Join top executives in San Francisco July 11-12 to hear how leaders are integrating and optimizing AI investments for success. Learn more
Fista deep learning platform for relational data, announced today at Snowflake Summit 2023 its integration of deep learning capabilities directly into the Snowflake data cloud via Snowpark Container Services.
Snowflake’s recently introduced Snowpark Container services expand the functionality of Snow Park. This update allows organizations to run third-party software and full-stack applications within their Snowflake accounts.
According to Snowflake, with this integration customers can maximize the potential of their data using cutting-edge tools while maintaining data security and eliminating the need for data movement.
In addition, Snowpark Container Services includes GPU support, which gives data science and machine learning teams a way to accelerate development and bridge the gap between model implementation and consistent data security and governance throughout the cycle of life of AI/ML.
Kumo is an early adopter of Snowpark Container Services, using the technology to implement advanced neural networks for enterprises.
Kumo’s predictive AI platform uses graphical neural network (GNN) technology, enabling developers, data scientists, analysts, and business owners to create and implement highly accurate predictions in manufacturing.
Design neural networks and artificial intelligence
Traditional machine learning requires extracting data from a data warehouse or lake, followed by manual development and optimization of features. The new integration, now available in private preview, allows joint users to work directly on raw Snowflake tables; generate predictions; and stores the results as additional tables within Snowflake.
“The new integration will run Kumo AI services directly on relational tables in the cloud without the intermediate steps typical of traditional machine learning, such as generating training sets and engineering features, using graph neural network technology,” Vanja Josifovski, co-founder and CEO of Kumo, told VentureBeat.
Josifovski highlighted that users can create and run a query that offers predictions, mirroring the process of querying past data for analysis, all without the need to export data from their Snowflake environment.
The announcement follows a recent collaboration between Nvidia and Snowflake that allows customers to customize their AI models through the cloud to meet their specific business needs.
The integration allows organizations to build generative AI applications using their own first-party data within the Snowflake Data Cloud environment, eliminating the need to move data offsite.
Facilitate deep learning-based predictive analytics in the cloud
According to Kumo’s Josifovski, Snowpark Container Services will allow customers to directly use Kumo’s predictive AI service within Snowflake to conduct graphical learning predictions on their enterprise data.
“An age-old question around machine learning and data warehousing has been where ML processing is done. By changing the paradigm to do ML processing in the Snowflake Data Cloud, our companies are enabling users to extend the use of machine learning and predictions to everyone who has access to the Data Cloud,” Josifovski told VentureBeat. “ This is done with a single security program which is much more simplified than working with multiple security programs.”
Modern AI methods rely heavily on linear algebra calculations, which are highly compatible with GPU computing. Previously, to use GPUs, Kumo had to pull data from the customer’s account and process it externally. With this integration, all data processing occurs directly within the customer’s Snowflake account, including GPU processing.
“The no-need-of-a-training-and-engineering-feature approach significantly shortens the AI/ML lifecycle,” he added. “We aim to relieve data scientists of repetitive and tedious tasks, focus on higher level tasks to define the right predictive task, evaluate the results and find the best way to get business value from forecasting.”
The company introduced a distinguishing feature through this offering: deep learning-powered relational data GNNs.
These deep learning-based GNNs can learn from the graph and associated attributes, which are determined by non-key columns of data. Once a graph is built, multiple AI/ML tasks can be efficiently trained on the same graph without creating separate training sets or numerous engineered features.
Kumo also offers an innovative, scalable autoML algorithm that alleviates the cumbersome process of tuning hyperparameters.
“Although GNNs are very effective for a wide range of predictive problems, they are also difficult to implement, scale and make efficient. Kumo’s AI platform eliminates the need to create graphs, which requires familiarity with GNNs and the creation of optimization tasks. To specify AI/ML activity, Kumo implemented a predictive query language,” Josifovski said.
Simplifying predictive analytics for citizen developers
Josifovski says predictive AI/ML currently requires highly skilled specialists with limited expertise. The lifecycle involves testing features, requiring substantial infrastructure support for training and inference (scoring).
He explained that the goal of the new integration is to offer users a streamlined workflow, regardless of their data science proficiency.
They can then easily apply graph predictive learning across different business domains such as customer acquisition, retention, retention, personalization, and fraud detection. His company claims that an entire AI-powered analysis can be completed in hours.
“Kumo allows users to query relational data without requiring a deep understanding of AI/ML concepts, while providing training and inference control for experienced data scientists,” Josifovski said. “In this way, the platform enables a broad range of users to use it, similar to how data warehouses are used for analytics today.”
Additionally, Kumo noted that the native integration with Snowflake makes it easy to install and use the product without requiring legal privacy and security reviews. This reduces barriers and significantly shortens the time to value.
The company is confident that this will accelerate the experimentation and implementation of detailed predictions, enabling and improving practices such as customer acquisition, personalization, entity resolution and other predictive activities.
“In companies, many teams send SQL queries to a data warehouse to get analytics that professionals use to track future actions,” Josifovski told VentureBeat. “Kumo will enable users to get actionable predictions in an automated way, without requiring professional interpretation.”
VentureBeat’s mission it is to be a digital city square for technical decision makers to gain insights into transformative business technology and transactions. Discover our Briefings.
