Learning Invariant Feature Hierarchies

By Yann LeCun. 2012

162 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Deep Learning: Methods and Applications (2014, background): “The power of the deep networks is shown to be their ability to extract appropriate features and do discrimination jointly [210].”
Emergence of Invariance and Disentanglement in Deep Representations (2017, methodology): “Other theoretical eﬀorts focus on complexity considerations, and explain the success of deep networks by ways of statistical or computational eﬃciency (Lee et al., 2017; Bengio et al., 2009; LeCun, 2012).”
A tutorial survey of architectures, algorithms, and applications for deep learning (2014, background): “Deep networks are shown to be powerful for computer vision and image recognition tasks because they extract appropriate features while jointly performing discrimination [192].”
Visual Representations: Defining Properties and Deep Approximations (2014, methodology): “atial regions, variously normalized and quantized. This process is repeated hierarchically in a convolutional neural network (CNN), with weights inferred from data, leading to representation learning [46,35,47,8,53,4]. There are more local descriptors than we can review here, and empirical compariso”
Automatic detection of invasive ductal carcinoma in whole slide images with convolutional neural networks (2014, background): “The motivation to do that are to accentuate differences between input features and accelerate gradient-based learning.(30)”
Deep learning for visual understanding: A review (2016, background): “Deep networks have been shown to be successful for computer vision tasks because they can extract appropriate features while jointly performing discrimination [9,13].”
Learning filter banks within a deep neural network framework (2013, background): “Deep Neural Networks are attractive because they have been shown to do feature extraction jointly with classification [2].”
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].”
Emergence of invariance and disentangling in deep representations (2017, background): “Other theoretical efforts focus on complexity considerations, and explain the success of deep networks by ways of statistical or computational efficiency (Lee et al., 2017; Bengio et al., 2009; LeCun, 2012).”
Automatic Analysis of Facial Affect: A Survey of Registration, Representation, and Recognition (2015, background): “There exists a variety of pooling techniques, such as binning features over local histograms, sampling the minimum or maximum value within a neighbourhood or computing the sum or average of the features across the neighbourhood [13], [14], [69].”
Deep convolutional neural networks for eigenvalue problems in mechanics (2018, background): “CNN architectures tend to be equivariant to general aﬃne input transformations of the nature of translations, rotations and small distortions [30, 31].”
Feature learning and deep architectures: new directions for music informatics (2013, background): “There is mathematical rigor behind the trade-off between depth and breadth of a processing structure (Bengio 2009; LeCun 2012), but a classic example ade- quately illustrates these principles.”
A Primal-Dual Method for Training Recurrent Neural Networks Constrained by the Echo-State Property (2013, methodology): “…paper, we build a “deep” version of the RNN by using an independent DNN to extract high-level features from the raw speech data, capitalizing on the well established evidence for the power of DNNs in carrying out automatic feature extraction (Hinton et al., 2012; Dahl et al., 2011; LeCun, 2012).”
Unsupervised Learning of Invariant Representations in Hierarchical Architectures (2013, background): “[50, 52, 47, 73, 6, 49]); it may provide even better performance by learning appropriate invariances from unsupervised experience instead of hard-wiring them.”
Deep learning for visual understanding (2017, unknown): “Deep networks have shown to be successful for computer vision tasks because they can extract appropriate features while jointly performing discrimination [15, 20].”