Deep learning
TL;DR: Deep learning is making major advances in solving problems that have resisted the best attempts of the artificial intelligence community for many years, and will have many more successes in the near future because it requires very little engineering by hand and can easily take advantage of increases in the amount of available computation and data.
Abstract: Deep learning allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction. These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition, object detection and many other domains such as drug discovery and genomics. Deep learning discovers intricate structure in large data sets by using the backpropagation algorithm to indicate how a machine should change its internal parameters that are used to compute the representation in each layer from the representation in the previous layer. Deep convolutional nets have brought about breakthroughs in processing images, video, speech and audio, whereas recurrent nets have shone light on sequential data such as text and speech.
Citations
More filters
TL;DR: A convolutional neural network (CNN) is developed that detects intraretinal fluid (IRF) on OCT in a manner indistinguishable from clinicians and can be trained to perform automated segmentations of clinically relevant image features.
Abstract: Evaluation of clinical images is essential for diagnosis in many specialties. Therefore the development of computer vision algorithms to help analyze biomedical images will be important. In ophthalmology, optical coherence tomography (OCT) is critical for managing retinal conditions. We developed a convolutional neural network (CNN) that detects intraretinal fluid (IRF) on OCT in a manner indistinguishable from clinicians. Using 1,289 OCT images, the CNN segmented images with a 0.911 cross-validated Dice coefficient, compared with segmentations by experts. Additionally, the agreement between experts and between experts and CNN were similar. Our results reveal that CNN can be trained to perform automated segmentations of clinically relevant image features.
275 citations
TL;DR: An adaptive multi-sensor data fusion method based on deep convolutional neural networks (DCNN) for fault diagnosis that can learn features from raw data and optimize a combination of different fusion levels adaptively to satisfy the requirements of any fault diagnosis task.
Abstract: A fault diagnosis approach based on multi-sensor data fusion is a promising tool to deal with complicated damage detection problems of mechanical systems. Nevertheless, this approach suffers from two challenges, which are (1) the feature extraction from various types of sensory data and (2) the selection of a suitable fusion level. It is usually difficult to choose an optimal feature or fusion level for a specific fault diagnosis task, and extensive domain expertise and human labor are also highly required during these selections. To address these two challenges, we propose an adaptive multi-sensor data fusion method based on deep convolutional neural networks (DCNN) for fault diagnosis. The proposed method can learn features from raw data and optimize a combination of different fusion levels adaptively to satisfy the requirements of any fault diagnosis task. The proposed method is tested through a planetary gearbox test rig. Handcraft features, manual-selected fusion levels, single sensory data, and two traditional intelligent models, back-propagation neural networks (BPNN) and a support vector machine (SVM), are used as comparisons in the experiment. The results demonstrate that the proposed method is able to detect the conditions of the planetary gearbox effectively with the best diagnosis accuracy among all comparative methods in the experiment.
275 citations
TL;DR: In this article, a compositing approach is optimized for narrow temporal-intervals and allows the derivation of time-series of consistent reflectance composites that capture field level phenologies.
275 citations
TL;DR: In this paper, a theory that encompasses general convolutional transforms, or in more technical parlance, general semi-discrete frames (including Weyl-Heisenberg filters, curvelets, shearlets, ridgelets, wavelets, and learned filters), general Lipschitz-continuous nonlinearities (e.g., rectified linear units, shifted logistic sigmoids, hyperbolic tangents, and modulus functions), and general pooling operators emulating, e.g. sub-sampling and averaging.
Abstract: Deep convolutional neural networks (DCNNs) have led to breakthrough results in numerous practical machine learning tasks, such as classification of images in the ImageNet data set, control-policy-learning to play Atari games or the board game Go, and image captioning. Many of these applications first perform feature extraction and then feed the results thereof into a classifier. The mathematical analysis of DCNNs for feature extraction was initiated by Mallat, 2012. Specifically, Mallat considered so-called scattering networks based on a wavelet transform followed by the modulus non-linearity in each network layer, and proved translation invariance (asymptotically in the wavelet scale parameter) and deformation stability of the corresponding feature extractor. This paper complements Mallat’s results by developing a theory that encompasses general convolutional transforms, or in more technical parlance, general semi-discrete frames (including Weyl–Heisenberg filters, curvelets, shearlets, ridgelets, wavelets, and learned filters), general Lipschitz-continuous non-linearities (e.g., rectified linear units, shifted logistic sigmoids, hyperbolic tangents, and modulus functions), and general Lipschitz-continuous pooling operators emulating, e.g., sub-sampling and averaging. In addition, all of these elements can be different in different network layers. For the resulting feature extractor, we prove a translation invariance result of vertical nature in the sense of the features becoming progressively more translation-invariant with increasing network depth, and we establish deformation sensitivity bounds that apply to signal classes such as, e.g., band-limited functions, cartoon functions, and Lipschitz functions.
275 citations
TL;DR: A CNN model extension is developed that redefines the concept of capsule units to become spectral–spatial units specialized in classifying remotely sensed HSI data and is able to provide competitive advantages in terms of both classification accuracy and computational time.
Abstract: Convolutional neural networks (CNNs) have recently exhibited an excellent performance in hyperspectral image classification tasks. However, the straightforward CNN-based network architecture still finds obstacles when effectively exploiting the relationships between hyperspectral imaging (HSI) features in the spectral–spatial domain, which is a key factor to deal with the high level of complexity present in remotely sensed HSI data. Despite the fact that deeper architectures try to mitigate these limitations, they also find challenges with the convergence of the network parameters, which eventually limit the classification performance under highly demanding scenarios. In this paper, we propose a new CNN architecture based on spectral–spatial capsule networks in order to achieve a highly accurate classification of HSIs while significantly reducing the network design complexity. Specifically, based on Hinton’s capsule networks, we develop a CNN model extension that redefines the concept of capsule units to become spectral–spatial units specialized in classifying remotely sensed HSI data. The proposed model is composed by several building blocks, called spectral–spatial capsules, which are able to learn HSI spectral–spatial features considering their corresponding spatial positions in the scene, their associated spectral signatures, and also their possible transformations. Our experiments, conducted using five well-known HSI data sets and several state-of-the-art classification methods, reveal that our HSI classification approach based on spectral–spatial capsules is able to provide competitive advantages in terms of both classification accuracy and computational time.
274 citations
References
More filters
TL;DR: A novel, efficient, gradient based method called long short-term memory (LSTM) is introduced, which can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units.
Abstract: Learning to store information over extended time intervals by recurrent backpropagation takes a very long time, mostly because of insufficient, decaying error backflow. We briefly review Hochreiter's (1991) analysis of this problem, then address it by introducing a novel, efficient, gradient based method called long short-term memory (LSTM). Truncating the gradient where this does not do harm, LSTM can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units. Multiplicative gate units learn to open and close access to the constant error flow. LSTM is local in space and time; its computational complexity per time step and weight is O. 1. Our experiments with artificial data involve local, distributed, real-valued, and noisy pattern representations. In comparisons with real-time recurrent learning, back propagation through time, recurrent cascade correlation, Elman nets, and neural sequence chunking, LSTM leads to many more successful runs, and learns much faster. LSTM also solves complex, artificial long-time-lag tasks that have never been solved by previous recurrent network algorithms.
72,897 citations
01 Jan 1998
TL;DR: In this article, a graph transformer network (GTN) is proposed for handwritten character recognition, which can be used to synthesize a complex decision surface that can classify high-dimensional patterns, such as handwritten characters.
Abstract: Multilayer neural networks trained with the back-propagation algorithm constitute the best example of a successful gradient based learning technique. Given an appropriate network architecture, gradient-based learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns, such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional neural networks, which are specifically designed to deal with the variability of 2D shapes, are shown to outperform all other techniques. Real-life document recognition systems are composed of multiple modules including field extraction, segmentation recognition, and language modeling. A new learning paradigm, called graph transformer networks (GTN), allows such multimodule systems to be trained globally using gradient-based methods so as to minimize an overall performance measure. Two systems for online handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of graph transformer networks. A graph transformer network for reading a bank cheque is also described. It uses convolutional neural network character recognizers combined with global training techniques to provide record accuracy on business and personal cheques. It is deployed commercially and reads several million cheques per day.
42,067 citations
TL;DR: Back-propagation repeatedly adjusts the weights of the connections in the network so as to minimize a measure of the difference between the actual output vector of the net and the desired output vector, which helps to represent important features of the task domain.
Abstract: We describe a new learning procedure, back-propagation, for networks of neurone-like units. The procedure repeatedly adjusts the weights of the connections in the network so as to minimize a measure of the difference between the actual output vector of the net and the desired output vector. As a result of the weight adjustments, internal ‘hidden’ units which are not part of the input or output come to represent important features of the task domain, and the regularities in the task are captured by the interactions of these units. The ability to create useful new features distinguishes back-propagation from earlier, simpler methods such as the perceptron-convergence procedure1.
23,814 citations
TL;DR: This work bridges the divide between high-dimensional sensory inputs and actions, resulting in the first artificial agent that is capable of learning to excel at a diverse array of challenging tasks.
Abstract: The theory of reinforcement learning provides a normative account, deeply rooted in psychological and neuroscientific perspectives on animal behaviour, of how agents may optimize their control of an environment. To use reinforcement learning successfully in situations approaching real-world complexity, however, agents are confronted with a difficult task: they must derive efficient representations of the environment from high-dimensional sensory inputs, and use these to generalize past experience to new situations. Remarkably, humans and other animals seem to solve this problem through a harmonious combination of reinforcement learning and hierarchical sensory processing systems, the former evidenced by a wealth of neural data revealing notable parallels between the phasic signals emitted by dopaminergic neurons and temporal difference reinforcement learning algorithms. While reinforcement learning agents have achieved some successes in a variety of domains, their applicability has previously been limited to domains in which useful features can be handcrafted, or to domains with fully observed, low-dimensional state spaces. Here we use recent advances in training deep neural networks to develop a novel artificial agent, termed a deep Q-network, that can learn successful policies directly from high-dimensional sensory inputs using end-to-end reinforcement learning. We tested this agent on the challenging domain of classic Atari 2600 games. We demonstrate that the deep Q-network agent, receiving only the pixels and the game score as inputs, was able to surpass the performance of all previous algorithms and achieve a level comparable to that of a professional human games tester across a set of 49 games, using the same algorithm, network architecture and hyperparameters. This work bridges the divide between high-dimensional sensory inputs and actions, resulting in the first artificial agent that is capable of learning to excel at a diverse array of challenging tasks.
23,074 citations
TL;DR: In this article, an effective way of initializing the weights that allows deep autoencoder networks to learn low-dimensional codes that work much better than principal components analysis as a tool to reduce the dimensionality of data is described.
Abstract: High-dimensional data can be converted to low-dimensional codes by training a multilayer neural network with a small central layer to reconstruct high-dimensional input vectors. Gradient descent can be used for fine-tuning the weights in such "autoencoder" networks, but this works well only if the initial weights are close to a good solution. We describe an effective way of initializing the weights that allows deep autoencoder networks to learn low-dimensional codes that work much better than principal components analysis as a tool to reduce the dimensionality of data.
16,717 citations