# Dataflow Matrix Machines as a Model of Computations with Linear Streams

## TL;DR

This paper presents dataflow matrix machines as a versatile, Turing complete model that generalizes recurrent neural networks, utilizing vector spaces of prefix trees to enhance expressiveness while maintaining simplicity.

## Contribution

It introduces a novel framework combining dataflow matrix machines with prefix tree vector spaces, expanding the expressive power of neural network models.

## Key findings

- Dataflow matrix machines are Turing complete.
- The vector space of prefix trees enables complex data representations.
- The approach simplifies the implementation of expressive neural models.

## Abstract

We overview dataflow matrix machines as a Turing complete generalization of recurrent neural networks and as a programming platform. We describe vector space of finite prefix trees with numerical leaves which allows us to combine expressive power of dataflow matrix machines with simplicity of traditional recurrent neural networks.