AI for Neuroscience: Dynamical VAE and its Application on Hippocampal Ensembles
Date:
- Represented the between-group distinguishability while maintaining the individual-level sequential structure with 98+% reconstruction accuracy, and revealed the unique dynamical moves under experiment conditions
- Visualized the latent dynamics of rats’ neuron activation during sequential events after compressing the noisy data by a Dynamical Variational Encoder based on Deep Markov Models and Recurrent Neural Networks