Boris Luis, Cuban student obtains scholarship for next Latin American Conference on Artificial Intelligence

March 5, 2024

Boris Luis Ramos, a fourth-year student in the Faculty of Physics at the University of La Habana, received the Grand Prize of the Event at the Hackathon of the 6th International Meeting on Artificial Intelligence and its Applications, held in Quito, Ecuador.

"I learned about this event because in March 2023 I attended the Latin American Conference on Artificial Intelligence (KHIPU) in Montevideo, Uruguay, and some of the speakers and sponsors were the same at both events. As soon as I found out about it, I applied online to present a poster and attend, and at the end of 2023 they notified me of my acceptance," he told Juventud Técnica.

For the competition, there was the option to form teams of up to four people, and Ramos was accompanied by three other young people from the region: Colombian Lina Montoya and Ecuadorians Ángel Yagual and Carlos Lincango; all with academic training in fields of artificial intelligence (AI), and data analysis and science.

"During the competition, you had to propose a solution using AI to a real problem and present a final product based on it. With this in mind, we considered that, once the problem was defined, we needed to look at it from as diverse perspectives as possible in order to reach the best solution to it," Ramos said. Coming from different disciplines and countries brought different perspectives to the problem and "each one contributed their grain of sand to enrich both the presentation and the final product that was presented."

Unlike other hackathons, where a problem to solve is normally posed, in this case to give more weight to creativity and data search, attendees were allowed to choose the situation and provide an answer through artificial intelligence.

Their work, titled "Prediction of river flows for hydroclimatological planning and management of associated risks," aims to contribute to the issuance of early warnings of problems such as droughts and floods, an aid for people commonly affected by them and which can generate economic benefits through the optimization of resources based on the alerts.

Ramos notes that his team, Hidro-Latinx Thinking, created an algorithm capable of predicting with good accuracy the flow of a river taking into account the climatological phenomena of El Niño and La Niña. For their project they based themselves on the case of the Magdalena River in Colombia, although he confirms that it can be applied to any other as long as the necessary data is available.

"A Deep Learning algorithm was proposed, which basically included an LSTM-type neural network model (Long short term memory), which is used to process and predict data sequences, and its performance was evaluated with RMSE, a metric commonly used to assess the precision of a regression model, comparing the model's predictions with the actual values of the test dataset.

"This model is an improvement over traditional recurrent neural networks given that they are equipped with a specialized memory cell structure that allows them to learn long-term dependencies, and also opens up the possibility for us to work with multivariate time series."

Regarding his experience of the collaborative environment of the event, he said that it is truly productive to meet and exchange with people from other countries and specialties. "That interaction generally makes it possible to create new contacts and collaborations, therefore it is extremely enriching and so was my experience at the Hackathon. Listening to the presentation of their projects and even talking in some cases about more personal experiences are things that bring new ideas and open your mind not only to new problems but also to new ways of facing them," said Ramos.

On his learning in Cuba and preparation for these events, he said that higher education in the country aims to have an integrative character. "In most cases, rather than teaching us to solve specific problems in our specialties, they equip us with the tools and with a way of thinking that allows us to address the most different problems that come our way," he said.

Although he did not have extensive training in programming or artificial intelligence, the degree places a strong emphasis on Scientific Work, and through it he was able to better prepare himself in these areas. "Similarly, the support of my mentor and the Faculty to be able to participate previously in events both national and international on mathematical modeling, data science and AI was decisive for me to properly prepare to solve the problem and obtain this Grand Prize at the Hackathon," added Ramos.

With this result, the Cuban and his companions obtained a scholarship and direct admission to attend the next Latin American Conference on Artificial Intelligence (KHIPU) which will be held during the first semester of 2025. The organization offers training in machine learning, deep learning and reinforcement learning through activities, talks and international-scale workshops.