The
5th Chinese Congress on Artificial Intelligence (CCAI2019) which was led by
Chinese Association for Artificial Intelligence (CAAI), was held on Sep 21-22,
2019, in Jiaozhou Qingdao, China. As the largest official AI conference in
China, CCAI has been held annually to promote the advancement of artificial
intelligence globally. The frame of congress named“Intellectual Change and Fusion (智变融合)”.
The CAAI Extenics Professional Committee (中國人工智能學會可拓學專業委員會) representatives participated
the CCAI 2019. I attended the Congress and summarized it for sharing AI trend
to all quality professionals. I am honor to have a chance to take photo with
Prof. Deyi Li (李德毅) (President of CAAI, Academician of CAE) at the end of the congress.
I
took a photo in front of the venue with the frame words“Intellectual Change and Fusion (智变融合)”.
Before
the Day 2 program, I visited the exhibition area and took some photo for
memory. I saw President Xi statement
about AI policy.
The exhibition hall demonstrated the AI strategies in different countries in the world in 2017 and 2018, respectively.
Then
“AI +” enable different industries were demonstrated.
I
visited different AI companies counter. The congress instant translation
service was employed iFLYTEK technology.
Baidu
and Tmall were also here.
I
took a photo with robot.
Robot
from Teksbotics and we needed to attract its attention for photo.
Motion
identification was demonstration for participants.
Day 2 (22 Sep 2019):
In
the beginning of the signing ceremony, Mr. Pan Feng (潘峰 - 胶州市委副书记) chaired this session.
After the signing ceremony, a video was played for transition.
The first speaker was Prof. Xu
Zongben (徐宗本，西安交通大学教授，中国科学院院士，数学家) and his topic entitled “AI and
mathematics: Harmony (AI与数学：融通共进)”. He said the fundamental of AI is mathematics.
Firstly Prof. Xu briefed the
relationship between agent and environment in machine learning framework. Agent could predict the output and the environment
would feedback to the agent about the output through the loss function. Because of big data, algorithm and computing
power integration, AI was from “unused” to “could use”. But the bottleneck was from “could use” to
“good use”. It needed to be autonomy intelligent (自主智能).
Then Prof. Xu raised some
challenges on mathematics in AI such as statistic in Big Data, Algorithm in Big
Data, mathematics theory in deep learning, etc.
“That was the problem from weak AI to strong AI and then to super AI” said
by Prof. Xu.
After that he discussed the
optimization of AI application especially in enforcement learning.
At the end, Prof. Xu shared his
perspectives included AI changed human life and mathematics was the fundamental
of AI model and algorithm. He concluded
that the interaction between AI and mathematics had promoted model-driven and
data-driven fusion.
The second speaker was Dr. Liu
Qingfeng (刘庆峰，科大讯飞股份有限公司董事长) and his presentation was “Your World,
Because of AI (你的世界, 因AI而能)”.
In the beginning, Dr. Liu
introduced iFLYTEK services in the coming winter Olympus including translation
system, recording, intelligent context, etc.
Then he briefed three characteristics of iFLYTEK speech recognition
included i) not connected to the internet, ii) no accent training and iii)
could translate into English, Japanese, Korean, Russian and so on.
And then Dr. Liu demonstrated the
speech synthesis technique and used Trump to speech Chinese as example.
Dr. Liu defined three standards
to identify AI benefit for implementation as follows:
i)
Any application scenario and case,
ii)
Any product for standardized promotion,
iii)
Any statistic data to proof its effect.The third speaker was Prof. So Young Sohn (韩国延世大学教授，韩国科学与技术研究院院士) and her presentation topic named “Innovation, Spatial Big Data and AI (创新、空间大数据以及人工智能)”. Firstly, she briefed her experience in research on data mining for manufacturing since 1995, and got Korea Technology Credit Guarantee fund in 2000.
Then she introduced AI application to Technology Credit Scoring and her research separated into two stages for Science of Patent Management Strategy for Effective Innovation.
Prof. So briefed her student on “Predicting the Pattern of Technology Convergence using Big-Data Technology in large-scale Triadic Patents” and “Technological Convergence in Standards for ICT” in 2015 and 2016, respectively. Recently, she researched “Deep Semantic Analysis for Forecasting New Technology Convergence” (2019). Finally, she summarized spatial big data studies included transportation, environment and residence.
The fourth speaker was Prof. Shai
Ben-David (School of Computer Science, U Waterloo - 加拿大滑铁卢大学教授，国际计算学习理论学会前主席) and his topic named “Unsupervised
learning: What can, what cannot and what should not be done (无监督学习理论的研究与应用)”. He would like to discuss several aspects of
the theory of unsupervised learning.
Then Prof. Shai highlighted three
directions of research included:
i)
Learning mixtures of Gaussians (a positive result).
ii)
Expectation maximization (EMX) in a set of random variables (a strong
negative result).
iii)
Clustering (calling out some common misconceptions and bad practices).
After that Prof. Shai discussed
some EMX problems including binary classification prediction, Multi-class
prediction, K-center clustering problems, etc.
He said the notions of learnability were defined in terms of learning
functions rather than learning algorithms.
Lastly, Prof. Shai proposed
semi-supervised-active clustering that considered algorithms to be interacted
with a domain expert by actively querying “same-cluster/diff-clusters” over
pairs of data points. At the end, he
introduced his book named “Understanding Machine Learning”.
Lunch venue is very large.
My lunch box so nice!
The second session was Robotic
Forum (机器人论坛) which was
chaired by Prof. Hou Yuguang (侯増广–中科院自动化所研究员)
The first speaker in Robotic
Forum was Prof. Cheng Hong (程洪–电子科技大学教授) and his presentation topic was “Human-machine
Intelligent System and Application (人机智能系统及应用)”.
Firstly, Prof. Cheng mentioned
the human-machine collaboration was the future of society. There were three
steps that started from Intelligent to Bionic, and then to Social.
Then he discussed human
intelligence (creative, complex and dynamic) and robot intelligence
(normalization, repeatability and logicality). He proposed the human-robot hybrid
intelligence system that could have the ability to achieve 1+1>2
hybrid-augmented intelligence by integrating human perception, cognitive
ability, robot computing and storage capacities. He also showed some
application trend in the diagram below.
Finally, he discussed AI +
Medical that had many application scenarios.
The second speaker was Prof. Ou
Yongsheng (欧勇盛–中国科学院深圳先进技术研究院智能仿生中心副主任) and his presentation entitled “Research on
Robot Intelligent Control Based on Human Behavior Imitation (基于人类行为模仿的机器人智能控制研究)”
Prof. Ou described three challenges
included high level planning of robotic technology, complicated learning and
intelligent control and simulated human soft behavior.
Lastly, Prof. Ou expected three breakthroughs
in the future included multi-sensing fusion, faster learning, and Human soft
operation through human-machine fusion.
The third speaker was Prof. Wang
Shuo (王硕–中科院自动化所研究员)
and his topic named “Underwater Bionic Robot-Working Arm System and Its
Autonomous Underwater Operation (水下仿生机器人-作业臂系统及其自主水下作业研究)”
In the beginning, Prof. Wang
introduced some challenges of robot underwater such as weak light and
communication, uncertainty and multi-target, etc.
Then Prof. Wang explained the design of that robot named “RobCutt-II”
Lastly, he demonstrated their
RobCutt-II in video and discussed the control of its motion.
The whole robot body trend was
demonstrated in the following diagram.
Dr. Liu demonstrated different
application about their robot body/arm.
Dr. Yan mentioned different application
scenarios of the integrated AI and education.
One of application was to detect
children focus during course. Their movement behaviors were recorded and
analyzed.
Finally, Dr. Yan shared some
traps during startup journey including technical-driven but neglect the demand,
focused on product but neglect service, and neglect government policy, etc. He
concluded that they could teach each children personally using AI in the
future.
Before end of this session, all
speakers were invited to discuss in the forum of robotic.
The next AI Innovative
Application Forum named “Innovative Application of Artificial Intelligence to
build a Better Life”. The first speaker
was Dr. Nie Zaiqing (聂再清–阿里巴巴研究员，阿里AI Labs北京研发中心总负责人) and his presentation title was
“Voice Assistant: Human-Machine Interaction Portal in the Age of Intelligent
Internet (语音助手：智联网时代人机交互入口)”.
Beginning, Dr. Nie briefed the voice
recognition breakthrough progress from 1993 to 2018. The error ratio reduced significantly. He also said voice recognition was the entry
of human-machine interaction.
Then he demonstrated the service
from Tmall spirit voice assistant to us.
The second speaker was Dr. Li Lei
(李磊–字节跳动资深科学家，人工智能实验室总监) and his topic named “Smart Writing Frontier Progress and Industry
Application (智能写作前沿进展与行业应用)”.
Dr. Li introduced the generating
text (phrase, sentence, and paragraph) that resembled human language (e.g.
naturalness) and satisfied desired properties (e.g. keyword occurrence). Then he mentioned different media robots
could perform smart writing.
After demonstrated different text
writing by robot including sport game and business advertisement, he pointed
out some challenges at the end.
The third speaker was Prof. Lu
Cewu (卢策吾–上海交通大学研究员)
and his presentation title was “Universal intelligent ontology (通用智能本体)”.
Prof. Lu briefed that Agent had
three basic characteristics. They were
transferable, extensible and multi-agent.
He explained every operation
could be parsing as element that was undividable manipulation primitive to form
a model. This unit had three elements that were Posture, Object and Status (P, O,
S).
The fourth speaker was Dr. Qin
Long (秦龙–先声智能CTO)
and his topic named “Artificial Intelligence Technology Reshapes Language
Education (人工智能技术重塑语言教育)”.
Dr. Qin introduced the correcting
students’ homework using LSTM machine learning to reduce teachers’
workload. He said they used this
technology to change the traditional learning to adaptive learning.
The fifth speaker was Dr. Wang
Jianzong (王健宗–平安科技副总工程师) and his presentation entitled “Federal
Intelligence Accelerated AI Landing (联邦智能加速AI落地)”.
In the beginning, Dr. Wang said
AI faced a new challenge that was the privacy law in different countries.
Federal Intelligence included
learning, reasoning, visualization, incentive and data center would be the
solution.
The sixth speaker was Dr. Yan Jun
(闫峻–医渡云 首席AI科学家) and his topic named “Technical Challenges
and Applications of Medical Big Data (医疗大数据的技术挑战与应用)”.
Firstly, Dr. Yan discussed the
challenges from medical big data included data application, processing and
integration. So that quality governance
was very important to ensure data integrity, accuracy, consistent and sequence.
Finally, he expected medical big
data could help to investigate new drug and insurance.
The last speaker was Dr. Zhang
Jiaxing (张家兴–蚂蚁金服人工智能部技术总监 资深算法专家) and his presentation topic was “AI assist
the Financial Services (AI助力金融服务)”
Firstly, Dr. Zhang introduced the
vision named BASIC that stand for Blockchain, AI, Security, IoT and
Computing.
He said they promoted Intelligent
plus service in their process.
After that the forum for Innovative
Application of Artificial Intelligence to build a Better Life started and all
speaker discussed the innovative application of AI.
The last session was finale forum
named "How to recognize the current development trend of artificial
intelligence in China" and chaired by Ms. Wang Xuechun (王雪纯).
Q1: What is the development stage of artificial intelligence in China?
Prof. Zhang Wei (张钹 - 清华大学人工智能研究院院长 中国科学院院士) said the first was the level of
scientific and technological development, and second was the ability to develop
science and technology. There is a mutual relationship. In which the ability to
develop science and technology mainly refers to three aspects: i) the ability
to discover scientifically; ii) the ability to develop technology and iii) the
ability to practice or industrialize.
Prof. Deyi Li (李德毅) (President of CAAI, Academician
of CAE) said Intelligent science involved two types of science, one was brain
science or neurology, and the other was cognitive science. The former was the
biological basis for studying human intelligence, while the latter was the
psychological activity of human intelligence.
Dr. Zhang Zhengyou (张正友) said current development of
artificial intelligence in China could summarize in two words, one was "Hot"
and the other was "Cool". “Hot”
was understood as popular and “Cool” had two meaning that one was style and the
other was to have a calmer judgment on AI.
Dr. Xiao Jing (肖京) said AI was an empowering tool.
It did not constitute an independent industry. It needs to rely on other
industries to generate added value.
Q2: How to understand the "change" of artificial intelligence
and what to do?
Prof. Deyi Li (李德毅) said the term
"intellectual and change integration" (智变融合) indicating that
"change" refers to transformation and upgrading, and
"intelligence" refers to diversity.
Q3: What are the current pain points in the field of artificial
intelligence applications?
Prof. Zhang Wei (张钹) said auto-driving as example
that can't be used in a dynamic and complicated environment.
Prof. Deyi Li (李德毅) said AI in the scenario
landing, the first step was to cut the pain points of the scene, this was the
most difficult point; the second step was able to integrate into the existing
system; after the first two steps were implemented, the third step could be a
large-scale technology promotion.
After
finished the Congress, I come back to hotel and took a photo with student
helpers. They were students from Qingdao Agricultural University.
Reference:
CCAI 2019 - http://ccai.cn/
CAAI - http://www.caai.cn/
20180728: CAAI Chinese Congress on Artificial Intelligence 2018 (中国人工智能大会) -
Day 1 - https://qualityalchemist.blogspot.com/2018/07/caai-chinese-congress-on-artificial.html
20180729: CAAI Chinese Congress on Artificial Intelligence 2018 (中国人工智能大会) -
Day 2 - https://qualityalchemist.blogspot.com/2018/07/caai-chinese-congress-on-artificial_29.html
20190921: CAAI Chinese Congress on Artificial Intelligence 2019 (中国人工智能大会) -
Day 1 - https://qualityalchemist.blogspot.com/2019/09/caai-chinese-congress-on-artificial.html
20190922:
CAAI Chinese Congress on Artificial Intelligence 2019 (中国人工智能大会) -
Day 2 - https://qualityalchemist.blogspot.com/2019/09/caai-chinese-congress-on-artificial_22.html
沒有留言:
發佈留言