Unlocking the Power of Data-Centric AI.
summary
Argues that improving data, not models, is the lever for better ML and cites ChinChilla as evidence that better data beats bigger models; recommends standardized data profiling (70+ quality metrics), reproducible pipelines with Airflow-like orchestrators and data version control, and three synthetic types—dummy, data-driven, simulated—for privacy, balancing underrepresented groups, and stress testing. Describes YData’s fabric catalog: automated profiling, synthetic generators, and versioned data pipelines (plus open-source profiling/synthetic tools) and claims these reduce cost and improve model performance.
AI-written summary · about
chapters (9)
- 0:00Introduction and speaker background
- 3:21Foundational models and data quality
- 5:33Data-centric AI approach
- 7:44Data quality objectives and metrics
- 9:09Data profiling best practices
- 12:00Synthetic data generation
- 16:34Reproducibility and data orchestration
- 19:51Data version control importance
- 20:53Closing and community resources
related talks
36mArchitecting Data: A Deep Dive Into the world of Synthetic Data
25mThe Importance of Synthetic Data in Data-Centric AI
26mThe Hidden Foundation: How Effective Data Management Determines AI System Success
27mData Creation: How Organizations Are Deliberately Creating Data To Power ML & Data Applications
01h 01mYou don't need AI; You need to know what your data is.
18mUnleash your Private Datasets to Power up AI
30mData-Centric AI Cookbook-let's prep that data
13mCombining algorithms and quality data to build better AI platforms