We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

Deep neural nets with a large number of parameters are very powerful machine learning systems. However, overfitting is a serious problem in such networks. Large networks are also slow to use, making it difficult to deal with overfitting by combining the predictions of many different large neural nets at test time. Dropout is a technique for addressing this problem. The key idea is to randomly drop units (along with their connections) from the neural network during training. This prevents units from co-adapting too much. During training, dropout samples from an exponential number of different "thinned" networks. At test time, it is easy to approximate the effect of averaging the predictions of all these thinned networks by simply using a single unthinned network that has smaller weights. This significantly reduces overfitting and gives major improvements over other regularization methods. We show that dropout improves the performance of neural networks on supervised learning tasks in vision, speech recognition, document classification and computational biology, obtaining state-of-the-art results on many benchmark data sets.

/pdf/dropout-a-simple-way-to-prevent-neural-networks-from-4gtr92okhv.pdf

Dropout: a simple way to prevent neural networks from overfitting

We present a new technique called “t-SNE” that visualizes high-dimensional data by giving each datapoint a location in a two or three-dimensional map. The technique is a variation of Stochastic Neighbor Embedding (Hinton and Roweis, 2002) that is much easier to optimize, and produces significantly better visualizations by reducing the tendency to crowd points together in the center of the map. t-SNE is better than existing techniques at creating a single map that reveals structure at many different scales. This is particularly important for high-dimensional data that lie on several different, but related, low-dimensional manifolds, such as images of objects from multiple classes seen from multiple viewpoints. For visualizing the structure of very large datasets, we show how t-SNE can use random walks on neighborhood graphs to allow the implicit structure of all of the data to influence the way in which a subset of the data is displayed. We illustrate the performance of t-SNE on a wide variety of datasets and compare it with many other non-parametric visualization techniques, including Sammon mapping, Isomap, and Locally Linear Embedding. The visualizations produced by t-SNE are significantly better than those produced by the other techniques on almost all of the datasets.

/pdf/visualizing-data-using-t-sne-2w1n304iq0.pdf

Visualizing Data using t-SNE

Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with L layers have L connections&#x2014;one between each layer and its subsequent layer&#x2014;our network has L(L+1)/2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less memory and computation to achieve high performance. Code and pre-trained models are available at https://github.com/liuzhuang13/DenseNet.

/pdf/densely-connected-convolutional-networks-i745msea9d.pdf

Densely Connected Convolutional Networks

Convolutional networks are at the core of most state of-the-art computer vision solutions for a wide variety of tasks. Since 2014 very deep convolutional networks started to become mainstream, yielding substantial gains in various benchmarks. Although increased model size and computational cost tend to translate to immediate quality gains for most tasks (as long as enough labeled data is provided for training), computational efficiency and low parameter count are still enabling factors for various use cases such as mobile vision and big-data scenarios. Here we are exploring ways to scale up networks in ways that aim at utilizing the added computation as efficiently as possible by suitably factorized convolutions and aggressive regularization. We benchmark our methods on the ILSVRC 2012 classification challenge validation set demonstrate substantial gains over the state of the art: 21:2% top-1 and 5:6% top-5 error for single frame evaluation using a network with a computational cost of 5 billion multiply-adds per inference and with using less than 25 million parameters. With an ensemble of 4 models and multi-crop evaluation, we report 3:5% top-5 error and 17:3% top-1 error on the validation set and 3:6% top-5 error on the official test set.

Rethinking the Inception Architecture for Computer Vision

Applications of multi-class classification, such as document categorization, often appear in cost-sensitive settings. Recent work has significantly improved the state of the art by moving beyond "flat" classification through incorporation of class hierarchies [4]. We present a novel algorithm that goes beyond hierarchical classification and estimates the latent semantic space that underlies the class hierarchy. In this space, each class is represented by a prototype and classification is done with the simple nearest neighbor rule. The optimization of the semantic space incorporates large margin constraints that ensure that for each instance the correct class prototype is closer than any other. We show that our optimization is convex and can be solved efficiently for large data sets. Experiments on the OHSUMED medical journal data base yield state-of-the-art results on topic categorization.

/pdf/large-margin-taxonomy-embedding-with-an-application-to-2v686wn5ra.pdf

Large margin taxonomy embedding with an application to document categorization

The goal of machine learning is to build automated systems that can classify and recognize complex patterns in data. The representation of the data plays an important role in determining what types of patterns can be automatically discovered. Many algorithms for machine learning assume that the data are represented as elements in a metric space. The performance of these algorithms can depend sensitively on the manner in which distances are measured. When data are represented as points in a multidimensional vector space, simple Euclidean distances are often used to measure the dissimilarity between different examples. However, such distances often do not yield reliable judgments; in addition, they cannot highlight the distinctive features that play a role in certain types of classification, but not others. Naturally, for different types of clustering, different ways of measuring dissimilarity were needed. In particular, different metrics for computing distances between feature vectors. This paper describes two algorithms for learning such distance metrics based on recent developments in convex optimization.

Convex Optimizations for Distance Metric Learning and Pattern Classification [Applications Corner]

A widely deployed method for reducing the training time of deep neural networks is to normalize activations at each layer. Although various normalization schemes have been proposed, they all follow a common theme: normalize across spatial dimensions and discard the extracted statistics. In this paper, we propose a novel normalization method that deviates from this theme. Our approach, which we refer to as Positional Normalization (PONO), normalizes exclusively across channels, which allows us to capture structural information of the input image in the first and second moments. Instead of disregarding this information, we inject it into later layers to preserve or transfer structural information in generative networks. We show that PONO significantly improves the performance of deep networks across a wide range of model architectures and image generation tasks.

/pdf/positional-normalization-36c72iryhe.pdf

Positional Normalization

Natural images are virtually surrounded by low-density misclassified regions that can be efficiently discovered by gradient-guided search --- enabling the generation of adversarial images. While many techniques for detecting these attacks have been proposed, they are easily bypassed when the adversary has full knowledge of the detection mechanism and adapts the attack strategy accordingly. In this paper, we adopt a novel perspective and regard the omnipresence of adversarial perturbations as a strength rather than a weakness. We postulate that if an image has been tampered with, these adversarial directions either become harder to find with gradient methods or have substantially higher density than for natural images. We develop a practical test for this signature characteristic to successfully detect adversarial attacks, achieving unprecedented accuracy under the white-box setting where the adversary is given full knowledge of our detection mechanism.

A New Defense Against Adversarial Images: Turning a Weakness into a Strength.

Embodiments are directed towards employing a playful incentive to encourage users to provide feedback that is useable to train a classifier. The classifier being associated with any of a variety of different settings, including but not limited to classifying: messages as ham/spam, images, advertising, bookmarking, music, videos, photographs, shopping, or the like. An animated image, such as a pet, provides an interface to the classifier that encourages and responds to user feedback. Users may share their classifiers or aspects thereof with other users to enable a community of knowledge to be applied to a classification task, while preserving privacy of the user feedback. One form of sharing may be within the context of a competitive game. Various evaluations may be performed on a classifier to indicate user feedback consistency, or quality. Classifiers may also be used to provide users with advertisements, products, or services based on the user's feedback.

Kilian Q. Weinberger

Papers

Large margin taxonomy embedding with an application to document categorization

Convex Optimizations for Distance Metric Learning and Pattern Classification [Applications Corner]

Positional Normalization

A New Defense Against Adversarial Images: Turning a Weakness into a Strength.

Playful incentive for labeling content