A Generative AutoML for Tabular Data
summary
A Fujitsu AutoML system generates executable machine‑learning pipelines for tabular data, outputs Python and Jupyter source code, and attaches line‑by‑line explanations with confidence scores. It trains on a corpus of dataset–pipeline pairs, uses meta‑feature driven pipeline ranking and weak supervision to label pipeline steps, applies mutation during training to propose candidates, then executes only the top few to cut compute cost and carbon footprint. Demo claims higher accuracy than other AutoMLs and runs on‑prem or cloud.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:28Traditional AutoML limits
- 5:03Automated ML pipeline
- 9:20Diverse tabular challenges
- 10:47Meta-learning for pipelines
- 13:16Model selection & execution
- 16:33Training across datasets
- 18:44Explaining pipelines
- 20:11Live demo
- 25:30Advantages over AutoML
- 27:15Q&A
related talks
27mFrom Data to Insights in Minutes: Accelerating Predictive Modelling (in Python) with AutoML
28mHow to boost data science routines with DataArt’s AutoML Platform?
08mBroadening AI Adoption with AutoML
01h 25mSolving complex ML problems in a few lines of code with AutoGluon
31mUse AutoML to create high-quality models with just a few clicks
31mPut Your Machine Learning on Autopilot
41mUse AutoML to Create High-Quality Models
18mAutomate the boring ML stuff with pipelines