On June 18th, the Huawei Cloud TechWave Global Technology Summit (Artificial Intelligence & Data) was successfully held in Shanghai. AI technology sub-forum “AI technology innovation accelerates the industry’s intelligent upgrade” As the theme, Huawei Cloud technical experts shared the latest AI technology and product developments, and industry guests shared AI’s landing practices in Internet media, medical, environmental and other industries, and how the Huawei Cloud Pangu NLP model can help companies achieve revenue. Leading figures from all walks of life gathered together to discuss how to implement the AI ​​industry!
Live photos of the AI-enabled sub-forum

HUAWEI CLOUD AI continues to innovate, accelerating industry intelligence upgrades

Fang Fan, a senior planner of Huawei Cloud EI, believes that artificial intelligence has entered thousands of industries and has begun to enter the practical stage. AI has entered the core production system of enterprises and began to create greater value. The overall development strategy of Huawei Cloud AI is to create a full-stack, full-scenario solution for cloud, edge, and end devices, providing ample and economical computing resources, easy-to-use, high-efficiency, and full-process AI platform. At the same time, the latest AI technology and practice accumulation are added to existing products and services, allowing customers to effectively implement AI; and facing the world, continue to cooperate extensively with academia, industry and industry partners to invest in open ecology and talent training.
Fang Fan, Senior Planner of Huawei Cloud EI
In order to be more committed to lowering the threshold of AI development, Huawei Cloud’s one-stop high-efficiency AI development platform ModelArts continues to upgrade and further upgrades its development capabilities: ModelArts CodeLab, which makes AI exploration and learning more efficient, through a large number of course cases in the AI ​​Gallery, which can be connected in seconds. Entry and practice; ModelArts IDE SDKs, cloud on and off, seamless collaboration, close to the usage habits of developers; ModelArts Studio, flexibly develop AI application workflow, accelerate model experiments and deployment iterations, and improve the overall efficiency of the AI ​​application process. In addition, in order to more conveniently implement AI applications, Huawei Cloud provides a unified framework for Runtime Framework on the cloud side, which can be developed once online, multiple reasoning, and real-time model updates and iterations.

In addition, in order to further reduce the difficulty of development for developers, HUAWEI CLOUD has built the AI ​​Gallery community, providing AI assets such as data sets, algorithms, pre-training models, workflows, models, papers, articles, courses, etc. related to many industries and fields. As with scenario cases, there are currently 2000+ AI assets, the most 6+ asset types in the industry. The newly released HUAWEI Cloud Pangu large model with hundreds of billions of parameters is also among them. It extracts and learns common knowledge for massive data, solves the problem of low-cost large-scale replication, and realizes “industrialized” development.

Classics will be passed on forever, and media AI will bring old movies to life

Zhou Suyue revealed that the black-and-white movie “Leifeng” adopts GPU parallel computing and AI iterative optimization and coloring. The self-learning iterative coloring process solves the two major problems of high computational complexity and difficult model building for black-and-white video coloring; further implementation of neural network automatic learning 1. The color timing consistency between frames, 2. The color space coordination within the frame, 3. The color semantic consistency of the object objects to achieve high-efficiency and high-quality AI coloring results. Zhou Suyue, Chief Expert of Malanshan Video Cultural and Creative Industrial Park Based on the HUAWEI cloud platform and media AI capabilities, Malanshan Video Cultural and Creative Park provides computing, storage, network, video media, AI, big data and other capabilities to meet the needs of cloud services in all aspects of video production, production, and distribution, greatly improving Improved the speed of video processing. The Malanshan Huawei Cloud Audio and Video Industry Innovation Center will combine 2021 industrial operations, joint innovation, industrial planning and ecological construction, from building an industrial park to building an industrial intelligent body, through the construction of the entire process of variety show production, the entire process of film production, and lightweight Four production lines for live broadcast and 3D music production lines, two platforms for video cultural and creative talent training and digital asset trading, as well as virtual production innovation, and the entire chain on the cloud, help Malanshan build a highly competitive China “V” valley.

Every second counts for the new crown epidemic, AI increases the efficiency of drug screening by ten times

The speech by Dr. Mingyue Zheng, a researcher at the Shanghai Institute of Materia Medica, Chinese Academy of Sciences, focused on the AI ​​large-scale virtual drug screening cloud service released by HUAWEI CLOUD EIHealth and the drug federation learning service based on HUAWEI CLOUD ModelArts. Dr. Mingyue Zheng, Researcher, Shanghai Institute of Materia Medica, Chinese Academy of Sciences Based on the HUAWEI CLOUD medical agent EIHealth, Researcher Zheng Mingyue and HUAWEI CLOUD team have carried out a number of AI drug research and development cooperation projects, such as the iPhord algorithm for predicting protein structure from the primary sequence for early research on drug targets, and the AI ​​prediction algorithm for drug relocation , Use the multi-omics automatic modeling tool AutoOmics to quickly discover biomarkers, use the personalized federated learning integration algorithm FedAMP to achieve efficient and safe integration of pharmaceutical enterprise data, and use Huawei Cloud’s automatic AI modeling tool for omics data to accurately predict drug attributes Wait. HUAWEI CLOUD EI Medical Intelligence (EIHealth) platform, based on the powerful AI capabilities of HUAWEI CLOUD AI Ascension Cluster Service and HUAWEI CLOUD ModelArts one-stop AI development and management platform, integrates many algorithms, tools, AI models and automated pipelines in the medical field . Through the medical agent platform, protein homology modeling, molecular dynamics simulation calculations, and large-scale virtual drug screening can be completed quickly, and tens of millions of simulation calculations can be completed in a short time, greatly shortening the screening time and improving drug screening More than 10 times the efficiency.

Intelligent algorithm empowers indoor air quality

Wu Ke, deputy dean of Daikin China Institute of Technology Development, explained how intelligent algorithms empower IAQ to monitor, predict, and control indoor air. Wu Ke, Vice President of Daikin China Institute of Technology Development After the new crown epidemic in 2020, people pay more attention to the indoor environment. Users gradually accept CO2 as an important indoor air (IAQ) indicator. The air retention of CO2 can cause discomfort to the human body and increase the possibility of spreading influenza. In addition, the country’s carbon is at its peak, the carbon neutrality policy is clearly directed, and the company’s energy saving requirements have become a must. IAQ is a highly professional field, and standards and testing methods need to be interpreted by professionals. In general commercial scenarios, there is a lack of professional personnel and effective data detection and control. In order to solve user pain points, Wu Ke led the team and Huawei Cloud EI team to dig deep into IAQ’s reserve technology, through big data and AI algorithms, to achieve air quality prediction, so as to accurately define the time point of the fresh air switch. In the air quality prediction project, the advantages of efficient integration and joint optimization of the HUAWEI CLOUD Pangu large model are used to improve the prediction accuracy and efficiency, and play an extremely important role in the prediction of intelligent control and air cleanliness indicators.

AI enables intelligent transformation of the financial industry

Zhang Song, Director of Huawei Cloud EI Product Solutions, mentioned that the key to the entire financial value chain is how to collect and transform physical world information, how to extract digital credit, and then evaluate financial assets based on digital credit. Four aspects are involved in this process:

1. Efficiency improvement: The difficulty lies in the huge amount of information collection, the difficulty in extracting unstructured data, and the many process breakpoints.

2. Risk control: Fraudulent transactions are grouped, the equity structure is nested, and the credit relationship is complicated.

3. Investing in research and improving efficiency: It involves many data sources, strong dependence on data service providers, and strong data interference signals.

4. Risk pricing: requires strong real-time calculations and high demand for combined pricing.

Zhang Song, Director of Huawei Cloud EI Product Solutions Based on customer demands, HUAWEI CLOUD proposed innovative solutions for the financial value chain: automated smart operation, smart risk control, and smart investment and research. First of all, the automated smart operation solution uses the capabilities of AI+RPA to create digital employees, perform repetitive tasks, and increase service efficiency by 20 times in financial marketing scenarios. Second, the smart risk control solution uses the Huawei Cloud GES graph computing engine and the RTD real-time decision engine to build relationships based on information such as capital flow, transfer records, accounts, corporate information, and industry information, and effectively improve risk control and marketing levels by mining relationships. Finally, the intelligent investment research solution based on the HUAWEI Cloud Pangu model integrates millions of financial news and Chinese and foreign stock data across the entire network, monitors hundreds of commodity prices, e-commerce commodity prices and sales, monitors various risk events, and uses nature. Language processing technology establishes a graph of the relationship between events and securities, and realizes event-driven risk prevention and control and investment opportunity mining.

How Huawei Cloud Pangu NLP Large Model Helps Companies Realize Revenue Growth

Dr. Yang Zhilin, co-founder of Recycle Intelligence, described the HUAWEI CLOUD Pangu NLP large model created in collaboration with HUAWEI CLOUD, which overcomes the problems of traditional large models in commercial implementation, and brings “improvement of employee execution ability in enterprise marketing and communication scenarios”. Solutions for three business scenarios, such as “target customer preference” and “compliance enhancement”, help companies to improve employee professional capabilities faster, find target customers faster, reduce compliance risks, and reshape customer experience And the steady growth of revenue. Dr. Yang Zhilin, Co-founder of Recycle Intelligence Among them, the “employee execution improvement” scenario solution uses the HUAWEI CLOUD Pangu NLP large model to digitally model the real sales communication process, open the black box of the communication process, and help managers to supervise the daily execution of sales staff, and make targeted guide. At the same time, it provides real-time assistance systems for front-line sales personnel to help them improve execution and productivity. The “target customer optimization” solution uses the HUAWEI CLOUD Pangu NLP large model to build an intention prediction model based on the communication dialogue data and the result of the order. Through the algorithm model, the high intention customers in the stock clues can be screened, or the customers who are more interested in a certain product can be screened Target customers, reduce the waste of energy on low-intention leads, and improve the efficiency of sales staff. The “Compliance Enhancement” solution leverages the paragraph-level semantic analysis capabilities of the HUAWEI CLOUD Pangu NLP model to help companies discover potential compliance risks in the communication between employees and customers, and greatly enhance business compliance.

Collaborative deployment of end-side cloud, accelerating the landing of industry AI applications

Finally, Xie Zhouyi, service director of HUAWEI CLOUD EI HiLens, explained how the AI ​​platform of HUAWEI CLOUD can solve this problem through the collaborative end-to-cloud solution. Xie Zhouyi, Director of Huawei Cloud EI Services In the process of AI applications being implemented in various industries, developers need to develop multiple sets of incompatible AI applications based on different platform development standards and interfaces, resulting in a long AI application development cycle and high costs. In addition, how to agile development and rapid deployment of AI applications has become an important problem to be solved in the AI ​​industry. Huawei Cloud’s HiLens platform is the industry’s first end-cloud collaborative AI application development and deployment platform. Based on the three capabilities of the management and operation platform HiLens Console, the development framework HiLens Framework, and the cloud development environment HiLens Studio, Huawei Cloud HiLens supports developers You can quickly develop your own skills without setting up the environment, and can simulate the effect of the algorithm online, and you can also use the rich AI skills preset on the platform. Through the unified development framework of HiLens end-side cloud, AI skills can be deployed to HiLens-supported end-side or edge devices for real-time reasoning. At the same time, it can also be combined with the capabilities of Huawei Cloud Cloud to easily integrate into third-party platform systems to form overall AI Business logic, build smart solutions in urban, financial, park, water, and industrial scenarios, and promote the rapid implementation of AI applications in various industries. Follow @Huaweiyun for more information