Elements of Learning Systems
summary
Covers two concrete systems: XGBoost and TVM. Claims XGBoost removes manual imputation by learning a automatic missing-value default direction inside tree training and scales across languages and platforms, cited as used in over half of winning Kaggle solutions. Describes TVM as an automated compiler that formulates a search over low‑level program variants, learns a statistical cost model from benchmark data to avoid exhaustive runs, and delivers surprising wins on nonstandard models including day-zero Apple S A M1 support.
AI-written summary · about
chapters (7)
- 0:00Introduction to learning systems
- 2:04XGBoost accessibility and scalability
- 7:59Handling missing values in XGBoost
- 8:20Deploying AI across diverse hardware
- 10:47AutoTVM automated optimization
- 17:09TVM case studies and impact
- 18:55Conclusion and future outlook
related talks
44mWinning with Simple, even Linear, Models
06mML for Systems and Systems for ML
43mToward Human-Centered Machine Learning
45mAGT and Machine Learning
39mMachine Learning Engineering
30mMachine Learning Techniques for Class Imbalances & Adversaries
22mMulti-level Machine Learning-Guided Autotuning for Efficient Code Generation on a Deep(…)
22mReal-world applications of machine learning: Beyond the hype