On the Turing Completeness of Modern Neural Network ArchitecturesDownload PDF

Sep 27, 2018 (edited Jan 10, 2019)ICLR 2019 Conference Blind SubmissionReaders: Everyone
  • Abstract: Alternatives to recurrent neural networks, in particular, architectures based on attention or convolutions, have been gaining momentum for processing input sequences. In spite of their relevance, the computational properties of these alternatives have not yet been fully explored. We study the computational power of two of the most paradigmatic architectures exemplifying these mechanisms: the Transformer (Vaswani et al., 2017) and the Neural GPU (Kaiser & Sutskever, 2016). We show both models to be Turing complete exclusively based on their capacity to compute and access internal dense representations of the data. In particular, neither the Transformer nor the Neural GPU requires access to an external memory to become Turing complete. Our study also reveals some minimal sets of elements needed to obtain these completeness results.
  • Keywords: Transformer, NeuralGPU, Turing completeness
  • TL;DR: We show that the Transformer architecture and the Neural GPU are Turing complete.
20 Replies