A Comparison of AutoML Tools for Machine Learning, Deep Learning and XGBoost
read more
Citations
Prediction of Maintenance Equipment Failures Using Automated Machine Learning
A comparison of machine learning methods for extremely unbalanced industrial quality data
A comparison of machine learning approaches for predicting in-car display production quality
Machine Learning Models for Slope Stability Classification of Circular Mode Failure: An Updated Database and Automated Machine Learning (AutoML) Approach
An Empirical Study on the Usage of Automated Machine Learning Tools
References
Efficient and robust automated machine learning
AMC: AutoML for Model Compression and Acceleration on Mobile Devices
Auto-WEKA: combined selection and hyperparameter optimization of classification algorithms
OpenML: networked science in machine learning
Auto-Keras: An Efficient Neural Architecture Search System
Related Papers (5)
A Case Study on Machine Learning Applications and Performance Improvement in Learning Algorithm
Frequently Asked Questions (14)
Q2. What are the future works mentioned in the paper "A comparison of automl tools for machine learning, deep learning and xgboost" ?
In future work, the authors intend to enlarge the comparison by considering more open-source AutoML technologies and datasets. In particular, the authors wish to analyze big data, where DL can potentially produce better predictions. The authors also plan to benchmark ML frameworks for specific infrastructure settings, such as involving edge computing.
Q3. What is the option for the XGB scenario?
As for the XGB scenario, rminer is the best overall option for binary and regression tasks, while H2O is recommended for multi-class tasks.
Q4. What was the criteria for the selection of the ML dataset?
The data selection criterion was defined as selecting the most downloaded datasets that did not include missing data and that reflected three supervised learning tasks: regression, binary and multi-class classification.
Q5. How many instances are used for the external testing?
For instance, if the data contains 100 instances, then in the first external 10-fold iteration 90 examples are used by the tool for fitting purposes (model selection and training), with the remaining 10 instances being used for the external testing.
Q6. What is the way to validate the model?
In order to create validation sets (to select the best ML algorithms and hyperparameters), the authors adopted an internal 5- fold validation.
Q7. What is the lexicographic choice for DL and XGB?
When considering both DL and XGB scenarios, the lexicographic choice favors rminer XGB for the binary classification and regression tasks, while AutoGluon DL is the selected tool for multi-class.
Q8. What is the scenario for the validation of the external test set?
Since neither Auto-Keras nor Auto-PyTorch natively support cross-validation during the fitting phase, the authors used a simpler holdout train (75%) and test (25%) set split to select and fit the models.
Q9. How many times did Auto-Sklearn require the maximum computational effort?
For GML, Auto-Sklearn always requires the maximum allowed computational effort (3,600 s), followed by TPOT (average of 858 s per external fold and dataset).
Q10. What is the tool in 3 of the datasets?
TransmogrifAI is the best tool in 3 of the datasets (churn, credit and qsar), also obtaining the best average AUC per dataset (88%).
Q11. What is the common use of Auto-PyTorch in the second DL?
Similarly to AutoKeras, the authors use Auto-PyTorch only in the second DL scenario.3) Auto-Sklearn: an AutoML library built on top of the Scikit-Learn ML framework.
Q12. Why was TPOT not included in the third comparison scenario?
TPOT was not included in the third comparison scenario (XGB, Section IV) because the tool does not allow the selection of a single algorithm, such as XGB.8) TransmogrifAI: an AutoML tool for structured data and that runs on top of Apache Spark [30].
Q13. What is the ML algorithm for the task type?
All ML algorithms (when available for the task type) were tested: AdaBoost (H = 4), Bernoulli (H = 2) and Multinomial NB (H = 2), Gaussian NB (H = 0), Decision Tree (DT) (H = 4), Extremely Randomized Trees (XRT) (H = 5), Gradient Boosting Machine (GBM) (H = 6), k-Nearest Neighbors (k-NN) (H = 3), Linear Discriminant Analysis (LDA) (H = 4), Linear SVM (LSVM) (H = 4), Kernel based SVM (KSVM) (H = 7), Passive Aggressive (H = 3), Quadratic Discriminant Analysis (QDA) (H = 2), Random Forest (RF) (H = 5) and a Multiple Linear Regression (MR) classifier (H = 10).
Q14. What is the tool for the regression tasks?
As for the regression tasks, the AutoML tool differences for each dataset are very small, corresponding to 1 pp in terms of NMAE for all three datasets.