Recently, the "Guangdong-Hong Kong-Macao Greater Bay Area (Whampoa) International Algorithm Calculation Competition" announced the final list. After fierce competition, Kuaishangtong AI laboratory kristonAI lab and the National University of Singapore doctoral team successfully advanced to the final of the "pre-trained language model application tuning algorithm" track.
In the digital economy era, algorithms have become the core support for promoting the deep integration of digital technology and the real economy. In order to comply with the current development trend of digital technology and actual application needs, Pazhou Laboratory (Whampoa) was entrusted by the Huangpu District Government of Guangzhou City to establish and host the first "Guangdong-Hong Kong-Macao Greater Bay Area (Whampoa) International Algorithm Calculation Competition".
This competition has the characteristics of "three highs", that is, high starting point, high bonus and high level of experts. The competition adopts a dual-track competition system, which is divided into two competition systems: competition system and arena system. There are ten very challenging competition questions. The competition organizer invites many famous academicians and experts and scholars in the field of artificial intelligence to form the competition expert committee. The theme direction, guiding the competition work, and standardizing the evaluation of competition questions provide strong support for the actual implementation of the competition. The competition focuses on next-generation digital technologies such as big data and artificial intelligence, and digital economic scenarios such as Internet+ and smart+, and gathers global high-tech technologies to promote original innovation and high-quality development of the digital economy. The "Pre-training Language Model Application Tuning Algorithm" track participated by kristonAI lab of Kuaishangtong AI laboratory and the doctoral team of the National University of Singapore belongs to the arena-based track. algorithm.
In recent years, the pre-trained language model, as a strategic barrier-level technology for AI companies around the world, has greatly promoted the development of natural language understanding, especially human-computer dialogue. However, due to the high operating costs of large-scale language models and the consideration of commercial profitability, the current mainstream is to implement the industry in the form of "Language-Model-as-a-Service" (Language-Model-as-a-Service, LMaaS). application. Therefore, the optimization of downstream tasks in the LMaas scenario has become the core technical challenge for the implementation of pre-trained large models. The pre-training language model, as the key research direction of Kuaishangtong kristonAI laboratory, has successfully advanced to the final in the "pre-training language model application tuning algorithm" track, which fully demonstrates the company's top technical strength and Influence. The latest breakthroughs in scientific research will be applied to the KICP man-machine dialogue system.
Kuaishangtong keeps pace with the rapid development of the digital economy, has been in the artificial intelligence industry for more than 13 years, and has achieved phased results in AI technology innovation and AI practical application implementation. The co-founder of Kuaishangtong, Professor Li Haizhou, academician of the Singapore Academy of Engineering, led the Kuaishangtong core R&D team composed of members from Stanford University, Cambridge University, National University of Singapore, Tsinghua University and other world-leading universities, mastered natural language processing (NLP), knowledge graph (KG), Speech Recognition (ASR), and Voiceprint Recognition (VPR) four core technologies, won the highest level certification in the global software field - CMMI5, China's highest artificial intelligence award - Wu Wenjun Artificial Intelligence Technology Progress Award and other top AI awards at home and abroad, Independent research and development of conversational AI drives industrial digital upgrade, which has been highly recognized by government departments.
Relying on independent research and development of core technologies, Kuaishangtong builds an AI PaaS open platform to open up more underlying algorithm capabilities and solutions to technical partners, empowering more than 1,000 development teams, and independently develops AI SaaS intelligent enterprise service product system, Solve the digitalization problems in medical beauty, education, home furnishing and other industries, promote the digital upgrade of the whole industry, and serve the high-quality development of the real economy. The AI SaaS intelligent enterprise service product system has realized the deep integration of AI technology innovation and AI real application implementation in the marketing application scenario, and built the first integrated enterprise marketing service for the whole life cycle of customers starting from enterprise marketing in the country, providing the whole industry with AI intelligent marketing solution. From "omni-channel access - intelligent reception - refined management - intelligent strategy optimization - private domain traffic operation" and other aspects, the whole link helps customers from marketing to operation, from customer acquisition to customer retention. At present, the Kuaishangtong AI SaaS full-link marketing solution already has important partners such as Dent Dental, Yonghe Zhifa, Xinhua Education, Gold Kitchen Cabinet, etc., and has profound technical accumulation and industry precipitation in digital technology innovation and application implementation.
In the future, Kuaishangtong will continue to use the two-wheel drive of "scientific and technological innovation + application implementation", and use original and leading scientific and technological achievements to accelerate the integration of technology and industry, form a "new momentum" for high-quality economic and social development, and promote industrial development. The integrated development of learning, research and application strives to provide strong support for the construction of digital China.