Update notes

Author: cxy0714

notes/Master.bib

@@ -1,3 +1,21 @@
+@article{cattaneo2025honest,
+  title={The honest truth about causal trees: Accuracy limits for heterogeneous treatment effect estimation},
+  author={Cattaneo, Matias D and Klusowski, Jason M and Yu, Ruiqi Rae},
+  journal={arXiv preprint arXiv:2509.11381},
+  year={2025}
+}
+
+@article{athey2016recursive,
+  title={Recursive partitioning for heterogeneous causal effects},
+  author={Athey, Susan and Imbens, Guido},
+  journal={Proceedings of the National Academy of Sciences},
+  volume={113},
+  number={27},
+  pages={7353--7360},
+  year={2016},
+  publisher={National Academy of Sciences}
+}
+
 @article{agrawal2024automated,
   title   = {Automated efficient estimation using monte carlo efficient influence functions},
   author  = {Agrawal, Raj and Witty, Sam and Zane, Andy and Bingham, Elias},

notes/main.typ

@@ -428,4 +428,36 @@ to measure the importance of variable $x$ by comparing the loss when including $
 
 @wang2024multi proposed an extension of LOCO under multiple source data and using semiparametric theory to provide the inference of their measure.
 
+
+
+== Ensemble learning
+
+Ensemble learning, as the name suggests, combines multiple basde models to improve prediction performance. Common ensemble learning methods include bagging, boosting, and stacking.
+
+=== Bagging
+BoostrapAggregating (bagging) is proposed by Leo Breiman @breiman1996bagging. The key idea is to generate multiple boostrap samples from the original data and train the base model on each boostrap sample then aggragate the predictions from all models, such as by averaging for regression or majority voting for classification.
+
+==== Theory properties of bagging
+
+- @breiman1996bagging Bagging can reduce the variance of unstable base models.
+- Peter Bühlmann and Bin Yu @buhlmann2002analyzing give some convergence rate analysis for bagging.
+- In the 2000s, many works on the theoretical understanding of bagging, seems no more work needed now.
+
+==== Random forest
+
+- Leo Breiman @breiman2001random proposed random forest, an ensemble learning method that builds multiple decision trees and merges their results to improve accuracy.
+
+==== Causal forest
+
+- @athey2016recursive proposed causal tree to estimate heterogeneous treatment effects namely the conditional average treatment effect (CATE) by extending decision tree, then @wager2018estimation using random forest to improve the estimation accuracy and provide asymptotic normality for inference.
+
+- @cattaneo2025honest establishes an inconsistency lower bound on the point wise convergence rate of causal tree, and challenges the $alpha$-regularity condition (each split leaves at least a fraction $alpha$ of available samples on each side) needed to establish the convergence rate in @wager2018estimation.
+
+=== Boosting
+
+- XGBoost 
+
+=== Stacking
+
+
 #bibliography("Master.bib")