Natural language inference (NLI) is the task of determining if a natural language hypothesis can be inferred from a given premise in a justifiable manner. NLI was proposed as a benchmark task for natural language understanding. Existing models perform well at standard datasets for NLI, achieving impressive results across different genres of text. However, the extent to which these models understand the semantic content of sentences is unclear. In this work, we propose an evaluation methodology consisting of automatically constructed "stress tests" that allow us to examine whether systems have the ability to make real inferential decisions. Our evaluation of six sentence-encoder models on these stress tests reveals strengths and weaknesses of these models with respect to challenging linguistic phenomena, and suggests important directions for future work in this area.
While Bayesian methods are extremely popular in statistics and machine learning, their application to massive datasets is often challenging, when possible at all. Indeed, the classical MCMC algorithms are prohibitively slow when both the model dimension and the sample size are large. Variational Bayesian methods aim at approximating the posterior by a distribution in a tractable family. Thus, MCMC are replaced by an optimization algorithm which is orders of magnitude faster. VB methods have been applied in such computationally demanding applications as including collaborative filtering, image and video processing, NLP and text processing... However, despite very nice results in practice, the theoretical properties of these approximations are usually not known. In this paper, we propose a general approach to prove the concentration of variational approximations of fractional posteriors. We apply our theory to two examples: matrix completion, and Gaussian VB.
Chatbots, taking advantage of the success of the messaging apps and recent advances in Artificial Intelligence, have become very popular, from helping business to improve customer services to chatting to users for the sake of conversation and engagement (celebrity or personal bots). However, developing and improving a chatbot requires understanding their data generated by its users. Dialog data has a different nature of a simple question and answering interaction, in which context and temporal properties (turn order) creates a different understanding of such data. In this paper, we propose a novelty metric to compute dialogs' similarity based not only on the text content but also on the information related to the dialog structure. Our experimental results performed over the Switchboard dataset show that using evidence from both textual content and the dialog structure leads to more accurate results than using each measure in isolation.
It is oftentimes impossible to understand how machine learning models reach a decision. While recent research has proposed various technical approaches to provide some clues as to how a learning model makes individual decisions, they cannot provide users with ability to inspect a learning model as a complete entity. In this work, we propose a new technical approach that augments a Bayesian regression mixture model with multiple elastic nets. Using the enhanced mixture model, we extract explanations for a target model through global approximation. To demonstrate the utility of our approach, we evaluate it on different learning models covering the tasks of text mining and image recognition. Our results indicate that the proposed approach not only outperforms the state-of-the-art technique in explaining individual decisions but also provides users with an ability to discover the vulnerabilities of a learning model.
Drug-drug interaction (DDI) is a vital information when physicians and pharmacists intend to co-administer two or more drugs. Thus, several DDI databases are constructed to avoid mistakenly combined use. In recent years, automatically extracting DDIs from biomedical text has drawn researchers' attention. However, the existing work utilize either complex feature engineering or NLP tools, both of which are insufficient for sentence comprehension. Inspired by the deep learning approaches in natural language processing, we propose a recur- rent neural network model with multiple attention layers for DDI classification. We evaluate our model on 2013 SemEval DDIExtraction dataset. The experiments show that our model classifies most of the drug pairs into correct DDI categories, which outperforms the existing NLP or deep learning methods.
In recent years we have seen rapid and significant progress in automatic image description but what are the open problems in this area? Most work has been evaluated using text-based similarity metrics, which only indicate that there have been improvements, without explaining what has improved. In this paper, we present a detailed error analysis of the descriptions generated by a state-of-the-art attention-based model. Our analysis operates on two levels: first we check the descriptions for accuracy, and then we categorize the types of errors we observe in the inaccurate descriptions. We find only 20% of the descriptions are free from errors, and surprisingly that 26% are unrelated to the image. Finally, we manually correct the most frequently occurring error types (e.g. gender identification) to estimate the performance reward for addressing these errors, observing gains of 0.2--1 BLEU point per type.
Autonomous agents must often detect affordances: the set of behaviors enabled by a situation. Affordance detection is particularly helpful in domains with large action spaces, allowing the agent to prune its search space by avoiding futile behaviors. This paper presents a method for affordance extraction via word embeddings trained on a Wikipedia corpus. The resulting word vectors are treated as a common knowledge database which can be queried using linear algebra. We apply this method to a reinforcement learning agent in a text-only environment and show that affordance-based action selection improves performance most of the time. Our method increases the computational complexity of each learning step but significantly reduces the total number of steps needed. In addition, the agent's action selections begin to resemble those a human would choose.
This paper describes our system, dubbed WS4A (Web Services for All), that participated in the fourth edition of the BioASQ challenge (2016). We used WS4A to perform the Question and Answering (QA) task 4b, which consisted on the retrieval of relevant concepts, documents, snippets, RDF triples, exact answers and ideal answers for each given question. The novelty in our approach consists on the maximum exploitation of existing web services in each step of WS4A, such as the annotation of text, and the retrieval of metadata for each annotation. The information retrieved included concept identifiers, ontologies, ancestors, and most importantly, PubMed identifiers. The paper describes the WS4A pipeline and also presents the precision, recall and f-measure values obtained in task 4b. Our system achieved two second places in two subtasks on one of the five batches.
Two semimetrics on probability distributions are proposed, given as the sum of differences of expectations of analytic functions evaluated at spatial or frequency locations (i.e, features). The features are chosen so as to maximize the distinguishability of the distributions, by optimizing a lower bound on test power for a statistical test using these features. The result is a parsimonious and interpretable indication of how and where two distributions differ locally. An empirical estimate of the test power criterion converges with increasing sample size, ensuring the quality of the returned features. In real-world benchmarks on high-dimensional text and image data, linear-time tests using the proposed semimetrics achieve comparable performance to the state-of-the-art quadratic-time maximum mean discrepancy test, while returning human-interpretable features that explain the test results.
This paper proposes an architecture for deep neural networks with hidden layer branches that learn targets of lower hierarchy than final layer targets. The branches provide a channel for enforcing useful information in hidden layer which helps in attaining better accuracy, both for the final layer and hidden layers. The shared layers modify their weights using the gradients of all cost functions higher than the branching layer. This model provides a flexible inference system with many levels of targets which is modular and can be used efficiently in situations requiring different levels of results according to complexity. This paper applies the idea to a text classification task on 20 Newsgroups data set with two level of hierarchical targets and a comparison is made with training without the use of hidden layer branches.