2020 AI Training Data Hackathon

Top Excellence Award

▲ (From left) Kim Gyu-jin and Son Hyun-soo, master's students, Department of Computer Science & Engineering

Kim Gyu-jin and Son Hyun-soo (supervisor: Professor Kim Ji-hwan), master's students in the University's Department of Computer Science & Engineering, received the Top Excellence Award at the AI Training Data Hackathon hosted by the NHN consortium on 29 January.

The competition formed part of the '2020 AI Training Data Construction (Phase 2) Project'. Teams of up to three were assessed on the recognition rate of the speech recognizers they built, using speech data collected from a range of age groups from children to the elderly.

The Sogang University team analysed the data provided by the organizers and, in order to reflect the characteristics of the linguistic expressions present in it, trained an advanced deep learning-based speech recognizer using the speech recognition toolkit 'Kaldi'.

Son Hyun-soo explained what prompted them to enter: "We wanted to verify objectively the performance of the speech recognizer we are researching in the Auditory Intelligence Laboratory." Kim Gyu-jin said: "The preparation period was only about a week, so we were short of time and could not be sure of a good result — I am delighted to have received the Top Excellence Award."

▲ (From left) Park Jin-woo, head of the NLP group at NHN Diquest; Kim Gyu-jin; Son Hyun-soo; Kang Rak-geun, CEO of NHN Diquest