Rafael Gomes Mantovani

Rafael Gomes Mantovani, André Luis Debiaso Rossi, Edesio Alcobaça et al.

Machine Learning (ML) algorithms have been increasingly applied to problems from several different areas. Despite their growing popularity, their predictive performance is usually affected by the values assigned to their hyperparameters (HPs). As consequence, researchers and practitioners face the challenge of how to set these values. Many users have limited knowledge about ML algorithms and the effect of their HP values and, therefore, do not take advantage of suitable settings. They usually define the HP values by trial and error, which is very subjective, not guaranteed to find good values and dependent on the user experience. Tuning techniques search for HP values able to maximize the predictive performance of induced models for a given dataset, but have the drawback of a high computational cost. Thus, practitioners use default values suggested by the algorithm developer or by tools implementing the algorithm. Although default values usually result in models with acceptable predictive performance, different implementations of the same algorithm can suggest distinct default values. To maintain a balance between tuning and using default values, we propose a strategy to generate new optimized default values. Our approach is grounded on a small set of optimized values able to obtain predictive performance values better than default settings provided by popular tools. After performing a large experiment and a careful analysis of the results, we concluded that our approach delivers better default values. Besides, it leads to competitive solutions when compared to tuned values, making it easier to use and having a lower cost. We also extracted simple rules to guide practitioners in deciding whether to use our new methodology or a HP tuning approach.

Gabriel Jonas Aguiar, Everton José Santana, Saulo Martiello Mastelini et al.

Several multi-target regression methods were devel-oped in the last years aiming at improving predictive performanceby exploring inter-target correlation within the problem. However, none of these methods outperforms the others for all problems. This motivates the development of automatic approachesto recommend the most suitable multi-target regression method. In this paper, we propose a meta-learning system to recommend the best predictive method for a given multi-target regression problem. We performed experiments with a meta-dataset generated by a total of 648 synthetic datasets. These datasets were created to explore distinct inter-targets characteristics toward recommending the most promising method. In experiments, we evaluated four different algorithms with different biases as meta-learners. Our meta-dataset is composed of 58 meta-features, based on: statistical information, correlation characteristics, linear landmarking, from the distribution and smoothness of the data, and has four different meta-labels. Results showed that induced meta-models were able to recommend the best methodfor different base level datasets with a balanced accuracy superior to 70% using a Random Forest meta-model, which statistically outperformed the meta-learning baselines.

Rafael Gomes Mantovani, André Luis Debiaso Rossi, Edesio Alcobaça et al.

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.

Rafael Gomes Mantovani, Tomáš Horváth, André L. D. Rossi et al.

Machine learning algorithms often contain many hyperparameters (HPs) whose values affect the predictive performance of the induced models in intricate ways. Due to the high number of possibilities for these HP configurations and their complex interactions, it is common to use optimization techniques to find settings that lead to high predictive performance. However, insights into efficiently exploring this vast space of configurations and dealing with the trade-off between predictive and runtime performance remain challenging. Furthermore, there are cases where the default HPs fit the suitable configuration. Additionally, for many reasons, including model validation and attendance to new legislation, there is an increasing interest in interpretable models, such as those created by the Decision Tree (DT) induction algorithms. This paper provides a comprehensive approach for investigating the effects of hyperparameter tuning for the two DT induction algorithms most often used, CART and C4.5. DT induction algorithms present high predictive performance and interpretable classification models, though many HPs need to be adjusted. Experiments were carried out with different tuning strategies to induce models and to evaluate HPs' relevance using 94 classification datasets from OpenML. The experimental results point out that different HP profiles for the tuning of each algorithm provide statistically significant improvements in most of the datasets for CART, but only in one-third for C4.5. Although different algorithms may present different tuning scenarios, the tuning techniques generally required few evaluations to find accurate solutions. Furthermore, the best technique for all the algorithms was the IRACE. Finally, we found out that tuning a specific small subset of HPs is a good alternative for achieving optimal predictive performance.