Neocognitron: A new algorithm for pattern recognition tolerant of deformations and shifts in position

By Kunihiko Fukushima, S. Miyake. 1982

How researchers see this paper

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

Citations in the corpus

Weight Agnostic Neural Networks (2019, background): “Convolutional networks [24, 56] are especially suited for image processing [16].”
Practical Black-Box Attacks against Machine Learning (2016, methodology): “set, 100 are used for training and 50 for validation. The architecture of our substitute DNN is a standard one for computer vision classiﬁcation tasks: a convolutional layer of 32 kernels of size 2x2 [9, 15], a max pooling layer of kernel 2x2 [24], a convolutional layer of 64 kernels of size 2x2, a ”
The hardware lottery (2020, background): “…been in place for a few decades: backpropagation (invented in 1963 (K & Piske, 1963), reinvented in 1976 (Linnainmaa, 1976), and again in 1988 (Rumelhart et al., 1988)), deep convolutional neural networks ((Fukushima & Miyake, 1982), paired with backpropagation in 1989 (LeCun et al., 1989)).”
Representation Learning: A Review and New Perspectives (2012, background): “riance to input translations, and this style of architecture (alternating selective feature extraction and invariance-creating pooling) has been the basis of convolutional networks, the Neocognitron (Fukushima and Miyake, 1982) and HMAX (Riesenhuber and Poggio, 1999) models, and argued to be the arc”
Vehicle Type Classification Using a Semisupervised Convolutional Neural Network (2015, background): “A small size saves time, but may lose too much information, which leads to low classification accuracy.”
Gradient-based learning applied to document recognition (1998, background): “Each unit in the third hidden layer in (cid:12)g-ure 2 may have input connections from several feature maps in the previous layer.”
Factored 3-Way Restricted Boltzmann Machines For Modeling Natural Images (2010, background): “…in vision applications: The outputs of linear ﬁlters are passed through a rectifying non-linearity and then similar ﬁlters have their outputs pooled by a “complex” cell which exhibits some local invariance due to the pooling (Fukushima and Miyake, 1982, LeCun et al., 1998, Serre et al., 2005).”
A Survey of the Usages of Deep Learning for Natural Language Processing (2020, background): “networks (CNNs) [14], [15], built upon Fukashima’s neocognitron [16], [17], derive the name from the convolution operation in mathematics and signal processing.”
Neocognitron: A hierarchical neural network capable of visual pattern recognition (1988, background): “As a result of this approach, a pattern-recognition system called the "neocognitron" has been developed (Fukushima, 1980; Fukushima & Miyake, 1982).”
Learning Invariant Feature Hierarchies (2012, methodology): “A number of similar models have been proposed with different learning algorithms, starting with Fukushima’s Neocognitron [14], and several others in the last 10 years [15,16].”
Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition (2007, background): “When multiple such levels are stacked, the resulting architecture is essentially identical to the Neocognitron [7], the Convolutional Network [13, 10], and the HMAX, or so-called “Standard Model” architecture [20, 17].”
1D Convolutional Neural Networks and Applications: A Survey (2019, background): “With these advances in neurocognitive science, Fukushima and Miyake [23] in 1982 proposed the predecessor of Convolutional Neural Networks (CNNs), at the time called as “Neocognitron” which is a self-organized, hierarchical network and has the capability to recognize stimulus patterns based on the d”
Foundations of Vision (2018, methodology): “This hierarchical model, referred to as HMAX, relies on a multilayered architecture (cf. Fukushima & Miyake, 1982) that capitalizes on the functional architecture of V1 simple cells and complex cells (Figure 1.17).”
Invariant Visual Object and Face Recognition: Neural and Computational Bases, and a Model, VisNet (2012, background): “The computational solution proposed to the superset/subset aspect of the binding problem will apply in principle to other multilayer competitive networks, although the issues considered here have not been explicitly addressed in architectures such as the Neocognitron (Fukushima and Miyake, 1982).”
Deep Convolutional Neural Networks for Image Classification: A Comprehensive Review (2017, background): “These initial CNNs were reminiscent of the neocognitron (Fukushima, 1979, 1980; Fukushima & Miyake, 1982).”
Large-scale Learning with SVM and Convolutional for Generic Object Categorization (2006, methodology): “These architectures are all based on stacking multiple layers of the following modules [9, 13]: feature extraction through multiple convolutional filters, point-wise nonlinearities, and spatial subsampling.”