# A meta-learning recommender system for hyperparameter tuning: predicting when tuning improves SVM classifiers

## TL;DR

This paper introduces a meta-learning recommender system that predicts when hyperparameter tuning improves SVM classifiers, reducing computational costs while maintaining model performance, and provides insights into the decision-making process.

## Contribution

It presents a novel meta-learning approach to determine when hyperparameter tuning is beneficial, applicable to SVMs and other algorithms, with extensive empirical validation.

## Key findings

- Accurately predicts when tuning improves performance
- Reduces optimization time without sacrificing accuracy
- Provides insights into meta-learner decision processes

## Abstract

For many machine learning algorithms, predictive performance is critically affected by the hyperparameter values used to train them. However, tuning these hyperparameters can come at a high computational cost, especially on larger datasets, while the tuned settings do not always significantly outperform the default values. This paper proposes a recommender system based on meta-learning to identify exactly when it is better to use default values and when to tune hyperparameters for each new dataset. Besides, an in-depth analysis is performed to understand what they take into account for their decisions, providing useful insights. An extensive analysis of different categories of meta-features, meta-learners, and setups across 156 datasets is performed. Results show that it is possible to accurately predict when tuning will significantly improve the performance of the induced models. The proposed system reduces the time spent on optimization processes, without reducing the predictive performance of the induced models (when compared with the ones obtained using tuned hyperparameters). We also explain the decision-making process of the meta-learners in terms of linear separability-based hypotheses. Although this analysis is focused on the tuning of Support Vector Machines, it can also be applied to other algorithms, as shown in experiments performed with decision trees.