A common problem in training neural networks is the vanishing and/or exploding gradient problem which is more prominently seen in training of Recurrent Neural Networks (RNNs). Thus several algorithms have been proposed for training RNNs. This paper proposes a novel adaptive stochastic Nesterov accelerated quasiNewton (aSNAQ) method for training RNNs. The proposed method aSNAQ is an accelerated method that uses the Nesterov's gradient term along with second order curvature information. The performance of the proposed method is evaluated in Tensorflow on benchmark sequence modeling problems. The results show an improved performance while maintaining a low per-iteration cost and thus can be effectively used to train RNNs.

The neural architecture search (NAS) algorithm with reinforcement learning can be a powerful and novel framework for the automatic discovering process of neural architectures. However, its application is restricted by noncontinuous and high-dimensional search spaces, which result in difficulty in optimization. To resolve these problems, we proposed NAS in embedding space (NASES), which is a novel framework. Unlike other NAS with reinforcement learning approaches that search over a discrete and high-dimensional architecture space, this approach enables reinforcement learning to search in an embedding space by using architecture encoders and decoders. The current experiment demonstrated that the performance of the final architecture network using the NASES procedure is comparable with that of other popular NAS approaches for the image classification task on CIFAR-10. The beneficial-performance and effectiveness of NASES was impressive even when only the architecture-embedding searching and pre-training controller were applied without other NAS tricks such as parameter sharing. Specifically, considerable reduction in searches was achieved by reducing the average number of searching to 100 architectures to achieve a final architecture for the NASES procedure. Introduction Deep neural networks have enabled advances in image recognition, sequential pattern recognition, recommendation systems, and various tasks in the past decades.

Variational language models seek to estimate the posterior of latent variables with an approximated variational posterior. The model often assumes the variational posterior to be factorized even when the true posterior is not. The learned variational posterior under this assumption does not capture the dependency relationships over latent variables. We argue that this would cause a typical training problem called posterior collapse observed in all other variational language models. We propose Gaussian Copula Variational Autoencoder (VAE) to avert this problem. Copula is widely used to model correlation and dependencies of high-dimensional random variables, and therefore it is helpful to maintain the dependency relationships that are lost in VAE. The empirical results show that by modeling the correlation of latent variables explicitly using a neural parametric copula, we can avert this training difficulty while getting competitive results among all other VAE approaches.

Significant advances have been made in Natural Language Proc essing (NLP) modelling since the beginning of 2018. The new approaches allow for accurate results, even when there is little labelled data, because these NLP mo dels can benefit from training on both task-agnostic and task-specific unlabelle d data. However, these advantages come with significant size and computational cos ts. This workshop paper outlines how our proposed convolutiona l student architecture, having been trained by a distillation process from a la rge-scale model, can achieve 300 inference speedup and 39 reduction in parameter count. In some cases, the student model performance surpasses its teacher on the studied tasks.

TorchGAN is a PyTorch based framework for writing succinct and comprehensible code for training and evaluation of Generative Adversarial Networks. The framework's modular design allows effortless customization of the model architecture, loss functions, training paradigms, and evaluation metrics. The key features of TorchGAN are its extensibility, built-in support for a large number of popular models, losses and evaluation metrics, and zero overhead compared to vanilla PyTorch. By using the framework to implement several popular GAN models, we demonstrate its extensibility and ease of use. We also benchmark the training time of our framework for said models against the corresponding baseline PyTorch implementations and observe that TorchGAN's features bear almost zero overhead.

We consider least squares regression estimates using deep neural networks. We show that these estimates satisfy an oracle inequality, which implies that (up to a logarithmic factor) the error of these estimates is at least as small as the optimal possible error bound which one would expect for MARS in case that this procedure would work in the optimal way. As a result we show that our neural networks are able to achieve a dimensionality reduction in case that the regression function locally has low dimensionality. This assumption seems to be realistic in real-world applications, since selected high-dimensional data are often confined to locally-low-dimensional distributions. In our simulation study we provide numerical experiments to support our theoretical results and to compare our estimate with other conventional nonparametric regression estimates, especially with MARS. The use of our estimates is illustrated through a real data analysis.

Event Representation Learning Enhanced with External Commonsense Knowledge Xiao Ding, Kuo Liao, Ting Liu, Zhongyang Li, Junwen Duan Research Center for Social Computing and Information Retrieval Harbin Institute of Technology, China {xding, kliao, tliu, zyli, jwduan }@ir.hit.edu.cn Abstract Prior work has proposed effective methods to learn event representations that can capture syntactic and semantic information over text corpus, demonstrating their effectiveness for downstream tasks such as script event prediction. On the other hand, events extracted from raw texts lacks of commonsense knowledge, such as the intents and emotions of the event participants, which are useful for distinguishing event pairs when there are only subtle differences in their surface realizations. To address this issue, this paper proposes to leverage external commonsense knowledge about the intent and sentiment of the event. Experiments on three event-related tasks, i.e., event similarity, script event prediction and stock market prediction, show that our model obtains much better event embeddings for the tasks, achieving 78% improvements on hard similarity task, yielding more precise inferences on subsequent events under given contexts, and better accuracies in predicting the volatilities of the stock market 1 . 1 Introduction Events are a kind of important objective information of the world. Structuralizing and representing such information as machine-readable knowledge are crucial to artificial intelligence (Li et al., 2018b, 2019). The main idea is to learn distributed representations for structured events (i.e. Figure 1: Intent and sentiment enhanced event embed-dings can distinguish distinct events even with high lexical overlap, and find similar events even with low lexical overlap. The function maps the summed vectors into an event embedding space.

Real world question answering can be significantly more complex than what most existing QA datasets reflect. Questions posed by users on websites, such as online travel forums, may consist of multiple sentences and not everything mentioned in a question may be relevant for finding its answer. Such questions typically have a huge candidate answer space and require complex reasoning over large knowledge corpora. We introduce the novel task of answering entity-seeking recommendation questions using a collection of reviews that describe candidate answer entities. We harvest a QA dataset that contains 48,147 paragraph-sized real user questions from travelers seeking recommendations for hotels, attractions and restaurants. Each candidate answer is associated with a collection of unstructured reviews. This dataset is challenging because commonly used neural architectures for QA are prohibitively expensive for a task of this scale. As a solution, we design a scalable cluster-select-rerank approach. It first clusters text for each entity to identify exemplar sentences describing an entity. It then uses a scalable neural information retrieval (IR) module to subselect a set of potential entities from the large candidate set. A reranker uses a deeper attention-based architecture to pick the best answers from the selected entities. This strategy performs better than a pure IR or a pure attention-based reasoning approach yielding nearly 10% relative improvement in Accuracy@3 over both approaches.

Neural network based approaches to automated story plot generation attempt to learn how to generate novel plots from a corpus of natural language plot summaries. Prior work has shown that a semantic abstraction of sentences called events improves neural plot generation and and allows one to decompose the problem into: (1) the generation of a sequence of events (event-to-event) and (2) the transformation of these events into natural language sentences (event-to-sentence). However, typical neural language generation approaches to event-to-sentence can ignore the event details and produce grammatically-correct but semantically-unrelated sentences. We present an ensemble-based model that generates natural language guided by events.We provide results---including a human subjects study---for a full end-to-end automated story generation system showing that our method generates more coherent and plausible stories than baseline approaches.

A key step in medical diagnosis is giving the patient a universally recognized label (e.g. Appendicitis) which essentially assigns the patient to a class(es) of patients with similar body failures. However, two patients having the same disease label(s) with high probability may still have differences in their feature manifestation patterns implying differences in the required treatments. Additionally, in many cases, the labels of the primary diagnoses leave some findings unexplained. Medical diagnosis is only partially about probability calculations for label X or Y. Diagnosis is not complete until the patient overall situation is clinically understood to the level that enables the best therapeutic decisions. Most machine learning models are data centric models, and evidence so far suggest they can reach expert level performance in the disease labeling phase. Nonetheless, like any other mathematical technique, they have their limitations and applicability scope. Primarily, data centric algorithms are knowledge blind and lack anatomy and physiology knowledge that physicians leverage to achieve complete diagnosis. This article advocates to complement them with intelligence to overcome their inherent limitations as knowledge blind algorithms. Machines can learn many things from data, but data is not the only source that machines can learn from. Historic patient data only tells us what the possible manifestations of a certain body failure are. Anatomy and physiology knowledge tell us how the body works and fails. Both are needed for complete diagnosis. The proposed Double Deep Learning approach, along with the initiative for Medical Wikipedia for Smart Machines, leads to AI diagnostic support solutions for complete diagnosis beyond the limited data only labeling solutions we see today. AI for medicine will forever be limited until their intelligence also integrates anatomy and physiology.