Dataflow Matrix Machines as a Model of Computations with Linear Streams
arXiv:1706.00648v11 citations
Originality Synthesis-oriented
AI Analysis
This provides a new programming platform for computations with linear streams, but it appears incremental as an extension of existing neural network models.
The paper introduces dataflow matrix machines as a Turing-complete generalization of recurrent neural networks, using a vector space of finite prefix trees with numerical leaves to combine expressive power with simplicity.
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.