AutoML : TensorFlowDNN and TensorFlowLinearRegressor are blacklisted by default

Question

Hello,

I am running an AutoML experiment for a regression task, and looking at the YAML file which is generated it seems that TensorFlowLinearRegressor and TensorFlowDNN models are listed as both 'supported_models' and 'blacklist_algos'.

I tried to deactivate the automatic blacklisting of models by specifying the parameter 'auto_blacklist' to False, and 'blacklist_models' and 'blacklist_algos' parameters to Null, but it doesn't change anything.

automl_settings = {  
    "primary_metric": 'normalized_mean_absolute_error',  
    "featurization": 'auto',  
    "verbosity": logging.INFO,  
    "n_cross_validations": 5,  
    "auto_blacklist": False,  
    "blacklist_models": None,  
    "blacklist_algos": None  
}  
run = experiment.submit(automl_config, show_output=True)  

The generated YAML file (excerpt):

"whitelist_models":null,  
"blacklist_algos":["TensorFlowDNN","TensorFlowLinearRegressor"],  
"supported_models":["ElasticNet","GradientBoosting","LightGBM","TensorFlowLinearRegressor","TensorFlowDNN","LassoLars","DecisionTree","RandomForest","FastLinearRegressor","OnlineGradientDescentRegressor","ExtremeRandomTrees","TabnetRegressor","XGBoostRegressor","KNN","SGD"],  
"private_models":[],  
"auto_blacklist":false  

Maybe the problem comes from the fact that Deep learning is set to 'Disabled' in the configuration settings, as shown on the following picture:

Are deep learning models not supported anymore by AutoML?

Answer 1

@ThierryL-3166 Thanks for the question.

As mentioned in the below document The following support models in AutoML TensorFlowDNN, TensorFlowLinearRegressor are deprecated.
https://learn.microsoft.com/en-us/python/api/azureml-automl-core/azureml.automl.core.shared.constants.supportedmodels.regression?view=azure-ml-py