Figure Eight releases enhanced enterprise-grade training data platform for large scale Machine...

Figure Eight releases enhanced enterprise-grade training data platform for large scale Machine Learning deployments

SHARE

US: Figure Eight, the essential Human-in-the-Loop Machine Learning platform, has released an enhanced enterprise-grade training data platform for customers with demanding security, scalability, and customization requirements in key industry sectors such as automotive, financial services, government, healthcare, media and entertainment, and technology industries.

Figure Eight’s platform has created over 10 billion human-generated computer vision and natural language processing training data labels over the past decade for hundreds of machine learning teams. Figure Eight is now extending its commitment to delivering the foundational technology layer the market is demanding with a new enhanced enterprise-grade training data platform.  The latest release includes the following new and enhanced capabilities:

  • On-Premise Solution. Figure Eight Dedicated is an on-premise solution designed for organizations with mission-critical data sovereignty or data governance requirements. The Figure Eight platform can run in an air-gapped network or natively on any of the major cloud platforms.
  • Integration with Internal and External Annotation Teams. The Figure Eight technology platform applies both machine learning and human annotation to transform unstructured text, image, audio and video data into customized, high-quality training data. The human annotation can be provided by a customer’s own internal labeling teams, external labeling teams provided by business process outsourcers (BPOs) or Figure Eight’s global network of annotators. The Figure Eight technology platform manages the two-sided workflow for both the human annotators and the machine learning teams to create the necessary training data to make the machine learning models work for the customer’s specific application.
  • Single Sign On (SSO). This feature lets users access the Figure Eight platform using one login for improved user management and security for large-scale deployments across an enterprise. Customers who choose to integrate via SSO can validate usernames and passwords against their corporate user database rather than Figure Eight managing separate passwords for each user. Federated authentication using Security Assertion Markup Language (SAML) lets users send authentication and authorization data between Figure Eight and their corporate network.
  • Customizable Annotator Groups. The Custom Channels feature allows customers to assemble customized groups of annotators based on previous performance on customer-specific annotation tasks. This improves human judgment quality by limiting annotation tasks to annotators who have demonstrated specific skills when working on previous annotation tasks. Specialty skills could be anything from in-depth knowledge of agriculture to specific medical knowledge to fashion expertise or language skills. Assembling industry-customized groups results in higher quality results, as well as lower costs by reducing the number of untrusted judgments from unskilled annotators and the ability to complete more complex annotation tasks.
  • Customizable Ontologies. This enhanced capability allows customers to create customized ontologies with up to 255 classes supporting computer vision image and video annotation like pixel-level semantic segmentation, bounding boxes, dots, polygons and lines. For example, automotive companies can detect pedestrians, cyclists, bicycles, motorbikes, delivery vans, road signs, or location-specific items like rickshaws or kangaroos. Retailers can detect different variations of the same product category such as ski boots, rain boots, or over-the-knee boots. Agricultural companies can detect different types of crops, weeds, crop ripeness, or specific categories of infestation like salmon lice or Colorado potato beetles.
  • Support for Multiple Languages. The Figure Eight platform supports multilingual annotation through geo-targeting, localized UI and language-specific contributor portals. Figure Eight’s global network of annotators spans 6 continents and supports languages critical to enterprises operating globally, including French, Spanish, German, Portuguese, Russian, Arabic, Japanese, Chinese, and others.
  • Professional Services. Figure Eight experts can provide consulting services in areas such as project management of large-scale annotation programs, training data design, and machine learning model creation and retraining.

“Training data is the bottleneck to making AI work at scale in the real world today. Enterprises need a secure and scalable technology platform to address this bottleneck,” said Robert Munro, CTO of Figure Eight. “With AI projects moving from labs run by data scientists to the office of the CIO, our enhanced enterprise-grade training data technology platform is part of our ongoing commitment to support customers serious about deploying machine learning models into production.  We’re continuing our technology investment into the platform to meet the security, scalability, and customization needs of enterprises in creating training data.”