Bubbles: a unifying framework for low-level statistical properties of natural image sequences.

By Aapo Hyvärinen, J. Hurri, Jaakko J. Väyrynen. 2003

113 citations recorded by the source corpus.

Subfields: Biophysics, Cognitive Neuroscience, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biophysics

In Cognitive Neuroscience

In Signal Processing

Citations in the corpus

Nonlinear Extraction of Independent Components of Natural Images Using Radial Gaussianization (2009, background): “…arising from ICA (and related sparse coding methods): an explanation for oriented receptive ﬁelds found in primary visual cortical neurons (e.g., Olshausen & Field, 1996, 1997; Bell & Sejnowski, 1997; van Hateren & Ruderman, 1998; Hyv¨arinen & Hoyer, 2000; Hyv¨arinen, Hurri, & V¨ayrynen, 2003).”
Learning Intermediate-Level Representations of Form and Motion from Natural Movies (2012, methodology): “In previous modeling work, Olshausen and Field (1996) showed that when a neural system is adapted to the statistics of natural images so as to produce a sparse representation, the receptive fields that emerge are localized, oriented, and multiscale, in line with the response properties of simple…”
Modeling Multiscale Subbands of Photographic Images with Fields of Gaussian Scale Mixtures (2009, background): “The local GSM model that underlies the FoGSM is closely related to other local hidden variable models for images [16], [8], [43], [31], [44].”
Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding (2020, methodology): “For these distributions, we also report the ﬁt parameters for the ∆ area, ∆ x, ∆ y statistics in Tables 12, 13, and 14, respectively, where the shape parameter α of the generalized Laplacian is bolded.”
Deep Learning of Invariant Features via Simulated Fixations in Video (2012, background): “In prior work [10, 11, 12], a single layer of features learned using temporal slowness results in translation-invariant edge detectors, reminiscent of com plex-cells.”
Unsupervised learning of visual invariance with temporal coherence (2011, background): “Much prior work has been devoted to learning invariant featu res sing image sequences [13, 8, 14, 12].”
A review of unsupervised feature learning and deep learning for time-series modeling (2014, methodology): “Data from EEG (top two signals), EOG (third and fourth signal), and combines sparseness, temporal coherence, topography and complex cell pooling in a single model [58].”