How to include neural network models in regression problems by using Auto ML?

Question

Hi everyone,

I faced an issue in Azure ML, I do highly appreciate you giving me guidance.

I have a regression problem for which I would like to set up Auto ML in Azure ML Studio to find the best model describing my data. For this purpose, I adjusted AutoML settings like below:

automl_settings = {
    "n_cross_validations": 10,
    "primary_metric": 'normalized_root_mean_squared_error',
    "enable_early_stopping": True,
    "experiment_timeout_hours": 1,  
    "max_concurrent_iterations": 4,
    "max_cores_per_iteration": -1,
    "verbosity": logging.INFO,
    "enable_dnn" : True,
    "enable_voting_ensemble": True,
    "enable_stack_ensemble": True,
    "enable_tf" : True, 
    "iterations" : 1000
}

As I put "enable_tf" : True in the AutoML settings above, I expect that AutoML will include (deep) neural network (NN) models in the search for finding the best regression model. However, I check the models used through iterations (by monitoring the job section), I see that NNs are not included whatsoever.

Any advice on including NNs in the AutoML search for regression problems is much appreciated.

Cheers