mikeriess / synbps Goto Github PK

Current HOMC implementation is hardcoded up to h = 4. This should be made scalable to larger values even though these processes will be extremely complex and heavy on memory.

A good clue on the pattern to be automated is found in homc_helpers.py

The current example is too messy and the aim of the framework should ideally be to abstract away the simulation part, allowing more focus on the models themselves.

A function: Run_experiments() should therefore be defined, as a wrapper the for loop and data collection.

• user should be able to specify eval metrics - OR - define a function themselves, with specific output type
• user should pass the model pipeline as a function

Add ML example to example notebook to make it easier for users to see how the framework is supposed to be used.

This should ideally be a RNN, as this model type is the main motivation for this framework.