Open SebieF opened 1 week ago
@SebieF Can we have more information about what must be contained inside the dialogue? Before that, it's hard to create a layout.
As a starter, I would suggest at least a dataset selection, a selection of the task to be done including the values to optimize (will discuss this this week in more detail with an example dataset), a model selection (Gaussian Processes, maybe Random Forest or other models later on) and some sort of Exploit vs. Explore value definition. For the latter two you can refer to: https://slamd-demo.herokuapp.com/materials/discovery
To execute the bayesian optimization pipeline, a dialog should be created that provides all the necessary configuration options. As BO is an iterative process, please think about if one dialog is enough, or multiple dialogs make more sense (e.g. one for starting the process and training the first model, another one for entering the experimental results and start a new iteration).
Acceptance criteria