Unsupervised model to infer latent generative processes from point processes occuring in $\mathbb{R}^d$ spcae. Developed with focus for inferring neural states from waveform features of unclustered spike events.

• Contrastive Predictive Coding (CPC)
  • Works for continuous time rather than point processes
• Temporal Neural Networks (e.g. Hawkes process)
  • designed for categorically labeled events
• CEBRA
  • designed for sorted spiking data
  • semi-supervised definition of matching states