Prediction as a candidate for learning deep hierarchical models of data

By Rasmus Berg Palm. 2012

444 citations recorded by the source corpus.

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

A Unified Gradient Regularization Family for Adversarial Examples (2015, background): “ACKNOWLEDGMENT The authors would like to thank the developer of Deeplearning Toolbox [21], the developers of Theano [11], [22] and the developers of Pylearn2 [20].”
MatConvNet: Convolutional Neural Networks for MATLAB (2014, methodology): “ter exposes individual CNN building blocks as MATLAB functions, as well as integrating with the native MATLAB GPU support, which makes it very convenient for fast prototyping. The DeepLearningToolbox [8] is a MATLAB toolbox implementing, among others, CNNs, but it does not seem to have been tested o”
Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning (2016, background): “Here, we explore another potential principle for exploiting video for unsupervised learning: prediction of future image frames [16, 28, 32, 33, 34, 46, 48].”
Low-Rank Approximations for Conditional Feedforward Computation in Deep Neural Networks (2013, methodology): “The neural network is built using Rasmus Berg Palm’s Deep Learning Toolbox [16].”
Empirical Mode Decomposition based ensemble deep learning for load demand time series forecasting (2017, methodology): “The scaling formula is: To implement the simulation, LIBSVM was used for the SVR model [54], while deep learning toolbox neural networks, including ANN, DBN, EDBN [27], EMD-ANN and method [55].”
Computer aided lung cancer diagnosis with deep learning algorithms (2016, methodology): “The architecture of our CNN contains 8 layers, and except input and output layer, every odd number layer is a convolution layer and every even number layer is a pooling and subsampling layer [6].”
Using machine learning for medium frequency derivative portfolio trading (2015, methodology): “We use scikit [11] implementation of SVM with a Gaussian Kernel, own implementation of logistic regression and implementation of [10] for neural network.”
An up-to-date comparison of state-of-the-art classification algorithms (2017, methodology): “For DL, we test three different methods, which are Neural Networks, DBN, implemented in Matlab using the Deep Learning Tool-box 6 ( Palm, 2012 ), and CNN, implemented in Python using the Pylearn2 7 library ( Goodfellow et al., 2013 ).”
Human Action Recognition Based on MOCAP Information Using Convolution Neural Networks (2014, methodology): “A typical CNN classifier architecture used as a classifier [14] consists of an alternating sequence of convolution and subsampling layers followed by a neural network for classification.”
Spectral-Spatial Classification of Hyperspectral Imagery with 3D Convolutional Neural Network (2017, unknown): “SAE-LR and DBN-LR were implemented with the MATLAB deep learning toolbox [45].”
Unsupervised Learning of Visual Structure using Predictive Generative Networks (2015, background): “In fact, Palm (2012) coined the term “Predictive Encoder” to refer to an autoencoder that is trained to predict future input instead of reconstructing current stimuli.”
Financial Time-Series Data Analysis Using Deep Convolutional Neural Networks (2016, unknown): “[29]- [30] used finance newspaper textual data to predict the price movement under deep learning.”
Robust Sound Event Classification Using Deep Neural Networks (2015, unknown): “The DNN classifier [37], when evaluated using SAI features, simply replaces the SVMclassifier, with the same input features.”
Highly accurate sequence-based prediction of half-sphere exposures of amino acid residues in proteins (2016, methodology): “In this study, we employed the deep neural network from the MATLAB toolbox, implemented by Palm (2012).”
Fast-classifying, high-accuracy spiking deep networks through weight and threshold balancing (2015, methodology): “The code used to train and convert the networks in this paper is a modification of the Matlab DeepLearnToolbox [25] and can be found online.”
Spatio-Temporal AutoEncoder for Video Anomaly Detection (2017, methodology): “Inspired by the previous works [11, 15, 18, 25, 29] which prove that predictive networks benefit learning video representations, we design a prediction branch in the decoder part to predict the future T frames after the input video clip.”
A novel deep learning approach for classification of EEG motor imagery signals (2017, methodology): “Matlab deep learning toolbox [35] was used for designing and testing the proposed networks.”