Exam DP-100 topic 1 question 36 discussion

Random seed does not help in tuning accuracy, it just make experiments reproducible. I think it should not be used. https://learn.microsoft.com/en-us/azure/machine-learning/component-reference/two-class-neural-network "For Random number seed, type a value to use as the seed. Specifying a seed value is useful when you want to ensure repeatability across runs of the same pipeline. Otherwise, a system clock value is used as the seed, which can cause slightly different results each time you run the pipeline."

Random seed is a parameter for binary classification, but I do not understand "Hyperparameters" has anything to do with it???

Could anyone please provide insight into why one might choose not to use Optimization Tolerance? No question with selecting The Initial learning weights diameter & Number of learning iterations. Random seed shall not be selected since it does not optimise during the training process but merely ensure consistent results.

Answer is correct https://learn.microsoft.com/en-us/azure/machine-learning/component-reference/two-class-neural-network?view=azureml-api-2#how-to-configure

Chat GPT answer:The Two-Class Neural Network module has the following valid parameters: The Initial learning weights diameter: This parameter controls the magnitude of the weights in the neural network at the start of the training process. Lambda: This parameter controls the amount of regularization applied to the neural network during training. Number of learning iterations: This parameter controls the number of iterations performed during the training process.

https://learn.microsoft.com/en-us/azure/machine-learning/component-reference/two-class-neural-network Correct