expected to rise. Deepmind WAAS able to reduce Google's data centers cooling energy consumption by 40% using machine running.

[00:12:03] Host Create clip Until now, we have looked at many solutions that lower emissions. But even if no CO two would be admitted at all anymore, today's level of carbon in the atmosphere will continue to have negative consequences in the future. The natural removal off go to buy absorption through plants, for example, won't suffice. That's why I actively removing 02 from the atmosphere and storing it in some other form. It's necessary. This is also known as carbon dioxide removal. According to researchers, the most promising approach is to sequester carbon dioxide, which can be done by capturing it and storing it underground. As an example, a Norwegian oil company has successfully sequestered, see or two from an offshore natural gas field on the ground for more than 20 years. ML can help identify and characterize potential storage locations for the more it can help monitor and maintain actives. Administration sides and help with simulations. If you enjoyed the video and want to learn more about these solutions, have a look at the block post serious linked in a description below where I created a list with many more solutions and collected links to relevant papers and projects.

[00:13:10] Host Create clip My name is polished over Thanks for watching and see you next time.