The global race for all things that involve artificial intelligence (AI) is at a fever pitch, and the business stakes are high. In the near future, AI technology will likely become a way of life for many people, with daily interactions with a smart speaker or a personalized AI that provides solutions derived from a user’s visual, tactile, and emotional cues. AI will also become the most potent enabler of competitive advantage throughout most areas of society and economy.
From a wide-range of consumer electronics to 5G equipment to powerful chipsets, Samsung is uniquely positioned to bring this world of connected AI services to life, based on the almost half a billion devices that the company sells every year.
In addition, for the past decades, Samsung has successfully introduced products and innovations by researching the lifestyle and behavior of its global consumers. In this aspect, there is no technology company better suited than Samsung to shape a new era of connected living that touches the lives of consumers and the entire society through AI. The AI that Samsung pursues is available anytime, anywhere and is continuously evolving to be always helpful, with the ultimate goal of value creation. We will begin an exciting journey open to boundless possibilities in new user experiences by integrating AI into the open IoT ecosystem that we are currently developing.
To achieve this vision, Samsung Research is developing next-generation AI technologies to be applied to all its products and services. Moreover, the primary goal is to secure cutting-edge AI core technologies and platforms - human-level AI with the ability to speak, recognize, and think - to provide new AI-driven experiences and value to its customers. Aligned with our goal above, we are conducting research in broad thematic areas such as AI core algorithms, on-device AI, next-generation virtual assistant platform, and so on.
Current deep learning algorithms, although extremely powerful, are limited in several ways. These algorithms require large amounts of data, struggle with learning from a continuous inflow of data, and cannot easily handle multimodal data. To address these issues, Samsung Research is developing new methods at the intersection of vision, language, and machine learning, with the goal of securing the world’s best core AI algorithms.
We are looking into approaches such as Bayesian Deep Learning for uncertainty modeling and probabilistic inference, and Reinforcement Learning to enhance system performance based on user feedback and for autonomous control. We are also working on Automated Machine Learning (AutoML) to automate model search and performance tuning, and Generative Adversarial Networks (GANs) to generate realistic scenes and to augment data.
Most of the current AI technologies depend on cloud services. However, to reduce latency, ensure privacy, and provide AI services anytime, anywhere, on-device AI capability is needed. On-device AI has many challenges such as meeting power constraints, on-device learning requirements and model reduction and optimization. In addition, researchers need software development kits (SDKs), and tool chains for on-device AI development and deployment. New algorithms and tools to meet these challenges are therefore continuously being developed.
For this, Samsung Research is researching technology for designing neural processors for hardware acceleration, making models light enough to be applied to all Samsung products, and for software acceleration.
Digital assistants have recently been more widely applied to various products and machines such as smart speakers, smart displays, and home robots. Further, multimodal assistants followed by proactive assistants are forecast to evolve in the future from the current voice-based assistants. Samsung Research is developing a comprehensive virtual assistant platform as a highly personalized multi-device ecosystem to transform and empower people’s daily lives. Its plan is to realize a “virtual assistant that optimally works, smartly understands, and interacts like a friend in various products anytime, anywhere”.
The goal of this project is to allow the agents to understand human dialog through voice recognition, to carry on natural conversations, to answer questions, to translate and interpret various languages, and to generate realistic speech through voice synthesis. As a result, users will get to enjoy the seamless provision of desired services at ease with any device in all aspects of daily lives.
Samsung Research has established a total of seven global AI centers in five countries, namely Korea (Seoul), the United Kingdom (Cambridge), Canada (Toronto and Montreal), Russia (Moscow), and the United States (Silicon Valley and New York). They will contribute to Samsung’s AI research with their unique regional strengths, and the AI technologies developed from these centers are being adopted in Samsung products and services. Samsung Research is expected to expand its AI centers to other technology- and talent-rich areas to cement its place as the leader of AI research in the industry.View More
Samsung AI Forum 2018VIEW MORE -