Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

/pdf/deep-residual-learning-for-image-recognition-b75jg61qrq.pdf

Deep Residual Learning for Image Recognition

We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm.

Adam: A Method for Stochastic Optimization

We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.

/pdf/imagenet-classification-with-deep-convolutional-neural-1190x9a6b9.pdf

ImageNet Classification with Deep Convolutional Neural Networks

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.

Long short-term memory

In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Very Deep Convolutional Networks for Large-Scale Image Recognition

While language tasks are naturally expressed in a single, unified, modeling framework, i.e., generating sequences of tokens, this has not been the case in computer vision. As a result, there is a proliferation of distinct architectures and loss functions for different vision tasks. In this work we show that a diverse set of"core"computer vision tasks can also be unified if formulated in terms of a shared pixel-to-sequence interface. We focus on four tasks, namely, object detection, instance segmentation, keypoint detection, and image captioning, all with diverse types of outputs, e.g., bounding boxes or dense masks. Despite that, by formulating the output of each task as a sequence of discrete tokens with a unified interface, we show that one can train a neural network with a single model architecture and loss function on all these tasks, with no task-specific customization. To solve a specific task, we use a short prompt as task description, and the sequence output adapts to the prompt so it can produce task-specific output. We show that such a model can achieve competitive performance compared to well-established task-specific models.

A Unified Sequence Interface for Vision Tasks

Using an unsupervised learning procedure, a network is trained on an ensemble of images of the same two-dimensional object at different positions, orientations and sizes. Each half of the network "sees" one fragment of the object, and tries to produce as output a set of 4 parameters that have high mutual information with the 4 parameters output by the other half of the network. Given the ensemble of training patterns, the 4 parameters on which the two halves of the network can agree are the position, orientation, and size of the whole object, or some recoding of them. After training, the network can reject instances of other shapes by using the fact that the predictions made by its two halves disagree. If two competing networks are trained on an unlabelled mixture of images of two objects, they cluster the training cases on the basis of the objects' shapes, independently of the position, orientation, and size.

/pdf/discovering-viewpoint-invariant-relationships-that-1fx9o3bcui.pdf

Discovering Viewpoint-Invariant Relationships That Characterize Objects

We describe a learning procedure for a generative model that contains a hidden Markov Random Field (MRF) which has directed connections to the observable variables. The learning procedure uses a variational approximation for the posterior distribution over the hidden variables. Despite the intractable partition function of the MRF, the weights on the directed connections and the variational approximation itself can be learned by maximizing a lower bound on the log probability of the observed data. The parameters of the MRF are learned by using the mean field version of contrastive divergence [1]. We show that this hybrid model simultaneously learns parts of objects and their inter-relationships from intensity images. We discuss the extension to multiple MRF’s linked into in a chain graph by directed connections.

/pdf/learning-causally-linked-markov-random-fields-2koq1l7hyi.pdf

Learning Causally Linked Markov Random Fields.

Geoffrey E. Hinton Department of Computer Science . U ni versi ty of Toran to Toronto, Canada M5S lA4 One way of simplifying neural networks so they generalize better is to add an extra t.erm 10 the error fUll c tion that will penalize complexit.y. \Ve propose a new penalt.y t.erm in which the dist rihution of weight values is modelled as a mixture of multiple gaussians . C nder this model, a set of weights is simple if the weights can be clustered into subsets so that the weights in each cluster have similar values . We allow the parameters of the mixture model to adapt at t.he same time as t.he network learns. Simulations demonstrate that this complexity term is more effective than previous complexity terms.

/pdf/adaptive-soft-weight-tying-using-gaussian-mixtures-1j4iszu7ms.pdf

Adaptive Soft Weight Tying using Gaussian Mixtures

Deep Learning methods aim at learning feature hierarchies. Applications of deep learning to vision tasks date back to convolutional networks in the early 1990s. These methods have been the subject of a recent surge of interest for two main reasons: when labeled data is scarce, unsupervised learning algorithms can learn useful feature hierarchies. When labeled data is abundant, supervised methods can be used to train very large networks on very large datasets through the use of high-performance computers. Such large networks have been shown to outperform previous state-of-theart methods on several perceptual tasks, including categorylevel object recognition, object detection and semantic segmentation. In “Stacked Predictive Sparse Decomposition for Classification of Histology Sections” (doi:10.1007/s11263-0140790-9) the authors propose the use of an unsupervised feature learning algorithm for the analysis of biological tissue imagery. Biomedical applications is one domainwhere unsupervised learning can be very useful because of the paucity of labeled images available at training time.

/pdf/guest-editorial-deep-learning-5f5d6660f6.pdf

Geoffrey E. Hinton

Papers

A Unified Sequence Interface for Vision Tasks

Discovering Viewpoint-Invariant Relationships That Characterize Objects

Learning Causally Linked Markov Random Fields.

Adaptive Soft Weight Tying using Gaussian Mixtures

Guest Editorial: Deep Learning