Clustering data in high dimensions is believed to be a hard problem in general.

The Paraphrase Database (PPDB; Ganitkevitch et al., 2013) is an extensive semantic resource, consisting of a list of phrase pairs with (heuristic) confidence estimates. However, it is still unclear how it can best be used, due to the heuristic nature of the confidences and its necessarily incomplete coverage. We propose models to leverage the phrase pairs from the PPDB to build parametric paraphrase models that score paraphrase pairs more accurately than the PPDB's internal scores while simultaneously improving its coverage. They allow for learning phrase embeddings as well as improved word embeddings. Moreover, we introduce two new, manually annotated datasets to evaluate short-phrase paraphrasing models. Using our paraphrase model trained using PPDB, we achieve state-of-the-art results on standard word and bigram similarity tasks and beat strong baselines on our new short phrase paraphrase tasks. 1,2

Word representations have proven useful for many NLP tasks, e.g., Brown clusters as features in dependency parsing (Koo et al., 2008). In this paper, we investigate the use of continuous word representations as features for dependency parsing. We compare several popular embeddings to Brown clusters, via multiple types of features, in both news and web domains. We find that all embeddings yield significant parsing gains, including some recent ones that can be trained in a fraction of the time of others. Explicitly tailoring the representations for the task leads to further improvements. Moreover, an ensemble of all representations achieves the best results, suggesting their complementarity.

Recent studies have been revisiting whole words as the basic modelling unit in speech recognition and query applications, instead of phonetic units. Such whole-word segmental systems rely on a function that maps a variable-length speech segment to a vector in a fixed-dimensional space; the resulting acoustic word embeddings need to allow for accurate discrimination between different word types, directly in the embedding space. We compare several old and new approaches in a word discrimination task. Our best approach uses side information in the form of known word pairs to train a Siamese convolutional neural network (CNN): a pair of tied networks that take two speech segments as input and produce their embeddings, trained with a hinge loss that separates same-word pairs and different-word pairs by some margin. A word classifier CNN performs similarly, but requires much stronger supervision. Both types of CNNs yield large improvements over the best previously published results on the word discrimination task.Index Terms-Acoustic word embeddings, segmental acoustic models, fixed-dimensional representations, query-by-example search.

No abstract

It has been previously shown that, when both acoustic and articulatory training data are available, it is possible to improve phonetic recognition accuracy by learning acoustic features from this multiview data with canonical correlation analysis (CCA). In contrast with previous work based on linear or kernel CCA, we use the recently proposed deep CCA, where the functional form of the feature mapping is a deep neural network. We apply the approach on a speakerindependent phonetic recognition task using data from the University of Wisconsin X-ray Microbeam Database. Using a tandem-style recognizer on this task, deep CCA features improve over earlier multiview approaches as well as over articulatory inversion and typical neural network-based tandem features. We also present a new stochastic training approach for deep CCA, which produces both faster training and better-performing features.Index Terms-multi-view learning, neural networks, deep canonical correlation analysis, XRMB, articulatory measurements INTRODUCTIONModern speech recognizers often use deep neural networks (DNNs) trained to predict the posterior probabilities of phonetic states [1]. In the two most common approaches, either (1) the DNN outputs are scaled by the state priors and used as an observation model in a hidden Markov model (HMM)-based recognizer (the hybrid approach [2]) or (2) the outputs of some layer of the network (possibly a narrow "bottleneck" layer or the final layer) are post-processed and used as acoustic features in an HMM system with a Gaussian mixture model (GMM) observation distribution (the tandem approach [3]). Working within the tandem approach, we investigate whether we can learn better DNN-based acoustic features via unsupervised learning using an external set of unlabeled multi-view data, in our case simultaneously recorded audio and articulatory measurements.The idea of feature learning using multi-view data has been explored previously using canonical correlation analysis (CCA) [4] and its nonlinear extension kernel CCA (KCCA) [5,6]. Here we propose to use the recently devloped deep CCA (DCCA) approach, which differs from linear/kernel CCA in that the feature mapping is implemented with a DNN rather than a linear/kernel function. Considering the earlier successes of CCA/KCCA, and the general success of DNNs for speech tasks, it is a natural question whether multi-view feature learning can benefit from the more flexible functional form of a DNN. We investigate this question, using data from the University of Wisconsin X-ray Microbeam Database (XRMB) [7], on speakerindependent phonetic recognition in a setting where no articulatory data is available for the recognizer training speakers. We find that DCCA indeed improves over previous CCA-based features, as well

We present a simple approach to improve direct speech-to-text translation (ST) when the source language is low-resource: we pre-train the model on a high-resource automatic speech recognition (ASR) task, and then fine-tune its parameters for ST. We demonstrate that our approach is effective by pre-training on 300 hours of English ASR data to improve Spanish-English ST from 10.8 to 20.2 BLEU when only 20 hours of Spanish-English ST training data are available. Through an ablation study, we find that the pre-trained encoder (acoustic model) accounts for most of the improvement, despite the fact that the shared language in these tasks is the target language text, not the source language audio. Applying this insight, we show that pre-training on ASR helps ST even when the ASR language differs from both source and target ST languages: pre-training on French ASR also improves Spanish-English ST. Finally, we show that the approach improves performance on a true low-resource task: pre-training on a combination of English ASR and French ASR improves Mboshi-French ST, where only 4 hours of data are available, from 3.5 to 7.1 BLEU.

Abstract-We study PCA, PLS, and CCA as stochastic optimization problems, of optimizing a population objective based on a sample. We suggest several stochastic approximation (SA) methods for PCA and PLS, and investigate their empirical performance.