Speaker: Tiberiu Tesileanu, Associate Research Scientist, Neuroscience, CCN
Topic: Biologically plausible circuits for segmenting time-series data
The signals reaching the brain typically exhibit significant temporal correlations. These correlations can be used to segment the signals in a content-independent way, in the sense that the dynamics that generated them rather than the specific realization of the signal is used for segmentation. This in turn helps the brain predict future sensory inputs, and sheds light on environmental changes. I will talk about two algorithms that we developed that perform this segmentation task and are biologically plausible, which we define as acting in a streaming setting and all learning rules being local. One algorithm is model-based and can be derived from an optimization problem involving a mixture of autoregressive processes. This algorithm relies on feedback in the form of a prediction error, and can also be used for forecasting future samples. In some brain regions, such as the retina, the feedback connections necessary to use the prediction error for learning are absent. For this case, I will describe a second, model-free algorithm that uses a running estimate of the autocorrelation structure of the signal to perform the segmentation. Both algorithms do well when tasked with segmenting signals drawn from autoregressive models with piecewise-continuous parameters. In particular, the segmentation accuracy is similar to that obtained from an oracle-like method in which the ground-truth parameters of the autoregressive models are considered known.