A Meta-Learning Method to Select the Kernel Width in Support Vector Regression

By Carlos Soares, Pavel Brazdil, Petr Kuba. 2004

201 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Automated Machine Learning: State-of-The-Art and Open Challenges (2019, background): “There exists a rich literature on using meta models for model configuration recommendations [114, 92, 65, 68, 44].”
Meta-Learning: A Survey (2018, neutral): “, 2013) min,max,μ,σ Gravity gravity(X) Inter-class dispersion (Ali and Smith-Miles, 2006a) ANOVA p-value pval X1X2 Feature redundancy (Kalousis, 2002) pvalXY (Soares et al., 2004) Coeff.”
Predicting hyperparameters from meta-features in binary classification problems (2018, methodology): “, 2014) and ranking predetermined hyperparameter sets (Soares et al., 2004), we propose the use of meta-learning to directly predict the optimal hyperparameters of various ML algorithms for training models in binary classification tasks.”
A comprehensive review on ensemble deep learning: Opportunities and challenges (2023, methodology): “The second fusion method is meta-learning (Soares et al., 2004), also known as ‘‘learning to learn”, which is the process of learning from learners.”
Taking Human out of Learning Applications: A Survey on Automated Machine Learning (2018, methodology): “• General configuration evaluation: the evaluation for other kinds of configurations can equip meta-learning in similar ways: in ExploreKit [9], ranking classifiers are trained to rank candidate features; in [111], metaregressor is trained to score kernel widths as hyperparameters for support vector”
Hyperparameter Importance Across Datasets (2017, methodology): “m similar datasets, and the resulting predictions are used to recommend a configuration for the dataset at hand. These techniques have been successfully used to recommend good hyperparameter settings [30, 35], to warm-start optimization procedures [13] or prune search spaces [44]. However, it is har”
Efficient Benchmarking of Algorithm Configuration Procedures via Model-Based Surrogates (2017, methodology): “…To study whether it is necessary for our model to distinguish different instances, we also considered another baseline, inspired by a metric used by Soares and Brazdil 2004: We computed the rank correlation between the true running times and the mean running times per configuration across…”
Algorithm Selection for Combinatorial Search Problems: A Survey (2012, background): “Other domains include Machine Learning (Soares et al., 2004; Leite et al., 2010), the most probable explanation problem (Guo & Hsu, 2004), parallel reduction algorithms Yu et al. (2004), Yu and Rauchwerger (2006) and simulation (Wang & Tropper, 2007; Ewald et al., 2010).”
Predicting the Performance of Learning Algorithms Using Support Vector Machines as Meta-regressors (2008, background): “According to [18], a property of NMSE which is interesting for Meta-Learning is that the values have no scale and are comparable across different datasets (which it would not be feasible if we had used a non-normalized error measure).”
Meta-QSAR: a large-scale application of meta-learning to drug design and discovery (2017, background): “A subset of these may be used for regression, and some measures are speciﬁcally deﬁned for regression targets (Soares et al., 2004).”
Meta-learning for evolutionary parameter optimization of classifiers (2012, methodology): “The only work in this direction uses meta-learning to predict kernel parameters for Support Vector Machines (Soares et al. 2004; Soares and Brazdil 2006).”