The benefits of integrating attention and object recognition are investigated.While attention is frequently modeled as pre-processor for recognition, we investigate the hypothesis that attention is an intrinsic component of recognition and vice-versa.This hypothesis is tested with a recognitionmodel, the hierarchical discriminant saliency network (HDSN), whose layers are top-down saliency detectors, tuned for a visual class according to the principles of discriminant saliency.
The HDSN has two possible implementations.In a 3 piece horse wall art biologically plausible implementation, all layers comply with the standard neurophysiological model of visual cortex, with sub-layers of simple and complex units that implement a combination of filtering, divisive normalization, pooling, and non-linearities.In a neuralnetwork implementation, all layers are convolutional and implement acombination of filtering, rectification, and pooling.
The rectificationis performed with a parametric extension of the now popular rectified linearunits (ReLUs), whose parameters can be tuned for the detection of targetobject classes.This enables a number of functional enhancementsover neural network models that lack a connection to saliency, including optimal feature denoising mechanisms for baseball scoreboards for sale recognition, modulation ofsaliency responses by the discriminant power of the underlying features,and the ability to detect both feature presence and absence.In either implementation, each layer has a precise statistical interpretation, and all parameters are tuned by statistical learning.
Each saliency detection layer learns more discriminant saliency templates than its predecessors and higher layers have larger pooling fields.This enables the HDSN to simultaneously achieve high selectivity totarget object classes and invariance.The resulting performance demonstrates benefits for all the functional enhancements of the HDSN.