早期的AutoML框架侧重于优化传统的机器学习管道及其超参数,而AutoML的另一个趋势是侧重于神经架构搜索。为了将这两个世界的优点结合起来,我们开发了Auto-PyTorch,它联合且鲁棒地优化网络架构和训练超参数,以实现全自动化深度学习(AutoDL)。Auto-PyTorch通过将多保真度优化与用于深度神经网络(DNN)的预热和集成的组合构建以及用于表格数据的常见基线相结合,在多个表格基准上取得了最先进的性能。
该API灵感来源于auto-sklearn,只需少量输入即可在给定数据集上拟合DL管道
>>> autoPyTorch = AutoNetClassification("tiny_cs", max_runtime=300, min_budget=30, max_budget=90) >>> autoPyTorch.fit(X_train, y_train, validation_split=0.3) >>> y_pred = autoPyTorch.predict(X_test) >>> print("Accuracy score", sklearn.metrics.accuracy_score(y_test, y_pred))
如果您对Auto-PyTorch感兴趣,可以在此处找到我们的开源实现
参考文献
- Lucas Zimmer, Marius Lindauer and Frank Hutter: Auto-PyTorch Tabular: Multi-Fidelity Meta Learning for Efficient and Robust AutoDL In: IEEE Transactions on Pattern Analysis and Machine Intelligence. 2021
- Mendoza, Hector and Klein, Aaron and Feurer, Matthias and Springenberg, Jost Tobias and Urban, Matthias and Burkart, Michael and Dippel, Max and Lindauer, Marius and Hutter, Frank: Towards Automatically-Tuned Deep Neural Networks In: AutoML: Methods, Sytems, Challenge. 2019

