- Version 3.1 blends AI technology with the database to address various challenges.
- The new version adds various database capabilities to address problems with earlier functions.
Splice Machine is a startup that offers offline and batch analysis tools that back up intelligent apps to handle operational workflows. Splice Machine launches version 3.1 today to usher in new features and functionality to support enterprises with real-time AI projects. This will include elasticity support to Kubernetes, GPU, and extensions to Spark’s libraries for machine learning.
According to a recent survey, only about 20% of enterprises have mature AI and machine learning drives. Deloitte claims that 62% of respondents have deployed some form of AI. Moreover, roughly 25% of companies see half their AI projects see a failure.
Features of Splice Machine 3.1:
Splice claims that version 3.1 combines a database with AI technologies. It’ll address the challenges that the data science team comes across during training, validation, and deployment of AI systems. For instance, it introduces native Spark structured streaming ingestion, which is a feature that enables streaming resources with visibly easier ingestion than before. The new version also adds new database capabilities which include foreign key processing, rich trigger support, and improved management, index on expressions, enhanced import-export capabilities along DB2 compatibility.
Monte Zweben is the founder and CEO of Splice. Streaming ingestion capability is useful especially for industry-based accounts which have a connection with a distributed control system. This is where the ingestion of data becomes available in a real-time environment. “With 3.1, we have made vital leaps in the database capabilities,” Zweben said. “[They’ll] successfully operationalize real-time AI applications and bring machine learning models into production.”
Goals of the company:
Splice Machine 3.1 aims to enhance the transparency around the data to build AI and machine learning models. The latest feature enables developers to send a query to the database back in time with syntax and a specific date while providing an audit and lineage for a regulator checking for bias or data drift. It derives the live sensor data for operational problems prediction to prevent outages and keep the machinery running.
Problems with AI training and machine learning:
A lot of organizations are struggling with the process of adoption of AI within the important stages of data collection, preprocessing, and preparation. Recent research states that less than about 4% of the organizations reported that they could train AI successfully with the data in question. Also, the system did not show any problems with data-related issues. These stem from the way data is produced and labeled in an internal system. In the list of problems, data bias, lack of resources, and errors topped the list of data management problems. These are the problems the organizations struggled with, most often.
“We are excited to be powering data engineers and data scientists with the tools they need,” Zweben added. “[We believe they’ll] break down the chasms that stop machine learning and AI projects from being successful.”
Splice Machine 3.1 is available in the form of fully managed cloud service on Amazon Web Service, Azure, and Google cloud platform which is also available for use on-premises.