Learning binary or real-valued time-series via spike-timing dependent plasticity
This work addresses time-series prediction for computational applications, but it is incremental as it builds on existing DyBM and VAR models with modifications.
The authors tackled the problem of learning binary or real-valued time-series by relaxing constraints in a dynamic Boltzmann machine (DyBM) to make it more suitable for computation and learning, showing that it can be considered as logistic regression for binary data and extending it to a Gaussian DyBM for real-valued data, which significantly improves predictive accuracy over the vector autoregressive (VAR) model in numerical experiments.
A dynamic Boltzmann machine (DyBM) has been proposed as a model of a spiking neural network, and its learning rule of maximizing the log-likelihood of given time-series has been shown to exhibit key properties of spike-timing dependent plasticity (STDP), which had been postulated and experimentally confirmed in the field of neuroscience as a learning rule that refines the Hebbian rule. Here, we relax some of the constraints in the DyBM in a way that it becomes more suitable for computation and learning. We show that learning the DyBM can be considered as logistic regression for binary-valued time-series. We also show how the DyBM can learn real-valued data in the form of a Gaussian DyBM and discuss its relation to the vector autoregressive (VAR) model. The Gaussian DyBM extends the VAR by using additional explanatory variables, which correspond to the eligibility traces of the DyBM and capture long term dependency of the time-series. Numerical experiments show that the Gaussian DyBM significantly improves the predictive accuracy over VAR.