Kakao Introduces AI Voice, Vision, and Knowledge Graph Sessions at @if(kakao)2021 Conference

Kakao Enterprise and Kakao Brain have unveiled AI-related technical presentation sessions at the @if(kakao)2021 conference. This event can be viewed in the Day2 tab for developers, and limited-edition emoticons are given to those who pre-register after signing up on the conference website.

Key Technical Sessions

The presentations cover various fields, including speech synthesis, computer vision, knowledge graphs, and MLOps platforms.

• Voice and Digital Humans: Kakao Enterprise introduces acoustic models and vocoders for natural speech, as well as style transfer techniques. Additionally, digital human technology based on Neural rendering and a Nickface generation model utilizing GANs will be presented.
• Vision and OCR: Tips for transfer learning on large-scale image datasets, issues in OCR service operations and model restructuring cases, and visual representation algorithms that learn without labels will be shared. Real-time interaction prediction technology for people and objects in images is also included.
• Knowledge Graphs: Industrial application cases are presented alongside the structure and solutions of Kakao Enterprise's knowledge graph platform. It also covers meaning-based learning algorithms that convert image content into graphs for search and question answering.
• Infrastructure and MLOps: Experiences in building a Kubernetes-based GPU sharing inference platform and case studies of problem-solving through large-scale deep learning will be presented. This session is targeted at developers interested in developing AI service serving platforms.