Neural Dynamics Discovery via Gaussian Process Recurrent Neural Networks
This work addresses the problem of accurately recovering dynamic portraits from neural data for researchers in neuroscience and machine learning, representing an incremental improvement with a novel hybrid method.
The paper tackled the challenge of extracting complex latent dynamics from high-dimensional noisy neural data by proposing a model that combines recurrent neural networks and Gaussian processes to capture nonlinear, non-Markovian dynamics and complex embeddings, outperforming state-of-the-art methods in reconstructing latent dynamics from simulated and experimental datasets, especially in low-sample scenarios.
Latent dynamics discovery is challenging in extracting complex dynamics from high-dimensional noisy neural data. Many dimensionality reduction methods have been widely adopted to extract low-dimensional, smooth and time-evolving latent trajectories. However, simple state transition structures, linear embedding assumptions, or inflexible inference networks impede the accurate recovery of dynamic portraits. In this paper, we propose a novel latent dynamic model that is capable of capturing nonlinear, non-Markovian, long short-term time-dependent dynamics via recurrent neural networks and tackling complex nonlinear embedding via non-parametric Gaussian process. Due to the complexity and intractability of the model and its inference, we also provide a powerful inference network with bi-directional long short-term memory networks that encode both past and future information into posterior distributions. In the experiment, we show that our model outperforms other state-of-the-art methods in reconstructing insightful latent dynamics from both simulated and experimental neural datasets with either Gaussian or Poisson observations, especially in the low-sample scenario. Our codes and additional materials are available at https://github.com/sheqi/GP-RNN_UAI2019.