Automatic Machine Learning by Pipeline Synthesis using Model-Based Reinforcement Learning and a Grammar
By Iddo Drori, Yamuna Krishnamurthy, Raoni Lourenço, Rémi Rampin, Kyunghyun Cho, Cláudio T. Silva, Juliana Freire. 2019
30 citations recorded by the source corpus.
Automatic machine learning is an important problem in the forefront of machine learning. The strongest AutoML systems are based on neural networks, evolutionary algorithms, and Bayesian optimization. Recently AlphaD3M reached state-of-the-art results with an order of magnitude speedup using reinforcement learning with self-play. In this work we extend AlphaD3M by using a pipeline grammar and a pre-trained model which generalizes from many different datasets and similar tasks. Our results demonstrate improved performance compared with our earlier work and existing methods on AutoML benchmark datasets for classification and regression tasks. In the spirit of reproducible research we make our data, models, and code publicly available.
Subfields: Artificial Intelligence
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
- Frank Hutter
- Marco F. Huber
- Marc-André Zöller
- Marius Thomas Lindauer
- Katharina Eggensperger
- Stefan Falkner
- Matthias Feurer
- Aaron Klein
Citations in the corpus
- Benchmark and Survey of Automated Machine Learning Frameworks (2019, methodology): “Similarly, Drori et al. (2019) and Rakotoarison et al. (2019) use meta-features of the data set and pipeline candidate to predict the performance of the pipeline.”
- Auto-Sklearn 2.0: Hands-free AutoML via Meta-Learning (2020, methodology): “, 2018) or reinforcement learning (Drori et al., 2019; Heffetz et al., 2020).”