Explore monitoring unstructured data in machine learning through this 13-minute lightning talk featuring Aparna Dhinakaran and Jason Lopatecki from Arize. Gain insights into the challenges of monitoring embeddings on unstructured data and learn how Arize tackles this complex issue. Discover the importance of embeddings in unstructured models, understand the need for common tools among ML teams, and delve into the concept of embeddings. Examine the real motivations behind AI and explore ML observability for unstructured data. Learn about indexing and monitoring embeddings, measuring drift in unstructured data, and utilizing interactive visualizations. Witness a product demo showcasing Arize's approach to handling unstructured data and implementing a data-centric AI workflow.
Monitoring Unstructured Data - Lightning Session
Overview
Syllabus
[] Introduction to the topic
[] Troubleshooting unstructured ML models is difficult
[] Challenges with monitoring unstructured data
[] How data looks like
[] Embeddings are the backbone of unstructured models
[] ML teams need a common tool
[] What are embeddings?
[] The real WHY behind AI
[] ML observability for unstructured data
[] Index and Monitor every Embedding
[] Measuring drift of unstructured data
[] Interactive visualizations
[] Fix underlying data issue
[] Data-centric AI workflow
[] Demo of the product
[] Wrap up
Taught by
MLOps.community
Related Courses
-
Monitoring Machine Learning Concepts
-
Troubleshooting Unstructured Data with Embeddings
-
ML Observability: A Critical Piece in the ML Stack
-
ML Observability - A Critical Piece for Making Models Work in the Real World
-
Minimize Risk and Accelerate MLOps with ML Monitoring and Explainability
-
Implementing Data Capture for ML Observability and Drift Detection
