 |
Европейская академия
|
| В начало | Энциклопедия ЕАЕН | |
Энциклопедия Европейской академии естественных наук |
||
Lu Zhengdong
Dr. Lu Zhengdong has long been engaged in the research of machine learning and artificial intelligence, and has made remarkable achievements in the fields of deep learning, natural language processing and semi-supervised learning. He is a world-leading scientist and technical expert in the field of deep learning (especially in the direction of natural language processing) and enjoys an international reputation.
Dr. Lu Zhengdong founded the deep learning team of Huawei Noah's Ark Lab from scratch in early 2013, and led Noah's Ark Lab to become an international first-class research institution in the field of neuro-linguistic intelligence within two years, with research level and achievements comparable to DeepMind, Google Brain and other international top teams. Dr. Lu Zhengdong has published more than 50 papers in many top conferences and journals of machine learning, data mining and natural language processing, such as NeurIPS, ICML, ACL, KDD, JMLR, etc., which have been cited more than 10,000 times, and three papers have been cited more than 1,000 times. Of the top 10 highly cited papers from the ACL (Top Conference on Natural Language Processing) over the past five years, two came from Dr. Lu's team. Dr. Lu's work has also been among the top 30 highly cited papers at NeurIPS (the top machine learning conference) for the past five years. Of the ten works in Greater China cited in the 2017 authoritative review of neuro-linguistic Intelligence in the Journal of Artificial Intelligence, four of the contributions of Dr. Lu and his team were highly rated. Dr. Lu Zhengdong is the inventor of a number of deep learning based natural language processing technology patents, a total of 18 patents, covering semantic matching, question and answer, multi-round conversation and automatic message reply.
From 2013 to 2014, Dr. Lu Zhengdong created the research direction of deep semantic matching with a series of work at NeurIPS conferences. The interaction-based deep semantic matching framework proposed has become one of the two main methods of semantic matching, and has become the basic model for questions and answers, dialogue and other directions. At the same time, in this series of work, Dr. Lu was the first to use convolutional neural networks (CNNS) for semantic representation, and is one of the pioneers of CNNS for natural language processing.
In 2013, Dr. Lu proposed a search-based automatic conversation technology that is now the basic technical framework for mainstream chatbots (such as Xiaoice) and question-and-answer robots.
In 2015, Dr. Lu Zhengdong proposed a reasoning model based on deep learning, Neural Reasoner, which solved two of the most difficult reasoning tasks out of 20 proposed by Facebook AI Research in 2014. He was invited to give a lecture at NIPS RAM Workshop, a top deep learning conference, and his paper was named one of the top ten arXiv papers in 2015.
Dr. Lu Zhengdong's team invented the world's first chatGPT type of generative chatbot "Neural Responding Machine" based entirely on deep learning, which caused wide response and is considered to be a pioneering work in generative dialogue.
In 2016, Dr. Lu Zhengdong proposed a question-answering system for all-neural networks called Neural Enquirer. His paper was published at the same time as similar work of Google Brain and Facebook AI Research and attracted wide attention, becoming an important work of neural symbolic intelligence.
In 2016, Dr. Lu Zhengdong proposed the copy mechanism of sequence generation and the corresponding model CopyNet, which has been widely recognized by the academia and the industry, and has become a general technology for dialogue systems, automatic summarization and machine translation.
Dr. Lu's contributions to machine translation also include his proposed coverage-based attention mechanism, which is now a foundational component of neural machine translation with automatic summarization and is adopted by Google's commercial translation model.
Dr. Lu Zhengdong, an internationally leading researcher in neurosymbolic intelligence, proposed a new framework for the deep integration of distributed representation and symbolic representation, which has achieved significant improvement in the question answering system with complex text understanding, dialogue understanding and reasoning, becoming the first successful commercial neurosymbolic system and attracting wide attention.
| Энциклопедия ЕАЕН |