Deep Direct Discriminative Decoders for High-dimensional Time-series Data Analysis
This provides a scalable solution for analyzing complex time-series data in fields like neuroscience, though it is an incremental improvement by combining existing methods.
The paper tackled the challenge of modeling high-dimensional time-series data with state-space models (SSMs) when observations deviate from normality, proposing a deep direct discriminative decoder (D4) that integrates deep neural networks into SSMs. It demonstrated improved performance over traditional SSMs and RNNs on simulated and real datasets, such as Lorenz attractors and rat hippocampus spiking neural data.
The state-space models (SSMs) are widely utilized in the analysis of time-series data. SSMs rely on an explicit definition of the state and observation processes. Characterizing these processes is not always easy and becomes a modeling challenge when the dimension of observed data grows or the observed data distribution deviates from the normal distribution. Here, we propose a new formulation of SSM for high-dimensional observation processes. We call this solution the deep direct discriminative decoder (D4). The D4 brings deep neural networks' expressiveness and scalability to the SSM formulation letting us build a novel solution that efficiently estimates the underlying state processes through high-dimensional observation signal. We demonstrate the D4 solutions in simulated and real data such as Lorenz attractors, Langevin dynamics, random walk dynamics, and rat hippocampus spiking neural data and show that the D4 performs better than traditional SSMs and RNNs. The D4 can be applied to a broader class of time-series data where the connection between high-dimensional observation and the underlying latent process is hard to characterize.