Meta-Learning

By Pavel Brazdil, Christophe G. Giraud-Carrier, Carlos Soares, Ricardo Vilalta. 2009

212 citations recorded by the source corpus.

How researchers see this paper

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

Citations in the corpus

Efficient Benchmarking of Hyperparameter Optimizers via Surrogates (2015, methodology): “In the field of meta-learning (Brazdil et al. 2008), regression models have been extensively used to predict the performance of algorithms across various datasets based on dataset features (Guerra, Prudêncio, and Ludermir 2008; Reif et al.”
Continual Learning for Robotics (2019, background): “Meta-learning Meta-learning [12] is a learning process that uses meta-data about past experiences, such as hyper-parameters, in order to improve its capacity to learn on new experiences.”
Democratizing Data Science through Interactive Curation of ML Pipelines (2019, background): “Meta-learning [7] tries to infer learning algorithms performances from the performance of learning algorithms across different datasets.”
Ensemble approaches for regression: A survey (2012, methodology): “One approach on this kind of analysis is metalearning [Brazdil et al. 2009].”
A survey of deep meta-learning (2020, methodology): “This broad notion includes more traditional algorithm selection and hyperparameter optimization techniques for Machine Learning (Brazdil et al., 2008).”
Hyperparameter Importance Across Datasets (2017, methodology): “on effects, and (unlike ablation analysis) does not rely on a speciﬁc default conﬁguration. Our methods are, however, by no means limited to functional ANOVA. Priors. The ﬁeld of meta-learning (e.g., Brazdil et al. (2008)) is implicitly based on priors: a model is trained on data characteristics (so”
On the predictive power of meta-features in OpenML (2017, background): “Among them, one that has been the focus of research for long is meta-learning (Brazdil et al., 2008; Bilalli et al., 2017; Lemke et al., 2015).”
Analysis of the AutoML Challenge Series 2015-2018 (2019, background): “Meta-learning One question is whether meta-learning (Brazdil et al., 2008) is possible, that is learning to predict whether a given classifier will perform well on future datasets (without actually training it), based on its past performances on other datasets.”
Alors: An algorithm recommender system (2017, background): “Collaborative ﬁltering (CF) oﬀers two main advantages compared to supervised ML-based algorithm selection (Brazdil et al., 2008; Xu et al., 2012a).”
Towards Knowledgeable Supervised Lifelong Learning Systems (2020, background): “, 2016), multitask learning (Caruana, 1998), metalearning (Brazdil et al., 2008; Lemke et al., 2015), lifelong machine learning (Chen & Liu, 2016, 2018) and continual learning (Parisi et al.”
Evaluation in artificial intelligence: from task-oriented to ability-oriented measurement (2017, background): “…very useful in the development of several machine learning techniques in the past two decades, such as ensemble methods (Zhou, 2012) or meta-learning (Brazdil et al, 2008), where the evaluation over multiple datasets was crucial to compare the improvements of new algorithms, parameters and tools.”
Surrogate Benchmarks for Hyperparameter Optimization (2014, methodology): “In the field of metalearning [6], regression models have been extensively used to predict the performance of algorithms across various datasets based on dataset features [11, 26].”