With the deepening of intelligent products in production and life, the demand for intelligent dialogue applications has exploded. Intelligent voice assistants such as Xiaobu, Xiaodu, Xiaoai, Tmall Genie, etc. are popular with end users; products such as intelligent customer service are the must-choice of enterprises today. marketing tool. Today's diverse needs of enterprises and users also pose many challenges to intelligent dialogue technology, such as open domain multi-round, multi-modal fusion, anthropomorphic emotion and so on.
Therefore, in recent years, the industry has successively proposed concepts such as "dialogue as a service" and "dialogue as a platform", and more and more domestic and foreign manufacturers have begun to deeply invest in the exploration and development of intelligent dialogue technology. Including the recently popular ChatGPT, it is inseparable from intelligent dialogue technology.
However, intelligent dialogue is a field that requires a high level of technology, and has relatively high development requirements for data quality and processing efficiency. On the other hand, it is relatively difficult to reuse and migrate many existing dialogue logics. Developers are faced with cumbersome workload and high development costs, which indirectly increases the cost of using smart dialogue.
In the face of these problems, what are the current solutions in academia and industry? What explorations have you made in the research and development of "smart dialogue" related technologies?
In order to empower more developers who pay attention to this field, OPPO invited a number of technical experts from academia and industry to plan content around intelligent dialogue, and it is scheduled to land in Beijing COFCO Land Plaza on the afternoon of March 11 , 2023 The OGeek Xiaobu Salon, with the theme of "Talking about "Smart Dialogue" and Initiating the "Future of Interaction" together , discussed the evolution direction of smart dialogue with many developers and explored the wonderful future of human-computer interaction.
Many developers have paid attention to the 2022 OGeek Technology Summit held online last year. After the summit, InfoQ also made continuous reports on the exciting content of the meeting, so they are no strangers to OGeek, a technology salon brand. "OGeek" is an industry technology salon brand sponsored by OPPO Digital Intelligence Engineering Division, aiming to build an open platform for technology exchange and sharing for technology enthusiasts. The salon mainly revolves around the technical concept of "science and technology for people and good for the world", focusing on providing safe and efficient cutting-edge technologies in data, computing power, algorithms, and cloud services for smart terminals, creating an industry ecology of technology interaction, and exploring the application of technology in the industry practice, breakthrough and future development direction.
This OGeek Xiaobu Salon is even more exciting, and developers must not miss it. Salon leaders gathered, Dr. Yang Zhenyu, Chief Researcher of OPPO Xiaobu Assistant, as the content producer of this salon, invited Dr. Huang Minlie, Associate Professor of the Department of Computer Science and Technology of Tsinghua University, Baidu AI Chief R&D Architect & Technical Director of Xiaodu Algorithm Team Dr. Xie Jian and Dr. Suo Hongbin, algorithm expert of OPPO Xiaobu's assistant; the content of the salon is hard-core throughout. In addition to discussing the technological evolution trend of ChatGPT from the perspective of intelligent dialogue, the sharing topics will also be fully covered from academic research to enterprise technology practice. Provide a feast of technology gluttony to the developers present.
The following is the content arrangement of this OGeek Xiaobu Salon:
Biography of the speaker: Dr. Yang Zhenyu, chief researcher of Xiaobu Assistant of OPPO, senior member of CCF, has taught in universities such as National University of Defense Technology, and his representative academic papers have been cited more than 800 times, and he was selected as ESI Top 0.1%. In recent years, he has been mainly engaged in the research and application of natural language processing and conversational AI-related algorithms. In 2018, he joined OPPO to lead NLP and dialogue-related work, helping the company's strategic product Xiaobu Assistant to achieve a breakthrough of 140 million monthly active users.
Biography of the speaker: Dr. Huang Minlie, associate professor of the Department of Computer Science and Technology, Tsinghua University, winner of the National Science Fund for Distinguished Young Scholars, deputy director of the Laboratory of Intelligent Technology and Systems. The main research directions are natural language generation, dialogue system, reading comprehension, etc. He has won the first prize of the Wu Wenjun Artificial Intelligence Science and Technology Progress Award of the Chinese Association for Artificial Intelligence (the first completer), the Hanwang Youth Innovation Award of the Chinese Information Society, and the Alibaba Innovation Cooperation Research Award. Published more than 100 papers in top international conferences or journals, obtained more than 10 patent authorizations, won 5 best papers or nominations in international mainstream conferences (IJCAI, ACL, SIGDIAL, etc.), and authored the first domestic book on natural language generation Book "Modern Natural Language Generation". Research and develop dialogue system platforms ConvLab and ConvLab2, the first Chinese open-domain dialogue pre-training model CDial-GPT, Chinese open-source dialogue model EVA, and the first emotional dialogue robot Emohaa.
演讲嘉宾履历简介：谢剑博士，百度 AI 主任研发架构师、小度算法团队技术负责人。持续深耕自然语言处理、搜索、推荐、广告、对话等大规模 AI 系统十余年，曾负责过百度凤巢、商业知心、多模搜索等系统的核心算法和产品。发表国内外专利三十余篇，EMNLP/AAAI/CIKM 等学术论文多篇。带领团队研发出家喻户晓的小度助手，相关技术获得国际权威对话比赛 DSTC10 冠军、EMNLP 2022 SereTOD 亚军等诸多荣誉。
演讲嘉宾履历简介：索宏彬博士，OPPO 小布助手算法专家，CCF 会员，曾在中国科学院、联想研究院、阿里达摩院工作，学术论文发表超过 40 篇，语音技术专利 30 项。近年来主要从事个性化语音交互、语音唤醒、语音识别和副语言属性识别算法研究与落地应用工作，2022 年加入 OPPO 主导语音技术研发工作，帮助公司级战略产品小布助手实现月交互突破 30 亿次。
届时来自学术界与工业界的多位博士将一起探讨以下话题：
关于智能对话技术的研究与探索，目前学术界和工业界的侧重点分别是什么？
当前 B 端企业和 C 端用户对于“智能对话”产品的核心需求分别有哪些？
目前智能对话领域最大的“技术挑战”是什么？如何应对这个挑战？
XR 等硬件设备的发展和元宇宙概念的火爆，将对智能对话等“人机交互”领域产生怎样的影响？
在你们看来，下一代“人机交互”的理想形态是怎么样的？“智能对话”未来在里面会扮演怎样的角色？
ChatGPT 的火爆袭来，对智能对话技术的发展将带来怎样的影响？
如果只选一个，未来 2～3 年内，您觉得智能对话等人工交互领域最有前景的方向是什么？为什么？
智能对话的实现逻辑，简单来说，就是依赖知识图谱对输入的文字进行实体和关系等语义的识别与理解，通过深度学习（包括各种序列学习）的框架得到候选输出、通过推理来做最后回答的排序和过滤来实现最后的输出，而技术专家们就是基于此进行了技术探索和迭代。
当下的 ChatGPT 不仅能够通过学习人类的语言来进行对话，还能根据聊天的上下文进行互动，这是智能对话技术向前发展的一大步，在 ChatGPT 爆火的当下，我们非常有必要去全面地探讨下该技术。所以本次 OGeek 小布沙龙，一定不能错过！
届时，在现场听了一下午技术专家们深度干货内容演讲的你，肯定会有一大堆问题想要提问吧！所以，我们除了在每个演讲后为各位准备了 QA 时间，在最后闭幕阶段也准备了惊喜环节~
话不多说，2023 年 3 月 11 日下午，来北京中粮置地广场与各位大佬面对面交流吧！点击阅读原文即刻报名本次活动~
本文分享自微信公众号 - 安第斯智能云（OPPO_tech）。
如有侵权，请联系 [email protected] 删除。
本文参与“OSC源创计划”，欢迎正在阅读的你也加入，一起分享。