Kakao Brain Publishes 9 AI-Related Papers at Global Conferences

[Asia Economy Reporter Buaeri] Kakao Brain announced on the 28th that it published a total of 9 AI-related papers at global conferences in the first half of this year.

In February, Kakao Brain published 3 papers at the AI conference 'AAAI 2021.'

One paper, co-researched with Pohang University of Science and Technology (POSTECH), proposed a method that significantly improves anomaly detection performance in the VQA (Visual Question Answering) task, which simultaneously understands text and images and provides answers.

Another paper, co-researched with Seoul National University (SNU), proposed a method to solve content-based image retrieval problems.

The remaining paper, co-researched with the AI research institute Mila, proposed a module that allows AI to combine and reason about visually learned concepts autonomously, enhancing video recognition performance.

At AAAI 2021, an additional paper co-researched with Seoul National University was published in a separate workshop program.

In March, a paper co-researched with Seoul National University was published in the international electronics journal 'IEEE Access.' It disclosed a deep learning model and related training dataset that can compare behaviors of different people to measure similarity.

Kakao Brain also achieved results in technology research applicable to various fields such as medical and natural sciences.

In March, it participated in a competition hosted by the journal 'Medical Image Analysis,' presenting a paper proposing an AI model capable of diagnosing lung nodules and won first place overall.

In the natural science journal 'Science Bulletin,' it published a paper developed jointly with Seoul National University and Chonnam National University on a model that can predict atmospheric pressure oscillation phenomena in advance.

In June, two papers were published at the academic conference 'CVPR 2021.' One paper, co-researched with Korea University, proposed an algorithm that automatically detects interactions between people and objects in images.

Kakao Brain stated, "We focus on areas where research is insufficient and aim to achieve groundbreaking results beyond existing studies," adding, "We will continue to produce steady results in fields such as video understanding and digital humans to contribute to a better world."