Developed by @coopss
- Version:
in-dev-03/18 - Latest Stable:
none
This is an implementation of a typing autocompletion algorithm that exploits the properties of stacked bidirectional neural networks. This was designed to allow for continuous adaptation to the user's behavior via retraining. The inherent properties of associative memory in stacked BRNNs will allow for diminishing value of the pretrained word completions to accommodate for the user's unique usage.