ML for Climate Meetup - 5/12

ML for Climate Meetup (ML4C)

Join us for our 2nd ML4Climate Meetup and explore the ways data science and machine learning can be harnessed to tackle climate change!

Dataiku

How do we leverage data science and machine learning to tackle climate change?The ML4Climate Meetup, organized by the Climate Ambassador Program, aims to lower the barriers between the data science community and climate subject-matter experts.Mark your calendar for December 5th, 7:00 PM, at the Dataiku Paris Office, for the 2nd edition of the ML4Climate meetup. Join us to delve into three impactful data science applications for climate action:PROGRAMPyronear: Collective Innovation for Wildfire Mitigation by Mateo Lostanlen - Co-Founder at PyronearWildfires are a global challenge that demand a proactive and collaborative approach to prevention, detection, and management. Pyronear is at the forefront of this battle, leveraging the power of open-source technology and community-driven development to mitigate the impact of forest fires. This presentation will delve into Pyronear's mission, the technological solutions we provide, and our unique model for building a community focused on reducing wildfire impact through collective intelligence and shared innovation.We will explore the core aspects of Pyronear's approach, which includes the development of low-tech, high-efficiency early detection systems utilizing cutting-edge artificial intelligence—all built and improved upon by a dedicated network of volunteers. The talk will emphasize how Pyronear's open-source ethos not only democratizes access to fire detection technology but also fosters a strong, collaborative community that spans firefighters, technologists, environmentalists, and local stakeholders.Participants will gain insights into how Pyronear's modular and energy-conscious design philosophy ensures ease of integration and operation, significantly aiding rapid deployment across varied landscapes. We will discuss the strategic placement of high-resolution cameras and micro-computers, the local processing of data to trigger early alerts, and the seamless communication protocols that link our detection system to firefighting authorities.Further, the presentation will outline the steps we have taken to cultivate a community dedicated to this cause. From hackathons to collaborative R&amp;D, and volunteer-driven initiatives, Pyronear is not just a technological solution but a movement that encourages community engagement at every level.Join us to understand how Pyronear is revolutionizing wildfire detection and management, and how you can be part of this transformative journey towards a safer, more resilient ecosystem.Reliably mapping the rooftop photovoltaic installations in France to support the decarbonation of the electric grid by Gabriel Kasmi - PhD Candidate at RTEIn 2022, France's photovoltaic (PV) installed capacity stood at 15.7 GWp, making it the country's third-largest source of renewable electricity. It represents year-on-year growth of 19.8%. However, the French electricity transmission system operator (TSO) has no measurements of the PV power production of a fifth of the installed capacity, which in a context of sustained growth could threaten the stability of the grid. To overcome this issue, the TSO needs to precisely know the localization and the characteristics of the unmetered PV systems. As these installations correspond to individual installations, accurate, comprehensive and up-to-date registries are lacking. To fill this gap, detecting and characterizing rooftop PV installations on overhead imagery using deep neural networks emerged as a promising issue. Yet, current methods are not reliable enough to be used for the constitution of such important datasets. This talk will present the method deployed by the French TSO to improve the reliability of deep learning-based methods for mapping rooftop PV installations. This method rests on three pillars: monitoring the accuracy of the PV registry, auditing the decision process of the model and improving the robustness of the model to acquisition conditions. To monitor the model’s outputs, we introduce the downstream task accuracy, which compares aggregated results with existing databases. To audit the model’s decision process we introduce the wavelet scale attribution method (WCAM), an explainability method that highlights the important scales, which are interpretable in terms of structural components, for the prediction. Finally, based on the WCAM, we introduce a new data augmentation strategy that outperforms existing robustification methods to improve the robustness of the model against varying acquisition conditions, which play a key role in the poor reliability of current models.Empowering Sustainability: Electric Power disaggregation for energy efficiency and CO2 emission reduction by Ghassene Jebali and Théo Delfolie - Research Engineers at Smart ImpulseClimate change is a key challenge of today's world. The pressing need to reduce CO2 emissions from electricity generation has become a paramount concern since it is responsible for 41% of the total emissions worldwide in 2020. To tackle energy efficiency challenges and reduce energy consumption, we, at Smart Impulse, aim at solving the energy disaggregation problem (also known as Non Intrusive Load Monitoring, NILM). Our goal is to enable energy managers to have a clear view of how their buildings are run, target energy savings and have enhanced monitoring over the long-term. In this talk, we will expose the foundation of energy disaggregation and present the research undertaken at Smart Impulse in order to solve the key challenges of NILM. These talks will be followed by a cocktail and a networking time! This event is co-organized and sponsored by the Dataiku Climate Ambassador Program.The Climate Ambassador Program (CAP) encourages Dataikers to lead and contribute to projects supporting Dataiku’s Climate Ambition.Since the beginning of the program, 60+ Ambassadors have been contributing to internal projects to accelerate decarbonization, bring awareness about climate, and contribute to AI for climate initiatives.

ML for Climate Meetup - 5/12

More events from ML for Climate Meetup (ML4C)

Discover more events from ML for Climate Meetup (ML4C), from Science & Tech to other experiences you might love.

Still looking for the right event?

Explore all events in Paris and filter by date, category, and more to find the perfect fit.

ML for Climate Meetup - 5/12

More events from ML for Climate Meetup (ML4C)

Discover more events from ML for Climate Meetup (ML4C), from Science & Tech to other experiences you might love.

You might also like...

Browse more events with different dates, prices, and formats to find your next great experience.

Still looking for the right event?

Explore all events in Paris and filter by date, category, and more to find the perfect fit.