# Takens-inspired neuromorphic processor: a downsizing tool for random recurrent neural networks via feature extraction

## TL;DR

This paper introduces a Takens-inspired method to reduce the size of random recurrent neural networks for time series prediction by leveraging feature extraction and attractor reconstruction, resulting in a more efficient hybrid processor.

## Contribution

It presents a novel technique combining Takens-based attractor reconstruction with machine learning to minimize neural network size for time series prediction.

## Key findings

- Reduced neural network size by a factor of 15.
- Demonstrated stabilization of an arrhythmic neural model.
- Derived scaling laws for prediction error.

## Abstract

We describe a new technique which minimizes the amount of neurons in the hidden layer of a random recurrent neural network (rRNN) for time series prediction. Merging Takens-based attractor reconstruction methods with machine learning, we identify a mechanism for feature extraction that can be leveraged to lower the network size. We obtain criteria specific to the particular prediction task and derive the scaling law of the prediction error. The consequences of our theory are demonstrated by designing a Takens-inspired hybrid processor, which extends a rRNN with a priori designed delay external memory. Our hybrid architecture is therefore designed including both, real and virtual nodes. Via this symbiosis, we show performance of the hybrid processor by stabilizing an arrhythmic neural model. Thanks to our obtained design rules, we can reduce the stabilizing neural network's size by a factor of 15 with respect to a standard system.