Computer vision has benefited from initializing multiple deep layers with weights pretrained on large supervised training sets like ImageNet. Natural language processing (NLP) typically sees initialization of only the lowest layer of deep models with pretrained word vectors. In this paper, we use a deep LSTM encoder from an attentional sequence-to-sequence model trained for machine translation (MT) to contextualize word vectors. We show that adding these context vectors (CoVe) improves performance over using only unsupervised word and character vectors on a wide variety of common NLP tasks: sentiment analysis (SST, IMDb), question classification (TREC), entailment (SNLI), and question answering (SQuAD). For fine-grained sentiment analysis and entailment, CoVe improves performance of our baseline models to the state of the art.

We propose a novel reinforcement learning approach for finite Markov decision processes (MDPs) with delayed rewards. In this work, biases of temporal difference (TD) estimates are proved to be corrected only exponentially slowly in the number of delay steps. Furthermore, variances of Monte Carlo (MC) estimates are proved to increase the variance of other estimates, the number of which can exponentially grow in the number of delay steps. We introduce RUDDER, a return decomposition method, which creates a new MDP with same optimal policies as the original MDP but with redistributed rewards that have largely reduced delays. If the return decomposition is optimal, then the new MDP does not have delayed rewards and TD estimates are unbiased. In this case, the rewards track Q-values so that the future expected reward is always zero. We experimentally confirm our theoretical results on bias and variance of TD and MC estimates. On artificial tasks with different lengths of reward delays, we show that RUDDER is exponentially faster than TD, MC, and MC Tree Search (MCTS). RUDDER outperforms rainbow, A3C, DDQN, Distributional DQN, Dueling DDQN, Noisy DQN, and Prioritized DDQN on the delayed reward Atari game Venture in only a fraction of the learning time. RUDDER considerably improves the state-of-the-art on the delayed reward Atari game Bowling in much less learning time. Source code is available at https://github.com/ml-jku/baselines-rudder, with demonstration videos at https://goo.gl/EQerZV.

Deep learning (DL) has achieved remarkable progress over the past decade and been widely applied to many safety-critical applications. However, the robustness of DL systems recently receives great concerns, such as adversarial examples against computer vision systems, which could potentially result in severe consequences. Adopting testing techniques could help to evaluate the robustness of a DL system and therefore detect vulnerabilities at an early stage. The main challenge of testing such systems is that its runtime state space is too large: if we view each neuron as a runtime state for DL, then a DL system often contains massive states, rendering testing each state almost impossible. For traditional software, combinatorial testing (CT) is an effective testing technique to reduce the testing space while obtaining relatively high defect detection abilities. In this paper, we perform an exploratory study of CT on DL systems. We adapt the concept in CT and propose a set of coverage criteria for DL systems, as well as a CT coverage guided test generation technique. Our evaluation demonstrates that CT provides a promising avenue for testing DL systems. We further pose several open questions and interesting directions for combinatorial testing of DL systems.

In this study, we explore capsule networks with dynamic routing for text classification. We propose three strategies to stabilize the dynamic routing process to alleviate the disturbance of some noise capsules which may contain "background" information or have not been successfully trained. A series of experiments are conducted with capsule networks on six text classification benchmarks. Capsule networks achieve competitive results over the strong baseline methods on 4 out of 6 datasets, which shows the effectiveness of capsule networks for text classification. We additionally show that capsule networks exhibit significant improvement when transfer single-label to multi-label text classification over the strong competitors. To the best of our knowledge, this is the first work that capsule networks have been empirically investigated for text modeling.

Vehicles of higher automation levels require the creation of situation awareness. One important aspect of this situation awareness is an understanding of the current risk of a driving situation. In this work, we present a novel approach for the dynamic risk assessment of driving situations based on images of a front stereo camera using deep learning. To this end, we trained a deep neural network with recorded monocular images, disparity maps and a risk metric for diverse traffic scenes. Our approach can be used to create the aforementioned situation awareness of vehicles of higher automation levels and can serve as a heterogeneous channel to systems based on radar or lidar sensors that are used traditionally for the calculation of risk metrics.

Binary neural networks (BNN) have been studied extensively since they run dramatically faster at lower memory and power consumption than floating-point networks, thanks to the efficiency of bit operations. However, contemporary BNNs whose weights and activations are both single bits suffer from severe accuracy degradation. To understand why, we investigate the representation ability, speed and bias/variance of BNNs through extensive experiments. We conclude that the error of BNNs are predominantly caused by the intrinsic instability (training time) and non-robustness (train & test time). Inspired by this investigation, we propose the Binary Ensemble Neural Network (BENN) which leverages ensemble methods to improve the performance of BNNs with limited efficiency cost. While ensemble techniques have been broadly believed to be only marginally helpful for strong classifiers such as deep neural networks, our analyses and experiments show that they are naturally a perfect fit to boost BNNs. We find that our BENN, which is faster and much more robust than state-of-the-art binary networks, can even surpass the accuracy of the full-precision floating number network with the same architecture.

Through the success of deep learning in various domains, artificial neural networks are currently among the most used artificial intelligence methods. Taking inspiration from the network properties of biological neural networks (e.g. sparsity, scale-freeness), we argue that (contrary to general practice) artificial neural networks, too, should not have fully-connected layers. Here we propose sparse evolutionary training of artificial neural networks, an algorithm which evolves an initial sparse topology (Erd\H{o}s-R\'enyi random graph) of two consecutive layers of neurons into a scale-free topology, during learning. Our method replaces artificial neural networks fully-connected layers with sparse ones before training, reducing quadratically the number of parameters, with no decrease in accuracy. We demonstrate our claims on restricted Boltzmann machines, multi-layer perceptrons, and convolutional neural networks for unsupervised and supervised learning on 15 datasets. Our approach has the potential to enable artificial neural networks to scale up beyond what is currently possible.

A critical research problem about information cascades, which is a central topic of social network analysis, is to predict the potential influence or the future growth of cascades. Recent developments of deep learning have provided promising alternatives, which no longer rely on heavy feature engineering efforts and instead learn the representation of cascade graphs in an end-to-end manner. In reality, however, the influence of a cascade not only depends on the cascade graph and the global network structure, but also largely relies on the content of the cascade and the preferences of users. In this work, we extend the deep learning approaches to cascade prediction by jointly modeling the content and the structure of cascades. We find that text information provides a valuable addition for the learning of cascade graphs, especially when some users (nodes) have rarely participated in the past cascades. To this end, a gating mechanism is introduced to dynamically fuse the structural and textual representations of nodes based on their respective properties. Attentions are employed to incorporate the text information associated with both cascade items and nodes. Empirical experiments demonstrate that incorporating text information brings a significant improvement to cascade prediction, and that the proposed model outperforms alternative ways to combine text and networks.

The year of 2018 should be affectionately referred to as the'Year of Sci-Fi Dreams'. This is primarily due to the fact that almost every unimaginable concept from the depths of science fiction cinema has started becoming a reality. Even though the advent of Deep Neural Networks and Artificial General Intelligence (AGI) has been making waves across the technology ecosystem (like the defeat of World Champion Lee Sedol in the game of'Go' by AGI AlphaGo-Zero), there have been a few innovations that have baffled even the brightest minds in the industry. So ignoring the various trends and innovations in the world of AI, here are three technological advancements in the field of AI and Machine Learning (ML) that will blow your mind! One of the biggest achievements in the field of Medicine was the creation of a fully functional human lung, just from stem cells.

Nvidia has released a bunch of new tools for savvy AI developers in time for the Computer Vision and Pattern Recognition conference in Salt Lake City on Tuesday. Some of them have been previously announced at its GPU Technology Conference (GTC) earlier this year. The beta platform for using graphics cards with the Kubernetes system is now available for developers to test out. It's aimed at enterprises dealing with heavy AI workloads that need to be shared across multiple GPU cloud clusters. Large datasets and models take a long time to train so using Kubernetes will speed up training and inference.