Multi Session Training Across Multiple Animals

[Barbod1380]

Assume we have 6 mice, and we are showing them 2 images, called image 'A' and image 'B'. In the demos, we trained models across animals, but this was for scenarios where we only had a single image (single manifold construction) and compared it across animals, as far as I understand.

One option is to create two different multi-session models: one for image 'A' and one for image 'B'. However, the problem is that I want to capture the relationship between these two manifolds. With two separate models, the constructed manifolds will change every time you run the algorithm, making it difficult to compare them properly.

Are there any options for addressing this problem?
Thanks.