Domino Data LabThe startup, which is developing a platform that focuses on companies with large data science teams, announced today that it has raised $ 43 million in equity funding. It happens when the company upgrades its portfolio of solutions and launches a new suite of technologies – Domino Model Monitor – to prevent AI models from misrepresenting distortion or deterioration.
In general, the accuracy of a model is best until use, a theory called concept drift. The statistical properties of the variables that the model tries to predict are bound to shift over time. For example, people's preferences change, or a competing company takes a step that turns certain assumptions upside down. Domino – whose customers include Dell, Allstate, UBS, Bristol Meyers, ConocoPhillips and Lockheed Martin – wants to reduce the drift with products that automate various model update processes.
With Domino Workbench, engineers can use existing tools to track, reproduce, and compare experiments while finding, discussing, and reusing work in one place. You can ramp up interactive workspaces to hardware of your choice and scale to more powerful compute resources when needed, while using an integrated package manager to organize the libraries and tools accessed during a project, and versioned records to keep track of the data used during the model Training and testing. With the Workbench reporting capabilities, administrators can schedule reports to be generated automatically and delivered to stakeholders while data pipelines perform tasks to keep models up to date. The on-demand launcher creates analysis forms that business users can use themselves.
On the model operations side, Domino customers can deploy models as on-demand APIs or export models for deployment on another infrastructure. The model monitor detects the data drift – the change in data that results in performance degradation – and monitors the performance of models in the wild, while alerting engineers to those who underperform. A free registration shows the status of models at a glance, regardless of where or how they were provided. Integration with Jira and other business tools ensures the level of verifiability and security required to comply with government regulations.
According to Domino, it is not uncommon for the accuracy of production models to deteriorate between 10% and 20% – especially in turbulent socio-economic times. "Companies must be more aware than ever of changes in buyer preferences, economic changes and other external factors that are beyond their control and that make their models redundant," a spokesman told VentureBeat. “Domino gives businesses the ability to … detect changes (using) a dashboard … Even if they don't want to monitor the dashboard, they can set thresholds so that … notifications can be sent to notify key people that it's time, either retraining It uses updated data from today's world or completely recreates it using a new algorithm. "
Domino also monitors the calculation and measures the business impact of model APIs, apps, and more, as project status is shown in terms of both progress and potential obstacles. With the latest updates to the platform backend, Domino customers can make externally hosted Git code repositories (on GitHub, Bitbucket, etc.) the primary location for project files. Further improvements are in preparation, including on-demand spark clusters for distributed processing, support for Kubernetes orchestration with Microsoft Azure and Red Hat Openshift, and the ability to export model images to Amazon Web Services (AWS) Sagemaker.
Domino, headquartered in San Francisco, is aimed at companies in the insurance, financial services, internet and technology, life sciences, manufacturing, media and retail, healthcare, oil and gas, and banking businesses to acquire part of the emerging “MLOps” market. It seems to be going well. This final round of financing, jointly led by Highland Capital Partners and Dell Capital, brings a total of over $ 123.6 million in August 2018 after a Series D of $ 40 million.