2013/01/15 by Felix Bauer, Roland Memisevic, Bauer, Felix +1 · 2 citations
Computer Science · Mathematics · Neuroscience · #Advanced Vision and Imaging #Computer science #FOS: Computer and information sciences #Feature (linguistics) #I.2.6 #Image Retrieval and Classification Techniques #Machine Learning (cs.LG) #Mathematics #Multiplicative function #Visual perception and processing mechanisms #cs.LG
paper · pdf · doi:10.48550/arxiv.1301.3391
published in arXiv (Cornell University) (Cornell University) · (new version:) added training formulae; added minor clarifications
openalex publication_date 2013/01/15 · arxiv created 2013/03/11 · arxiv updated 2013/03/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We present a feature learning model that learns to encode relationships between images. The model is defined as a Gated Boltzmann Machine, which is constrained such that hidden units that are nearby in space can gate each other's connections. We show how frequency/orientation "columns" as well as topographic filter maps follow naturally from training the model on image pairs. The model also helps explain why square-pooling models yield feature groups with similar grouping properties. Experimental results on synthetic image transformations show that spatially constrained gating is an effective way to reduce the number of parameters and thereby to regularize a transformation-learning model.