Scalable Gaussian process-based transfer surrogates for hyperparameter optimization

By Martin Wistuba, Nicolas Schilling, Lars Schmidt-Thieme. 2017

150 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics

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 Computational Theory and Mathematics

Citations in the corpus

Automated Machine Learning: State-of-The-Art and Open Challenges (2019, background): “Another way to define learning from model evaluations is through surrogate models [129].”
Meta-Learning: A Survey (2018, neutral): “One can then learn how to combine these per-task predictions to warm-start or guide optimization techniques on a new task tnew (Feurer et al., 2018a; Perrone et al., 2017; Springenberg et al., 2016; Wistuba et al., 2018), as discussed in Section 2.”
P4ML: A Phased Performance-Based Pipeline Planner for Automated Machine Learning (2018, methodology): “While Feurer et al. (2015a) use random forests as a surrogate, recent advances have been achieved by using matrix factorization (Fusi et al., 2017) or scaling Gaussian process surrogates to large collections of metadata (Wistuba et al., 2018).”
Auto-Sklearn 2.0: Hands-free AutoML via Meta-Learning (2020, methodology): “A generalization of this result given by Krause and Golovin (2014) also tightens this bound to close 99% of the gap between the worst ADTM score and the score the optimal portfolio P∗ of size |P∗| would achieve, by extending the portfolio constructed by the greedy algorithm to size 5 · |P|.”
Taking Human out of Learning Applications: A Survey on Automated Machine Learning (2018, unknown): “ptimization, classiﬁcation-based optimization and optimistic optimization: Bayesian optimization [99], [100], [101]: Bayesian optimization builds a probabilistic model, e.g., Gaussian process [102], [103], tree-based model [104], [105], or deep network [106], that maps the conﬁgurations to their per”
Inductive Transfer for Neural Architecture Optimization (2019, background): “The use of inductive transfer in the metalearning community is very common (Vanschoren, 2018; Wistuba et al., 2018).”