The paper improves part-of-speech tagging with multi-task learning and character-level word representations.
problem Improving part-of-speech tagging accuracy.
method Developed a new character-level word representation using feedforward neural network, pretraining with existing word vectors, and an additional prediction of neighbour labels as an auxiliary loss.
result The methods significantly improved POS tagging performance on English and Russian languages.
End-to-end ASR model combines word and character representation for improved performance.
problem Difficulty in training with word-level supervision due to sparsity of examples.
method Multi-task learning framework combining word and character representations.
result Improved word-error rate (WER) by interpolating between word-level and character-level models.
This paper reports the performances of shallow word-level convolutional neural networks (CNN), our earlier work (2015), on the eight datasets with relatively large training data that were used for testing the very deep character-level CNN in Conneau et al. (2016). Our findings are as follows. The shallow word-level CNN…
The study examines how character and word-level representations improve sentence-level tasks.
problem Improving the quality of word and sentence representations using character-level information.
method Feature-wise sigmoid gating mechanism for combining character and word-level representations.
result Modeling characters improves final word and sentence representations, especially for less frequent words.
Predicting the click-through rate of an advertisement is a critical component of online advertising platforms. In sponsored search, the click-through rate estimates the probability that a displayed advertisement is clicked by a user after she submits a query to the search engine. Commercial search engines typically rel…
Dynamic segmentation algorithm improves NMT performance by favoring character-level processing.
problem Suboptimal static segmentation choices in NMT systems.
method Adaptive Computation Time algorithm for dynamic segmentation, trainable end-to-end.
result The model prefers character-level processing when given the freedom to navigate different segmentation levels.
We describe a simple neural language model that relies only on character-level inputs. Predictions are still made at the word-level. Our model employs a convolutional neural network (CNN) and a highway network over characters, whose output is given to a long short-term memory (LSTM) recurrent neural network language mo…
We introduce multiplicative LSTM (mLSTM), a recurrent neural network architecture for sequence modelling that combines the long short-term memory (LSTM) and multiplicative recurrent neural network architectures. mLSTM is characterised by its ability to have different recurrent transition functions for each possible inp…
A deep learning approach generates math word problems in multiple languages.
problem Template-based mechanisms for generating mathematical word problems lack customizability and creativity.
method Character Level Long Short Term Memory Network (LSTM) and POS tags are used to generate and resolve constraints in generated problems.
result The approach generates accurate math word problems in English and Sinhala with over 90% accuracy.
End-to-end character-level model for text generation without delexicalization.
problem Generating text without delexicalization and tokenization.
method Character-level sequence-to-sequence model with attention mechanism, copy mechanism, and transfer learning.
result Competitive performance in text generation metrics.
In this paper, we propose a probabilistic parsing model, which defines a proper conditional probability distribution over non-projective dependency trees for a given sentence, using neural representations as inputs. The neural network architecture is based on bi-directional LSTM-CNNs which benefits from both word- and …
Classifies CAD model descriptions and names from product websites.
problem Distinguishing product descriptions from other text and identifying product names.
method Paragraph vectors, character-level LSTM, word embeddings LSTM tagger.
result Promising results for distinguishing product descriptions and names.
Neural machine translation aims at building a single large neural network that can be trained to maximize translation performance. The encoder-decoder architecture with an attention mechanism achieves a translation performance comparable to the existing state-of-the-art phrase-based systems on the task of English-to-Fr…
State-of-the-art sequence labeling systems traditionally require large amounts of task-specific knowledge in the form of hand-crafted features and data pre-processing. In this paper, we introduce a novel neutral network architecture that benefits from both word- and character-level representations automatically, by usi…
We propose a generalization of neural network sequence models. Instead of predicting one symbol at a time, our multi-scale model makes predictions over multiple, potentially overlapping multi-symbol tokens. A variation of the byte-pair encoding (BPE) compression algorithm is used to learn the dictionary of tokens that …
Deep transformer models outperform RNNs in character-level language modeling.
problem Improving character-level language modeling performance.
method A deep (64-layer) transformer model with fixed context and auxiliary losses.
result Achieved state-of-the-art performance on text8 and enwik8 benchmarks.
Proposes BU-SPO method to improve text classification robustness.
problem Vulnerability of deep models in text classification.
method Bigram and unigram based adaptive Semantic Preservation Optimization (BU-SPO) method.
result Achieves highest attack success rates and semantic similarity by changing the smallest number of words.
Charmer improves character-level adversarial attacks for language models.
problem Efficiency and effectiveness of character-level adversarial attacks for language models.
method Query-based adversarial attack method that maintains semantic similarity.
result Charmer achieves high attack success rate and similar adversarial examples.
Bayesian algorithm improves word representations using semantic taxonomy.
problem Improving word representations in semantic taxonomy.
method Bayesian Hierarchical Words Representation (BHWR) learning algorithm combining Variational Bayes and semantic taxonomy modeling.
result BHWR produces better representations for rare words.
HW2MP-GAN tackles ancient handwritten text recognition.
problem Automatic text recognition from ancient handwritten records.
method Conditional Generative Adversarial Network (HW2MP-GAN) with Sliced Wasserstein distance and U-Net architectures.
result HW2MP-GAN outperforms state-of-the-art models in image-to-image translation and handwritten recognition.
A transformer model improves spell correction with hierarchical attention.
problem Improving spell correction accuracy and speed.
method Multi encoder-single decoder transformer architecture with hierarchical attention.
result Significant improvement in CER, WER, and SER error rates.
Estimator Vectors learns OOV word embeddings using subword and context clues.
problem Lack of OOV word representations in neural network models.
method Jointly learns word, subword, and context clue representations.
result Strong estimates for OOV words via combined subword and context clue embeddings.
Recent work on learning multilingual word representations usually relies on the use of word-level alignements (e.g. infered with the help of GIZA++) between translated sentences, in order to align the word embeddings in different languages. In this workshop paper, we investigate an autoencoder model for learning multil…
GROVER improves word representations by gradually adding random noises during training.
problem Improving word representations for better model performance.
method Gradually adding random noises to word embeddings during training.
result GROVER improves model performances on most text classification datasets.
This paper proposes using contextualized word representations for better concept taxonomies.
problem Current taxonomy learning systems define concepts as single words, limiting their semantic understanding.
method Defines concepts as synsets, learns density-based approximations of contextualized word representations, and measures similarity and hypernymy.
result Contextualized word representations can improve the accuracy of concept taxonomies.
Paper finds better words for topic models by reranking top words.
problem Top words in topic models are not always representative.
method Reranking words by considering marginal probability over every topic.
result Reranked top words are more representative of topics.
The abstract explains how word and relation representations capture semantic meaning.
problem Understanding how word and relation representations capture semantic meaning.
method Theoretical justification and extension of geometric relationships between word embeddings and knowledge graph representations.
result The geometric relationships between word embeddings correspond to semantic relations between words and entities in knowledge graphs.
Neural language models (LMs) based on recurrent neural networks (RNN) are some of the most successful word and character-level LMs. Why do they work so well, in particular better than linear neural LMs? Possible explanations are that RNNs have an implicitly better regularization or that RNNs have a higher capacity for …
This paper assesses biases in contextualized word representations.
problem Analyzing biases in contextualized word representations.
method Proposes assessing bias at the contextual word level, capturing contextual effects of bias.
result Demonstrates evidence of bias in contextual word models, including racial bias and exacerbated effects for intersectional minorities.
Probabilistic FastText captures multiple word senses and sub-word structures.
problem Capturing multiple word senses and sub-word structures in word embeddings.
method Probabilistic FastText uses Gaussian mixture densities to represent words, sharing statistical strength across sub-word structures and capturing different word senses.
result Probabilistic FastText outperforms existing models on word-similarity benchmarks and discerning different meanings.
New methods for unsupervised learning of word and entity representations.
problem Learning distributed representations of words and entities from text and knowledge bases.
method MVLSA for words and NVSE for entities, both unsupervised learning methods.
result MVLSA and NVSE outperform state-of-the-art models in word and entity representation learning.
We propose a new method for learning word representations using hierarchical regularization in sparse coding inspired by the linguistic study of word meanings. We show an efficient learning algorithm based on stochastic proximal methods that is significantly faster than previous approaches, making it possible to perfor…
Trains word embeddings from music and text data to link music contexts.
problem Varying vocabulary size and musical relevance in word embeddings.
method Combines general text and music-specific data to train word embeddings.
result Trained embeddings better associate music contexts with compositions.
The recently introduced continuous Skip-gram model is an efficient method for learning high-quality distributed vector representations that capture a large number of precise syntactic and semantic word relationships. In this paper we present several extensions that improve both the quality of the vectors and the traini…
Real-valued word representations have transformed NLP applications; popular examples are word2vec and GloVe, recognized for their ability to capture linguistic regularities. In this paper, we demonstrate a {\em very simple}, and yet counter-intuitive, postprocessing technique -- eliminate the common mean vector and a f…
Paper proposes a new method for sentence embeddings using weighted word vectors.
problem Improving sentence embeddings for natural language processing tasks.
method A simple sentence embedding method using weighted average of word vectors followed by soft projection.
result Demonstrates effectiveness on clinical semantic textual similarity task.
Top2Vec finds topic vectors from documents and words without needing stop words or custom settings.
problem Topic modeling weaknesses, including needing known topics, stop words, and custom settings.
method Joint document and word semantic embedding to find topic vectors automatically.
result Top2Vec finds more informative and representative topics than probabilistic models.
Survey of word embedding techniques for NLP.
problem Creating effective word representations for natural language processing.
method Describes recent strategies for fixed-length, dense word embeddings.
result Word embeddings encode syntactic and semantic information and improve NLP tasks.
Fruit fly brain network learns word embeddings using sparse binary codes.
problem Learning semantic word representations from text.
method Inspired by mushroom body neural network, sparse binary hash codes.
result Fruit fly network achieves comparable NLP performance with reduced resources.
Word embeddings are representations of individual words of a text document in a vector space and they are often use- ful for performing natural language pro- cessing tasks. Current state of the art al- gorithms for learning word embeddings learn vector representations from large corpora of text documents in an unsu- pe…
Trellis networks improve sequence modeling performance.
problem Sequence modeling challenges.
method Temporal convolutional network with weight tying and direct input injection.
result Trellis networks outperform state-of-the-art methods on benchmarks.
Vector representations of words have heralded a transformational approach to classical problems in NLP; the most popular example is word2vec. However, a single vector does not suffice to model the polysemous nature of many (frequent) words, i.e., words with multiple meanings. In this paper, we propose a three-fold appr…
This paper classifies tweets into positive and negative sentiments using distributed word and sentence representations.
problem Classifying tweets into positive and negative sentiments.
method Used distributed representations of words and sentences, and LSTM and CNN networks for classification.
result Achieved accuracies as high as 81%.
CPGBN analyzes text sequences, capturing word order for better topic extraction.
problem Lack of word order in traditional text representations.
method CPFA processes words as sequences, CPGBN adds hierarchical topic modeling.
result CPGBN extracts high-quality latent representations capturing word order.
Approaches KL divergence for learning multi-sense word distributions.
problem Capturing the polysemy and uncertainty of words in word embeddings.
method Modeling words as multi-sense Gaussian mixtures and using KL divergence for learning.
result The proposed approach effectively captures word entailment and distribution similarity.
Model learns multilingual word representations robust to noise.
problem Learning multilingual word representations in noisy environments.
method Fit a generative latent variable model to a multilingual dictionary.
result Competitive multilingual embeddings across various tasks.
A new method for document network embedding interprets and generalizes well.
problem Lack of interpretability and generalization to new documents in existing methods.
method Introduces Topic-Word Attention (TWA) and Inductive Document Network Embedding (IDNE) to generate document representations.
result Achieves state-of-the-art performance on various networks and produces meaningful representations.
BERT-based word embeddings improve active learning for text datasets.
problem Efficiently labelling large text datasets for machine learning.
method Evaluation of text representation mechanisms (BERT vs. bag of words) in active learning.
result BERT-based word embeddings significantly improve active learning performance.