Post sponsored by NewzEngine.com

MIL OSI Translation. Region: Russian Federation –

In Stavropol, the third artificial intelligence hackathon (hacks-ai.ru) for solving the problems of the Russian Emergencies Ministry, which took place from September 3 to 5, was completed.

Within 48 hours, 12 teams gathered from all over the country solved the case from the Ministry of Emergencies of Russia, Roshydromet and Rosleskhoz. 9 teams reached the final. They were developing an algorithm to prevent forest fires at an early stage and to predict the average daily increase in the area of ​​a large landscape fire.

The winners of the third artificial intelligence hackathon are the following teams:

1. Profit (St. Petersburg, Moscow, Stavropol)

Team members: Pavel Alekseev, Tatiana Litovchenko, Egor Pustovit.

The team developed an algorithm to prevent fires at an early stage. Based on this algorithm, data (meteo, geography, surface type – water, forest, stones, vegetation) are loaded for each coordinate of the cellular automaton (1×1 km) and predictions of the spread of fire are performed. We used machine learning algorithms lightgbm, catboost, KDTree.

2. AIForestFire (Stavropol Territory)

Team members: Vladimir Antonov, Nikolay Svistunov, Igor Struchkov, Mikhail Gurchinsky, Sergey Ryabtsev.

The solution that the team proposed is to predict not only the total area of ​​the fire, but also to determine its exact shape. By converting polygons into specific images, in the future, it will be possible to predict the behavior of the fire in the future from the picture of the current state of the fire. The benefit of this approach is that you can add an unlimited number of new features without changing the principle of the model.

3. “There is a way out” (Stavropol Territory)

Team members: Vladimir Shalnev, Mikhail Dukhno, Sofya Grobova.

The team’s solution is a neural network that receives a line of coordinates as input and displays the current plan of the area based on them, then calculates the further path of fire propagation. The work used TensorFlow, Pandas, Geopandas, Numpy.

“I would like to express my gratitude to the organizers of the event for the opportunity to participate as a case holder and to look at our tasks and ways of solving them from a different angle, as well as for the development of the topic of data science and artificial intelligence in Russia. And, of course, many thanks to all participants in the competition for a detailed study of the subject area, deep analysis of data and dependencies, and fresh ideas, which we will definitely take into our own development and verification. I would like to wish the participants to ask more questions and argue with the case holders, substantiating their hypotheses with data analysis, ”said Alexander Goryunov, head of the digital development department of the Information and Analytical Center of the EMERCOM of Russia.

The event was held within the framework of the federal project “Artificial Intelligence”. Hackathons and lectures on artificial intelligence – a series of educational events, in the framework of which, in the period from 2021 to 2024. It is planned to hold 116 hackathons on artificial intelligence in all regions of Russia.

EDITOR’S NOTE: This article is a translation. Apologies should the grammar and / or sentence structure not be perfect.

MIL OSI Russia News