Explore how machine learning-powered data cleaning streamlines the AI pipeline in this 30-minute video from Snorkel AI. Learn about the challenges data scientists face in preparing, cleaning, and transforming raw data before model training. Discover automated approaches to data cleaning, including record linkage pipelines, probabilistic cleaning models, and imputation techniques. Gain insights into differential privacy synthesis with structure and methodologies for automating data cleaning infrastructure. Understand how ML-driven data cleaning can significantly reduce the labor-intensive exercises that impede end-to-end AI pipelines, ultimately accelerating the data science workflow.
Overview
Syllabus
Intro
Data Prep is the Impediment for Al
Record Linkage Pipeline
Dirty Data Beyond Integration
Automating Cleaning with ML
A Probabilistic Cleaning Model
Use Case: Imputation
ML for Cleaning Automation
Differential Privacy Synthesis with Structure
Methodology Overview
Automating Data Cleaning Infrastructure
Taught by
Snorkel AI
