Skip to content

[Question] Where can I find the initial set of candidate pipelines? #1757

@amorimlb

Description

@amorimlb

Hi, first of all, congratulations for such a useful and inspiring work!
I am working on a research project related to Auto-sklearn. Can you help me with the following information?
On the paper from Auto-sklearn 2.0, (https://arxiv.org/pdf/2007.04074) you describe in section 3.1 how you construct a portfolio given a finite set of candidate pipelines C. My question is where I can find this initial set of candidate pipelines. Are they organized somewhere in the directories here in the repository?

I have found some pipeline description in "configurations.csv" files, such as autosklearn/metalearning/files/accuracy_binary.classification_dense/configurations.csv, but I am not sure if these are the optimized portfolios or the initial set of candidate pipelines C, that you mention in the paper.

Thank you!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions