A hierarchical framework for object recognition

Object recognition in the presence of background clutter and distractors is a central problem both in neuroscience and in machine learning. However, the performance level of the models that are inspired by cortical mechanisms, including deep networks such as convolutional neural networks and deep belief networks, is shown to significantly decrease in the presence of noise and background objects [19, 24]. Here we develop a computational framework that is hierarchical, relies heavily on key properties of the visual cortex including mid-level feature selectivity in visual area V4 and Inferotemporal cortex (IT) [4, 9, 12, 18], high degrees of selectivity and invariance in IT [13, 17, 18] and the prior knowledge that is built into cortical circuits (such as the emergence of edge detector neurons in primary visual cortex before the onset of the visual experience) [1, 21], and addresses the problem of object recognition in the presence of background noise and distractors. Our approach is specifically designed to address large deformations, allows flexible communication between different layers of representation and learns highly selective filters from a small number of training examples.