Stacked generalization: when does it work?

By Kai Ming Ting, Ian H. Witten. 1997

Subfields: Statistical Learning Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistical Learning Theory

1. Pengfei Liu
2. Ahmed Ali Abbasi
3. Xuanjing Huang
4. Conan C. Albrecht
5. James Hansen
6. Anthony Vance
7. Jinlan Fu

Citations in the corpus

• Deep learning in neural networks: An overview (2014, neutral): “In principle, many UL methods (Section 5.6.4) could be stacked like the AEs above, the history-compressing RNNs of Section 5.10, the Restricted Boltzmann Machines (RBMs) of Section 5.15, or hierarchical Kohonen nets (Section 5.6.4), to facilitate subsequent SL. Compare Stacked Generalization (Ting &”
• ExplainaBoard: An Explainable Leaderboard for NLP (2021, background): “System combination (Ting and Witten, 1997; González-Rubio et al., 2011; Duh et al., 2011) is a technique to improve performance by combining the output from multiple existing systems.”
• SpanNER: Named Entity Re-/Recognition as Span Prediction (2021, methodology): “a, Stacked Generalization) is a general method of using a high-level model to combine lower-level models to achieve greater predictive accuracy (Ting and Witten, 1997).”
• MetaFraud: A Meta-Learning Framework for Detecting Financial Fraud (2012, methodology): “The training data for the top-level classifiers was generated by running the bottom-level classifiers using 10-fold cross-validation on the training instances (Dzeroski and Zenko 2004; Ting and Witten 1997).”
• Cascade Generalization (2000, unknown): “0Boosting with the default parameter of 10, that is aggregating over 10 trees, and St cked Generalizationas it is defined in Ting and Witten (1997) and described in an earlier section.”
• Privacy-preserving Transfer Learning for Knowledge Sharing (2018, unknown): “To solve this problem, we next propose to use a variant of stacked generalization (Ting and Witten 1997).”
• Using Meta-Learning to Support Data Mining (2004, unknown): “Research in this area investigates what base-learners and meta-learners produce best empirical results (e.g., [20],[26]); how to represent class predictions (class labels versus class-posterior probabilities [48]); and how to define meta-features ([3],[15]).”