Modelling Higher Cognitive Functions with Hebbian Cell Assemblies
Modelling Higher Cognitive Functions with Hebbian Cell Assemblies
复制标题
使用赫布细胞组件模拟高级认知功能
DOI：
--
复制
发表时间：
1999
期刊：
影响因子：
--
通讯作者：
M. Chady
中科院分区：
文献类型：
--
作者：
M. Chady
这项工作的目标是开发一个模型，更高的认知行为使用连接主义范式。在阿尔蒂神经网络领域内的这种工作的例子是非常稀少的，主要是由于在神经硬件中实现符号处理的困难。像组合性和变量绑定这样的问题，以及选择合适的表示方案，是实现超越简单模式识别的智能行为的主要障碍。尽管存在能够进行高级推理的连接主义系统（参见。Shastri和Ajjanagadde [9]，或Barnden [2]），他们总是妥协他们的灵活性，最重要的是，他们的学习能力。这种系统中的所有连接都是固定的，设计者已经仔细地预先安排好了。我们想要实现的是一个自组织系统，它能够从连续的输入数据流中学习，并根据以前的经验做出概括性的预测。
The objective of this work is to develop a model of higher cognitive behaviour using the connectionist paradigm. Examples of such work within the eld of arti cial neural networks are exceptionally scarce, mainly due to the di culty of implementing symbol processing in neural hardware. Problems like compositionality and variable binding, as well as the selection of a suitable representation scheme, are a major obstacle to achieving the kind of intelligent behaviour which would extend beyond simple pattern recognition. Although there are connectionist systems which are capable of advanced inferencing (cf. Shastri and Ajjanagadde [9], or Barnden [2]), they always compromise their exibility and, most importantly, their capability to learn. All connections in such systems are xed, having been carefully prearranged by the designer. What we want to achieve is a self-organising system which is able to learn from a continuous stream of input data and make generalising predictions based on its previous experience.