Huawei Cloud APAC’s Post

Transcript

Hi everyone. My name is William Farm in Arms, Chief Product Officer of High Cloud. Welcome to Huawei Cloud Tech talk. To unlock the value of the enterprise data is always a key objective of business leaders. With new disruptors like generative AI, there are new opportunities as well as challenges. Today, I'm going to talk about how to unleash the new business values with the date and AI convergence. Mackenzie says data-driven companies are 23 times more likely to acquire customers and the 19 times more likely to be profitable. However, to unlock the value of data is not easy. According to Intelligence 20-30 report, the data volume generates the worldwide will grow to over 1000 in 20-30 and more than 80% is unsearchable. Data or 80% of infant data is in the dark in the sense that it's inaccessible and not being used to drive business decisions or to improve customer experience or operational efficiencies. For structured data, traditional business intelligence can effectively utilize the data warehouse for data analysis. Algorithm driven methodologies can also assist in the data processing and support decision making. However, we Gen. AI, we can import new data intelligence. Methodologies to leverage all types of data. With the help of AI, enterprise can not only obtain reports on past activities, but also predict their business future and facilitate their business operations. On the load to data intelligence, there are some significant challenges enterprise faces. First, we need to adjust data silos. Enterprise data is scattered in different systems and cannot be easily shared and integrated. As a result, internal information and data processing are not smooth, affecting enterprise. Using making and efficiency to provide as much data training material as possible for a large model training, technical departments must migrate or replicate data in this state subtle across domains. As result, the data population time accounts for more than 50% of entire large model training process, greatly affecting service efficiency. Secondly, the performance of the large models is only as good as the data there have been changed on the air industry often says garbage in, garbage out and high quality data. Can improve model performance accelerator changing process, and enhance the generalization capability or models high quality data is the key to determine the difference in large models capabilities. Third, digital intelligence makes data engineering and model engineering collaboration a new paradigm. We found that there is work of the data engineers and the model engineers are were separate data engineers obtain data and process data from the data lake and the warehouse and the air modeling engineering. Parliament obtained data to fine tuning and influence modeled by themselves. There's no unified development and governance platform for date and air development. As a result, the collaboration efficiency of data collaboration and the model training is low. Last but not least is about data security and compliance. Enterprise data containing sensitive personal information, which must be anonymized to adhere with the regulation like GDPR, HIPAA, essential, so data intelligence. Chatham new to provide a comprehensive data governance capabilities like data encryption, data lineage and data well watermark essential. Bed and a I have become the core focus their child social progress and enterprise business growth. However, there are applications that have long been confined to a few professionals and large enterprise. Our goal is to democratize the data and AI. On the one hand, democratize data and AI means making data and AI more accessible and easier to use. Complex technologies and the high cost of use are no longer allowed to prevent people from accessing. And using the latest technologies of data I on the other hand, it also means promoting fair competition, providing more equal opportunities for SMEs and individual participants. For the data for AI, first, we are committed to provide our one stop unified data management platform that can easily consolidate, clean and store all types of data, helping customers eliminate silos and reduce duplication. Secondly, health enterprise to generate. And the major high quality data, people need to extract the insights and make reliable decisions based on high quality data. Air model training and evaluation also require high quality data sets. Third, compliance, security issues, confidential and sensitive information like your company financial or customer details must be handled carefully to avoid a privacy disclosure. For the Afro data, we first integrate the AI capability into our full data stack products, then transform enterprise. Data-driven to knowledge base in the last 30 by using Gen. AI technologies there is new data visualization capabilities and the user can use natural language to interact with data systems to unlock the data value with the data in the AI convergence, we build our data intelligence platform. The platform provides a unified storage based on OBS. It's semantically flexible data storage repository for all types of data from diverse enterprise sources with legal. Mostly it's unified and metadata service. It again store and process data of any structure. On top of that, there is a data architect fabric which is lake house layer. It provides a single service computing infrastructure that supports different analytical workload such as data engineering, data science and large language model. The digital intelligence platform streamlines the data governance, production line data and AI development production line model us to build a unified. Development environment enabling data and AI development to enter a new phase. It also provides Data US insight as a modern BI with the natural user interface supporting enterprise to do data, apps and Arabs efficiently on one unified platform.