Are sparse representations really relevant for image classification?

Recent years have seen an increasing interest in sparse representations for image classification and object recog- nition, probably motivated by evidence from the analysis of the primate visual cortex. It is still unclear, however, whether or not sparsity helps classification. In this pa- per we evaluate its impact on the recognition rate using a shallow modular architecture, adopting both standard fil- ter banks and filter banks learned in an unsupervised way. In our experiments on the CIFAR-10 and on the Caltech- 101 datasets, enforcing sparsity constraints actually does not improve recognition performance. This has an impor- tant practical impact in image descriptor design, as enforc- ing these constraints can have a heavy computational cost.