The ability to look multiple times through a series of pose-adjusted glimpses is fundamental to human vision. This critical faculty allows us to understand highly complex visual scenes. Short term memory plays an integral role in aggregating the information obtained from these glimpses and informing our interpretation of the scene. Computational models have attempted to address glimpsing and visual attention but have failed to incorporate the notion of memory. We introduce a novel, biologically inspired visual working memory architecture that we term the Hebb-Rosenblatt memory. We subsequently introduce a fully differentiable Short Term Attentive Working Memory model (STAWM) which uses transformational attention to learn a memory over each image it sees. The state of our Hebb-Rosenblatt memory is embedded in STAWM as the weights space of a layer. By projecting different queries through this layer we can obtain goal-oriented latent representations for tasks including classification and visual reconstruction. Our model obtains highly competitive classification performance on MNIST and CIFAR-10. As demonstrated through the CelebA dataset, to perform reconstruction the model learns to make a sequence of updates to a canvas which constitute a parts-based representation. Classification with the self supervised representation obtained from MNIST is shown to be in line with the state of the art models (none of which use a visual attention mechanism). Finally, we show that STAWM can be trained under the dual constraints of classification and reconstruction to provide an interpretable visual sketchpad which helps open the 'black-box' of deep learning.

In this paper, we propose a capsule-based neural network model to solve the semantic segmentation problem. By taking advantage of the extractable part-whole dependencies available in capsule layers, we derive the probabilities of the class labels for individual capsules through a recursive, layer-by-layer procedure. We model this procedure as a traceback pipeline and take it as a central piece to build an end-to-end segmentation network. Under the proposed framework, image-level class labels and object boundaries are jointly sought in an explicit manner, which poses a significant advantage over the state-of-the-art fully convolutional network (FCN) solutions. Experiments conducted on modified MNIST and neuroimages demonstrate that our model considerably enhance the segmentation performance compared to the leading FCN variant.

Abstract-- In this paper we presented a novel constructive approach for training deep neural networks using geometric approaches. We show that a topological covering can be used to define a class of distributed linear matrix inequalities, which in turn directly specify the shape and depth of a neural network architecture. The key insight is a fundamental relationship between linear matrix inequalities and their ability to bound the shape of data, and the rectified linear unit (ReLU) activation function employed in modern neural networks. We show that unit cover geometry and cover porosity are two design variables in cover-constructive learning that play a critical role in defining the complexity of the model and generalizability of the resulting neural network classifier. In the context of cover-constructive learning, these findings underscore the age old tradeoff between modelcomplexity and overfitting (as quantified by the number of elements in the data cover) and generalizability on test data. Finally, we benchmark on algorithm on the Iris, MNIST, and Wine dataset and show that the constructive algorithm is able to train a deep neural network classifier in one shot, achieving equal or superior levels of training and test classification accuracy with reduced training time. I. INTRODUCTION Artificial neural networks have proven themselves to be useful, highly flexible tools for addressing many complex problems where first-principles solutions are infeasible, impractical, orundesirable.

Various forms of representations may arise in the many layers embedded in deep neural networks (DNNs). Of these, where can we find the most compact representation? We propose to use a pruning framework to answer this question: How compact can each layer be compressed, without losing performance? Most of the existing DNN compression methods do not consider the relative compressibility of the individual layers. They uniformly apply a single target sparsity to all layers or adapt layer sparsity using heuristics and additional training. We propose a principled method that automatically determines the sparsity of individual layers derived from the importance of each layer. To do this, we consider a metric to measure the importance of each layer based on the layer-wise capacity. Given the trained model and the total target sparsity, we first evaluate the importance of each layer from the model. From the evaluated importance, we compute the layer-wise sparsity of each layer. The proposed method can be applied to any DNN architecture and can be combined with any pruning method that takes the total target sparsity as a parameter. To validate the proposed method, we carried out an image classification task with two types of DNN architectures on two benchmark datasets and used three pruning methods for compression. In case of VGG-16 model with weight pruning on the ImageNet dataset, we achieved up to 75% (17.5% on average) better top-5 accuracy than the baseline under the same total target sparsity. Furthermore, we analyzed where the maximum compression can occur in the network. This kind of analysis can help us identify the most compact representation within a deep neural network.

This paper proposes Dirichlet Variational Autoencoder (DirVAE) using a Dirichlet prior for a continuous latent variable that exhibits the characteristic of the categorical probabilities. To infer the parameters of DirVAE, we utilize the stochastic gradient method by approximating the Gamma distribution, which is a component of the Dirichlet distribution, with the inverse Gamma CDF approximation. Additionally, we reshape the component collapsing issue by investigating two problem sources, which are decoder weight collapsing and latent value collapsing, and we show that DirVAE has no component collapsing; while Gaussian VAE exhibits the decoder weight collapsing and Stick-Breaking VAE shows the latent value collapsing. The experimental results show that 1) DirVAE models the latent representation result with the best log-likelihood compared to the baselines; and 2) DirVAE produces more interpretable latent values with no collapsing issues which the baseline models suffer from. Also, we show that the learned latent representation from the DirVAE achieves the best classification accuracy in the semi-supervised and the supervised classification tasks on MNIST, OMNIGLOT, and SVHN compared to the baseline VAEs. Finally, we demonstrated that the DirVAE augmented topic models show better performances in most cases.

Automatic human affect recognition is a key step towards more natural human-computer interaction. Recent trends include recognition in the wild using a fusion of audiovisual and physiological sensors, a challenging setting for conventional machine learning algorithms. Since 2010, novel deep learning algorithms have been applied increasingly in this field. In this paper, we review the literature on human affect recognition between 2010 and 2017, with a special focus on approaches using deep neural networks. By classifying a total of 950 studies according to their usage of shallow or deep architectures, we are able to show a trend towards deep learning. Reviewing a subset of 233 studies that employ deep neural networks, we comprehensively quantify their applications in this field. We find that deep learning is used for learning of (i) spatial feature representations, (ii) temporal feature representations, and (iii) joint feature representations for multimodal sensor data. Exemplary state-of-the-art architectures illustrate the progress. Our findings show the role deep architectures will play in human affect recognition, and can serve as a reference point for researchers working on related applications.

Human perception of 3D shapes goes beyond reconstructing them as a set of points or a composition of geometric primitives: we also effortlessly understand higher-level shape structure such as the repetition and reflective symmetry of object parts. In contrast, recent advances in 3D shape sensing focus more on low-level geometry but less on these higher-level relationships. In this paper, we propose 3D shape programs, integrating bottom-up recognition systems with top-down, symbolic program structure to capture both low-level geometry and high-level structural priors for 3D shapes. Because there are no annotations of shape programs for real shapes, we develop neural modules that not only learn to infer 3D shape programs from raw, unannotated shapes, but also to execute these programs for shape reconstruction. After initial bootstrapping, our end-to-end differentiable model learns 3D shape programs by reconstructing shapes in a self-supervised manner. Experiments demonstrate that our model accurately infers and executes 3D shape programs for highly complex shapes from various categories. It can also be integrated with an image-to-shape module to infer 3D shape programs directly from an RGB image, leading to 3D shape reconstructions that are both more accurate and more physically plausible.

This project contains an overview of recent trends in deep learning based natural language processing (NLP). It covers the theoretical descriptions and implementation details behind deep learning models, such as recurrent neural networks (RNNs), convolutional neural networks (CNNs), and reinforcement learning, used to solve various NLP tasks and applications. The overview also contains a summary of state of the art results for NLP tasks such as machine translation, question answering, and dialogue systems. There are various ways to contribute to this project. Refer to the issue section of the GitHub repository to learn more about how you can help. Or you can make suggestions by submitting a new issue.

"NEVER THINK THERE IS ANYTHING IMPOSSIBLE FOR THE SOUL. IT IS THE GREATEST HERESY TO THINK SO. IF THERE IS A SIN, THIS IS THE ONLY SIN; TO SAY THAT YOU ARE WEAK, OR OTHERS ARE WEAK" - By Swami vivekanand Is Deep Learning now overtaking the Machine Learning algorithm?? Let us first know what is Machine Learning? Machine Learning was coined by "Arthur Samuel" in the year 1959. As we know to perform any Machine Learning algorithm we require a humongous amount of data and very high computation power. But at that time we are not able to generate or store that amount of data, additionally, the computer also doesn't have enough computation power.

Intel's CES press event wasn't as bombastic as last year's blowout keynote, but the chip giant actually had a lot more news to share. For one, Intel rounded out its ninth generation Core CPUs with six more processors, going from lowly i3 models to high-end i9s. And after years of waiting for 10nm chips, it seems like we'll get several this year, from desktop Ice Lake CPUs to Lakefield chips that will bring together Sunny Cove cores with some Atom CPUs to balance high and low-power performance. And finally, we got a glimpse at Intel's next Nervana AI processor, which will focus more on inference capabilities and not just deep learning. Follow all the latest news from CES 2019 here!