Artificial Intelligence (AI) was highly popular in the 1980’s but because of undelivered promises, it went dormant for over 30 years. However, thanks to ground-breaking recent research in deep learning, AI is making a huge come back, and its use in the industry is spreading rapidly. Almost all global corporations now are investing in and developing AI technologies and AI platforms, to enhance their products and services, with the ultimate goal of providing new user experience.
New technologies invented and developed at Samsung Research are already being applied to our products and services, providing new AI-driven experience and value to our customers. Furthermore, we are conducting investigations in the frontiers of AI research, to open up new business opportunities and to contribute to the society.
Specifically, we strive to secure competitive AI solutions such as Virtual Personal Assistants (VPA), Visual Understanding, and On-Device AI. For this, we are developing new methods for natural language and speech processing, visual object recognition, image processing, and inference and prediction. We are also building an AI platform to incorporate all of the above (and more) in an integrated development/service platform.
Virtual Personal Assistants (VPA): We are developing VPA technologies to enable natural and accurate conversation between the user and our conversational agent. The goal of this project is to allow our agent to understand human dialogue through voice recognition, carry on natural conversation, answer questions, translate and interpret various languages, and generate realistic speech through voice synthesis.
Visual Understanding: We are developing visual understanding technologies to extract meaningful descriptions from images, with the goal of providing users with timely and accurate information, and to allow natural user interaction. We are developing (1) deep-learning-based image understanding methods for use in our products and services, which can recognize large number of objects (on the order of 1,000 objects) and understand the semantic content of the image, and (2) robot vision technology to allow the robot to monitor its surroundings and the human users, and to allow the robot to navigate freely and safely in complex in-door environments.
On-Device AI: We are developing a high-efficiency AI engine to enable AI services to run in embedded system environments, where resources such as memory, computing power, and energy supply are severely limited. Our focus in this area is software/hardware codesign and optimization to support various applications, AI processors, and SDK for the integrated software/hardware system. Our efforts are expected to enable the development of new personal, wearable devices that can run high-end AI algorithms and services.