Scikit-Learn can do THAT?!
summary
Shows concrete capabilities and results: scikit-learn implements pipeline caching that can halve grid‑search time on a single machine, supports sparse inputs without densifying by offering options like variance‑only scaling, and implements partial_fit out-of-core with a numerically stable online variance algorithm that avoids overflow. The library accepts sample weights and routes them via metadata routing, loads sentence‑transformer and CLIP embeddings for image and text classification, and powers a 600k‑abstract vector search with semi‑supervised label‑spreading that actually works.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:05Beyond fit and predict
- 3:34Caching pipelines
- 5:20Embeddings and images
- 6:45Standard scaler complexity
- 13:04Partial fit and out-of-core
- 17:42Numerical stability
- 19:26Invisible features matter
- 20:32Building a search engine
- 24:45Semi-supervised learning
- 27:36All-in-one toolkit
- 28:39Q&A
related talks
01h 17mMachine Learning with Scikit-Learn
21mKeynote: Easier Machine Learning Thoughts From Scikit-Learn
39mImplementing distributed grid search for deep learning using scikit learn and joblib
42mThe Machine Understands Text — Learning and Search at Micro and Macro Scale
43mDeploying Machine Learning using sklearn pipelines
39mA full Machine learning pipeline in Scikit-learn vs in scala-Spark: pros and cons
31mAn update on the latest scikit-learn features
44mApache Spark and ML workflows