LSM6DSOX MLC - Access to datasets, decision tree and workflow to extend existing ST models

[SamuPre]

Hello,

I am working with the Machine Learning Core (MLC) of the LSM6DSOX and I would like to better understand how to extend or reuse existing ST models.

In particular, I am using preconfigured models available from ST (e.g. Activity Recognition), which are provided as .ucf files.

My goal is to:

• reuse an existing ST model as a baseline

• add my own labeled data (custom activities)

• generate a new combined model without starting completely from scratch

For this reason, I would like to ask:

1. Is it possible to access or obtain the original datasets used by ST to train the provided MLC models?

2. Is there a way to extract or inspect the underlying decision tree from a .ucf configuration?

3. Is there an official workflow to extend an existing model (e.g. retrain by adding new data) rather than creating a new model from zero?

I understand that .ucf files represent the final register configuration, but I am interested in working at a higher level (model/data level) if possible.

Any guidance or recommended approach would be greatly appreciated.

Thank you for your support.

[Federica Bossi]

Hi @SamuPre ,

Questions 1 and 2 are related to ST Confidential information that we can't share on ST community.

About question 3: No, there isn't an official workflow to extend an existing model.