This paper describes a PP-attachment ambiguity resolution with a layered neural network which have function of learning vectorial symbol representations. The proposed model does not update only link weight but also vectorial symbol representations. We show qualitative difference between a proposed model and an ordinary layered neural network, which has more hidden units (i.e. more parameters) to have more flexibility but does not update symbol representations.
|ジャーナル||Research Reports on Information Science and Electrical Engineering of Kyushu University|
|出版ステータス||出版済み - 3 1 2005|
All Science Journal Classification (ASJC) codes
- Computer Science(all)
- Electrical and Electronic Engineering