Are cortical models really bound by the "binding problem"?
Are cortical models really bound by the "binding problem"?
复制标题
DOI：
10.1016/s0896-6273(00)80824-7
复制
发表时间：
1999-09-01
期刊：
影响因子：
16.2
通讯作者：
Poggio, T
中科院分区：
文献类型：
--
作者：
Riesenhuber, M;Poggio, T
视觉信息在皮层中的处理通常被描述为Hubel和wiesel假设的从简单到复杂的层次结构的延伸——一个由越来越复杂和不变的神经元表征组成的前馈序列。近年来，一些研究人员对这类模型在混乱场景中执行高级视觉处理（如视点不变物体识别）的能力提出了质疑，他们反过来提出了另一类基于皮质区域内和跨皮质区域内大细胞集合同步的模型。这种新颖而有争议的观点的主要隐含论据是，分层模型不能处理高级视觉的计算需求，并遭受所谓的“绑定问题”。在这里，我们回顾了现状，并讨论了理论和实验证据，表明层次模型的感知弱点是未经证实的。特别是，我们在这里展示了在混乱的场景中识别多个物体，可以说是视觉中最困难的任务之一，可以在分层前馈模型中完成。有两个问题使物体识别变得特别困难：
Processing of visual information in cortex is usually described in terms of an extension of the simple-to-complex hierarchy postulated by Hubel and Wiesel—a feed forward sequence of more and more complex and invariant neuronal representations. The capability of this class of models to perform higher-level visual processing such as viewpoint-invariant object recognition in cluttered scenes has been questioned in recent years by several researchers, who in turn proposed an alternative class of models based on the synchronization of large assemblies of cells, within and across cortical areas. The main implicit argument for this novel and controversial view was the assumption that hierarchical models cannot deal with the computational requirements of high-level vision and suffer from the so-called “binding problem.” Here, we review the present situation and discuss theoretical and experimental evidence showing that the perceived weaknesses of hierarchical models are unsubstantiated. In particular, we show here that recognition of multiple objects in cluttered scenes, arguably among the most difficult tasks in vision, can be done in a hierarchical feedforward model. Two problems in particular make object recognition difficult: