We will study neuromorphic NLP by (1) better understanding the
language computation in the brain and (2) by building bio-inspired
real-time NLP systems. Such constraints are typically ignored by the
NLP community. Input to the NLP systems will use off-the-shelf speech
processing modules or text as input and output. We propose that the
NLP system be implemented in a spike-based architecture using PyNN and
Nest, Brian, or SpiNNaker.
Here is a version of
the working group proposal.