search

kdmalc / fl-gestures

Develop a personalized sequential federated learning algorithm for real-time gesture classification
1 stars 0 forks source link

Test having each gesture be its own point in the latent space #23

Open kdmalc opened 1 month ago

kdmalc commented 1 month ago

Obvious way is to flatten/reshape gesture into 1D, but then we lose all the temporal information (it just gets concatenated when flattened).

Can we instead use convolutions to reduce to 1D?

What do people do in the literature?