In recent years, deep reinforcement learning has pushed the boundaries of Artificial Intelligence to an unprecedented level, achieving what was expected to be possible only in a decade and outperforming human intelligence in a number of highly complex tasks. Paramount examples of this potential have appeared over the past few years, with such algorithms mastering games and tasks of increasing complexity, from playing Atari to learning to walk and beating world grandmasters at the game of Go [16, 23, 24, 31-33]. Such impressive success would be impossible without using neural networks to approximate value functions and / or policy functions in reinforcement learning algorithms. While neural networks, in particular deep neural networks, provide a powerful and versatile tool to approximate high dimensional functions [4, 12, 17], their intrinsic nonlinearity might also lead to trouble in training, in particular in the context of reinforcement learning. For example, it is well known that nonlinear approximation to value function might cause divergence in the classical temporal-difference learning due to instability [40].

We present an alternative layer to convolution layers in convolutional neural networks (CNNs). Our approach reduces the complexity of convolutions by replacing it with binary decisions. Those binary decisions are used as indexes to conditional probability distributions where each probability represents a leaf in a decision tree. This means that only the indices to the probabilities need to be determined once, thus reducing the complexity of convolutions by the depth of the output tensor. Index computation is performed by simple binary decisions that require fewer CPU cycles compared to conventionally used multiplications. In addition, we show how convolutions can be replaced by binary decisions. These binary decisions form indices in the conditional probability distributions and we show how they are used to replace 2D weight matrices as well as 3D weight tensors. These new layers can be trained like convolution layers in CNNs based on the backpropagation algorithm, for which we provide a formalization. Our results on multiple publicly available data sets show that our approach outperforms conventional CNNs. Beyond the formalized reduction of complexity and the improved qualitative performance, we show empirically a significant runtime improvement compared to convolution layers. DOWNLOAD EXAMPLES: https://drive.google.com/open?id=1gqLD5N--tqVNCixenXiptcaR5hAP5IYJ

A standard ODE solver can be used to solve all the hidden states and final states (output from the neural network), starting from an initial state (input to the neural network). The continuous neural network described in (2) exhibits several advantages over its discrete counterpart described in (1), in terms of memory efficiency, parameter efficiency, explicit control of the numerical error of final output, etc. One missing component in the current Neural ODE network is the various regularization mechanisms commonly employed in discrete neural networks. These regularization techniques have been demonstrated to be crucial in reducing generalization errors, and in improving the robustness of neural networks to adversarial attacks. Many of these regularization techniques are based on stochastic noise injection. For instance, dropout [3] is widely adopted to prevent overfitting; injecting Gaussian random noise during the forward propagation is effective in improving generalization [4, 5] as well as robustness to adversarial attacks [6, 7].

Recently, neural network based approaches have achieved significant improvement for solving large, complex, graph-structured problems. However, their bottlenecks ] still need to be addressed, and the advantages of multi-scale information and deep architectures have not been sufficiently exploited. In this paper, we theoretically analyze how existing Graph Convolutional Networks (GCNs) have limited expressive power due to the constraint of the activation functions and their architectures. We generalize spectral graph convolution and deep GCN in block Krylov subspace forms and devise two architectures, both with the potential to be scaled deeper but each making use of the multi-scale information in different ways. We further show that the equivalence of these two architectures can be established under certain conditions. On several node classification tasks, with or without the help of validation, the two new architectures achieve better performance compared to many state-of-the-art methods.

Automated text generation has been applied broadly in many domains such as marketing and robotics, and used to create chatbots, product reviews and write poetry. The ability to synthesize text, however, presents many potential risks, while access to the technology required to build generative models is becoming increasingly easy. This work is aligned with the efforts of the United Nations and other civil society organisations to highlight potential political and societal risks arising through the malicious use of text generation software, and their potential impact on human rights. As a case study, we present the findings of an experiment to generate remarks in the style of political leaders by fine-tuning a pretrained AWD- LSTM model on a dataset of speeches made at the UN General Assembly. This work highlights the ease with which this can be accomplished, as well as the threats of combining these techniques with other technologies.

Humans are able to conceive physical reality by jointly learning different facets thereof. To every pair of notions related to a perceived reality may correspond a mutual relation, which is a notion on its own, but one-level higher. Thus, we may have a description of perceived reality on at least two levels and the translation map between them is in general, due to their different content corpus, one-to-many. Following success of the unsupervised neural machine translation models, which are essentially one-to-one mappings trained separately on monolingual corpora, we examine further capabilities of unsupervised deep learning methods used there and apply these methods to sets of notions of different level and measure. Using the graph and word embedding-like techniques, we build one-to-many map without parallel data in order to establish a unified latent mental representation of the outer world, by combining notions of different kind into a unique conceptual framework. Due to latent similarity, by aligning two embedding spaces in purely unsupervised way, one obtains a geometric relation between objects of cognition on the two levels, making it possible to express a natural knowledge using one description in the context of the other.

Gary Marcus has emerged as one of deep learning's chief skeptics. In a recent interview, and a slightly less recent medium post, he discusses his feud with deep learning pioneer Yann LeCun and some of his views on how deep learning is overhyped. I find the whole thing entertaining, but at many times LeCun and Marcus are talking past each other more than with each other. Marcus seems to me to be either unaware of or ignoring certain truths about machine learning and LeCun seems to basically agree with Marcus' ideas in a way that's unsatisfying for Marcus. The temptation for me to brush 10 years of dust off of my professor hat is too much to ignore.

During this webinar, we show the future of HCS image analysis enabled by Genedata Imagence, and how this solution efficiently automates HCS image analysis. Genedata Imagence does not require technical expertise beyond that of an assay biologist to annotate images, and our deep learning-based workflow makes this an extremely efficient process enabling higher throughput and improved quality. We show how one can rapidly detect and define all cellular phenotypes in a high-content screen in a few simple steps, with the final goal to precisely quantify relevant pharmacology.

Although Deep Learning has been around for several years now, that's still an unanswered question. Much of the difficulty in designing a good neural net stems from the fact that they're still black boxes. We have some high-level idea of how they work, but we don't really know how they achieve the results that they do. There are a few clues we can start with. From an intuitive perspective, all of these additions to the state-of-the-art of Deep Learning make sense.

Although Deep Learning has been around for several years now, that's still an unanswered question. Much of the difficulty in designing a good neural net stems from the fact that they're still black boxes. We have some high-level idea of how they work, but we don't really know how they achieve the results that they do. There are a few clues we can start with. From an intuitive perspective, all of these additions to the state-of-the-art of Deep Learning make sense.