In today’s rapidly evolving healthcare landscape, the ability to securely and efficiently share data is critical to driving better patient outcomes, operational efficiencies, and groundbreaking research. In this session, Komodo Health will explore how Delta sharing unlocks new opportunities across the life sciences ecosystem, with de-identified longitudinal patient data without compromising patient privacy. We will share insights into customers' experiences leveraging de-identified patient data to reduce the burden of disease while improving the overall patient experience. Attendees will learn practical approaches to compliantly share data in life sciences.

This session will cover the challenges and the solution that The Trade Desk went through to scale their ML models for NLP for 100 million web pages per day.

TTD's contextual targeting team needs to analyze 100 million web pages per day. Fifty percent of the webpages are non-English. Half of the content was not being properly analyzed and targeted intelligently. TTD attempted to build a model using Spark NLP, however the package could not scale and was not cost-effective. GPU utilization was low and the solution was cost prohibitive. TTD engaged with Databricks in early 2022 to build an NLP model on Databricks. Our teams partnered closely together. We were able to build a solution using distributed inference (150-200 GPUs running at 80%+ utilization); Each day, Databricks translated two hundred times faster across 50 million web pages that are in for over 35 + languages and at a fraction of the cost. This solution enables TTD teams to standardize on English for contextual targeting ML models. TTD can now be a one-stop shop for their customers' global advertising needs.

The Trade Desk is headquartered in Ventura, California. It is the largest independent demand-side platform in the world, competing against Google, Facebook, and others. Unlike traditional marketing, programmatic marketing is operated by real-time, split-second decisions based on user identity, device information, and other data points. It enables highly personalized consumer experiences and improves return-on-investment for companies and advertisers.

Talk by: Xuefu Wang and Mark Lee

Connect with us: Website: https://databricks.com Twitter: https://twitter.com/databricks LinkedIn: https://www.linkedin.com/company/databricks Instagram: https://www.instagram.com/databricksinc Facebook: https://www.facebook.com/databricksinc

Typical deployment architectures introduce challenges to fully using the shared metadata platform across products, environments, and servers. Data privacy and information security requirements add even more levels of complexity. IBM® InfoSphere® Information Server provides a comprehensive, metadata-driven platform for delivering trusted information across heterogeneous systems. This IBM Redbooks® publication presents guidelines and criteria for the successful deployment of InfoSphere Information Server components in typical logical infrastructure topologies that use shared metadata capabilities of the platform, and support development lifecycle, data privacy, information security, high availability, and performance requirements. This book can help you evaluate information requirements to determine an appropriate deployment architecture, based on guidelines that are presented here, and that can fulfill specific use cases. It can also help you effectively use the functionality of your Information Server product modules and components to successfully achieve your business goals. This book is for IT architects, information management and integration specialists, and system administrators who are responsible for delivering the full suite of information integration capabilities of InfoSphere Information Server.