Holographic Reduced Representations: Convolution Algebra for Compositional Distributed Representations
Holographic Reduced Representations: Convolution Algebra for Compositional Distributed Representations
复制标题
全息简化表示：组合分布式表示的卷积代数
DOI：
--
复制
发表时间：
1991
期刊：
影响因子：
--
通讯作者：
T. Plate
中科院分区：
文献类型：
--
作者：
T. Plate
描述了使用分布式表示形式来表示综合结构结构的问题。该方法使用循环卷积将其关联，该项目由V ec-tors表示。可以将任意变量绑定，各个长度，帧和减少表示形式的短序列被压缩到XED宽度向量中。这些表示本身就是项目，可用于构建组成结构。可以通过使用具有良好重构属性的单独的关联内存来清理卷积记忆的嘈杂重构。
A solution to the problem of representing compo-sitional structure using distributed representations is described. The method uses circular convolution to associate items, which are represented by v ec-tors. Arbitrary variable bindings, short sequences of various lengths, frames, and reduced representations can be compressed into a xed width vector. These representations are items in their own right, and can be used in constructing compositional structures. The noisy reconstructions given by convolution memories can be cleaned up by using a separate associative memory that has good recon-structive properties.