Une procedure d'apprentissage pour reseau a seuil asymmetrique (A learning scheme for asymmetric threshold networks)

By Yann LeCun. 1985

Subfields: Model Architectures, Optimization & Training

How researchers see this paper

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

In Model Architectures

In Optimization & Training

Citations in the corpus

Natural Language Processing (Almost) from Scratch (2011, positive): “Following the classical “back-propagation” derivations (LeCun, 1985; Rumelhart et al., 1986) and the modular approach shown in (Bottou, 1991), any feed-forward neural network with L layers, like the ones shown in Figure 1 and Figure 2, can be seen as a composition of functions f l θ(·), correspondin”
Multi-column deep neural networks for image classification (2012, background): “(2) It was shown [14] that such multi-layered DNN are hard to train by standard gradient descent [36, 18, 28], the method of choice from a mathematical/algorithmic point of view.”
Deep, Big, Simple Neural Nets for Handwritten Digit Recognition (2010, background): “More than a decade ago, artiﬁcial neural networks called Multilayer Perceptrons or MLPs (Werbos, 1974; LeCun, 1985; Rumelhart et al., 1986) were among the ﬁrst clas-siﬁers tested on MNIST.”
Deep learning in neural networks: An overview (2014, background): “Related work was published several years later (LeCun, 1985, 1988; Parker, 1985).”
Weight Uncertainty in Neural Networks (2015, methodology): “Using Monte Carlo sampling to evaluate the expectations, a backpropagation-like (LeCun, 1985; Rumelhart et al., 1988) algorithm is obtained for variational Bayesian inference in neural networks – Bayes by Backprop – which uses unbiased estimates of gradients of the cost in (1) to learn a…”
Measuring Invariances in Deep Networks (2009, methodology): “Given a set of p input patterns x, i = 1, · · · , p, the weight matrices W1 and W2 are adapted using backpropagation [16, 17, 18] to minimize the reconstruction error ∑p i=1 ∥”
To recognize shapes, first learn to generate images. (2007, background): “Backpropagation was discovered independently by several different researchers (Bryson and Ho, 1975; Werbos, 1974; Parker, 1985; LeCun, 1985; Rumelhart et al., 1986) and it was the first effective way to learn neural networks that had one or more layers of adaptive hidden units.”
Reinforcement Learning: An Introduction (1998, methodology): “error-backpropagation stage (LeCun, 1985; Rumelhart, Hinton, and Williams, 1986) of learning by multi-layer ANNs, and the current deep-learning stage with its emphasis on representation learning (e.g., Bengio, Courville, and Vincent, 2012; Goodfellow, Bengio, and Courville, 2016).”
A Neural Syntactic Language Model (2005, methodology): “Using standard back-propagation (LeCun, 1985; Rumelhart, Hinton, & Williams, 1986; Werbos, 1974) it is straightforward to compute the gradient for every parameter of the model.”
Inside-Outside and Forward-Backward Algorithms Are Just Backprop (tutorial paper) (2016, background): “Back-propagation is popular for optimizing the parameters of neural networks, which involve nonlinear functions (LeCun, 1985; Rumelhart et al., 1986).”
Implicit and explicit representations (2009, unknown): “For example, the multi-layer perceptron [18, 23, 4] is a set of feed-forward layers where units from a layer possess directed connections to the subsequent layer.”
Weight Uncertainty in Neural Network (2015, methodology): “This leads to a simple approximate learning algorithm similar to backpropagation (LeCun, 1985; Rumelhart et al., 1988).”
Merging chrominance and luminance in early, medium, and late fusion using Convolutional Neural Networks (2015, methodology): “For example, the same concept was proposed in Werbos PhD thesis [Werbos, 1974], and was simultaneously applied by Yann Le Cun45 in his PhD thesis [LeCun, 1985; Le Cun, 1986] to propagate “virtual target values” and one year earlier by Parker in his report “Learning-logic” [Parker, 1985].”
Large Scale Machine Learning (2004, methodology): “The first one is the Multi Layer Perceptron (MLP), which became useful with the introduction of the back-propagation training algorithm by LeCun (1985) and Rumelhart et al. (1986). The second one, proposed more recently, is the Support Vector Machine (SVM) algorithm proposed by Vapnik (1995). SVMs s”
Forward Models: Supervised Learning with a Distal Teacher (1992, methodology): “of algorithms such as Boltzmann learning (Hinton & Sejnowski, 1986) and backpropagation (LeCun, 1985; Parker, 1985; Rumelhart, Hinton, & Williams, 1986, Werbos, 1974) have provided the means for training networks with adaptive nonlinear internal units.”