Multilabel Classification with AutoSklearn2Classifier - sgd component not valid error

[anthonyromyn]

Hello Auto-sklearn team,

Describe the bug

AutoSklearn2Classifier with y-variable of type multilabel-indicator erroring with component sgd. Demonstrated below with example data.

To Reproduce

Steps to reproduce the behavior:

1. Using the sample multilabel classification data from https://automl.github.io/auto-sklearn/master/examples/20_basic/example_multilabel_classification.html
2. autosklearn.classification.AutoSklearnClassifier is working fine with y-variable of type multilabel-indicator
3. Importing AutoSklearn2Classifier (from autosklearn.experimental.askl2 import AutoSklearn2Classifier)
4. AutoSklearn2Classifier works fine with y-variable of type binary (e.g. using just the first column of the y-matrix)
5. but AutoSklearn2Classifier with y-variable of type multilabel-indicator is producing the error below

Actual behavior, stacktrace or logfile

Text-based error: ValueError: The provided component 'sgd' for the key 'classifier' in the 'include' argument is not valid. The supported components for the step 'classifier' for this task are ['bernoulli_nb', 'decision_tree', 'extra_trees', 'gaussian_nb', 'k_nearest_neighbors', 'lda', 'liblinear_svc', 'mlp', 'multinomial_nb', 'passive_aggressive', 'qda', 'random_forest']

Environment and installation:

• Google Colab Jupyter Notebook
• Python 3.8.16
• Auto-sklearn version 0.15.0

Best, Anthony