Nonlinear image representation using divisive normalization

By Siwei Lyu, Eero P. Simoncelli. 2008

171 citations recorded by the source corpus.

Subfields: Biophysics, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

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 Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

Group Normalization (2018, neutral): “Normalization layers in deep networks had been widely used before the development of BN. Local Response Nor-malization (LRN) [40, 28, 32] was a component in AlexNet [32] and following models [64, 53, 58].”
Building high-level features using large scale unsupervised learning (2011, methodology): “This exper-iment shows that patches, being detected as faces by the OpenCV face detector, account for less than 3% of the 100,000 sampled patches”
Density Modeling of Images using a Generalized Normalization Transformation (2015, methodology): “Finally, we ﬁnd that the GDN transformation can be efﬁciently inverted using a ﬁxed point iteration: Other iterative inverse solutions have been proposed for this purpose (Malo et al., 2006; Lyu & Simoncelli, 2008), but these only apply to special cases of the form in eq.”
Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition (2012, methodology): “They used a combination of a rectified sigmoid followed by subtractive and divisive local normalization inspired by computational neuroscience models of vision (Lyu and Simoncelli, 2008; Pinto et al., 2008).”
Convolutional networks and applications in vision (2010, background): “layer is inspired by visual neuroscience models [3], [4].”
Blind Image Quality Assessment Based on High Order Statistics Aggregation (2016, background): “This normalization can be also regarded as a simpler version of divisive normalization transform (DNT) [38] to mimic the early nonlinear processing in human visual system (HVS) and reduce redundancies in local features.”
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Robust Semantic Pixel-Wise Labelling (2015, methodology): “We perform local contrast normalization (LCN) as a pre-processing step to the input [23, 15].”
Blind Image Quality Assessment Using Joint Statistics of Gradient Magnitude and Laplacian Features (2014, methodology): “As in [38] and [39], we propose to normalize the GM and LOG coefficients to obtain stable statistical image representations.”
What is the best multi-stage architecture for object recognition? (2009, methodology): “The local contrast normalization layer is inspired by computational neuroscience models [24, 20].”
Traffic sign recognition with multi-scale Convolutional Networks (2011, methodology): “The local normalization operations are inspired by visual neuroscience models [13], [14].”
Positional Normalization (2019, methodology): “Normalization methods for neural networks can be roughly categorized into two regimes: normalization of weights [50, 55, 59, 73] and normalization of activations [29, 31, 37, 41, 47, 49, 61, 68, 74].”
Representation learning for mammography mass lesion classification with convolutional neural networks (2016, background): “It has been empirically shown that such divisive normalization reduces statistical dependencies [28,25], which in turn accentuates differences between input features and accelerates gradientbased learning [29].”
Kernel Methods for Deep Learning (2009, methodology): “Computation in infinite neural networks was first studied in the context of Gaussian processes (Neal, 1996). In later work, Williams (1998) derived analytical forms for kernel functions that mimicked the computation in large networks.”