Towards human-guided machine learning
summary
A system called P4ML is integrated with a UI to provide a baseline P4ML integrated with a UI that returns ranked pipelines and ensembles while exposing data distributions and performance metrics; task analysis (top‑down and bottom‑up) on a political‑science replication and a neuroscience consortium study yields requirements emphasizing feature and instance selection, model type/tuning and interpretation. Existing AutoML tools (TPOT, Auto‑Keras, scikit‑learn/WEKA variants) are assessed against those requirements; many features—user constraints files, feature generation UI, bias detection—are specified but not yet fully implemented.
AI-written summary · about
chapters (5)
related talks
08mBroadening AI Adoption with AutoML
21mDemocratizing AI: AutoML Transforms Predictive Analytics
27mFrom Data to Insights in Minutes: Accelerating Predictive Modelling (in Python) with AutoML
05mModel LineUpper: Supporting Interactive Model Comparison at Multiple Levels for AutoML
32mPresence and continuity – on AI driven automation and decision making
11mAutomated ML: A journey from CRISPR.ML to Azure ML
28mHow to boost data science routines with DataArt’s AutoML Platform?
31mUse-Case Driven ML Platforms