Latent Sequence Decompositions

We present the Latent Sequence Decompositions (LSD) framework. LSD decomposes sequences with variable lengthed output units as a function of both the input sequence and the output sequence. We present a training algorithm which samples valid extensions and an approximate decoding algorithm. We experiment with the Wall Street Journal speech recognition task. Our LSD model achieves 12.9% WER compared to a character baseline of 14.8% WER. When combined with a convolutional network on the encoder, we achieve 9.6% WER.

• Publication year

  2016

• Venue

  International Conference on Learning Representations

• Publication date

  2016-10-10

• Fields of study

  Mathematics, Computer Science

• Identifiers
  arXiv 1610.03035
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Under review as a conference paper at ICLR 2020 many domain adaptation methods

  2019cited by this paper
• Published as a conference paper at ICLR 2018 S IMULATING A CTION D YNAMICS WITH N EURAL P ROCESS N ETWORKS

  2018cited by this paper
• Advances in all-neural speech recognition

  2016cited by this paper
• Achieving Open Vocabulary Neural Machine Translation with Hybrid Word-Character Models

  2016cited by this paper
• Wav2Letter: an End-to-End ConvNet-based Speech Recognition System

  2016influential reference
• Very deep convolutional networks for end-to-end speech recognition

  2016cited by this paper
• Latent Predictor Networks for Code Generation

  2016cited by this paper
• Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation

  2016cited by this paper
• Order Matters: Sequence to sequence for sets

  2015cited by this paper
• Task Loss Estimation for Sequence Prediction

  2015influential reference
• Show, Attend and Tell: Neural Image Caption Generation with Visual Attention

  2015cited by this paper
• Attention-Based Models for Speech Recognition

  2015cited by this paper
• RECURRENT NEURAL NETWORKS

  2015cited by this paper
• Neural Machine Translation of Rare Words with Subword Units

  2015cited by this paper
• Listen, attend and spell: A neural network for large vocabulary conversational speech recognition

  2015influential reference
• End-to-end attention-based large vocabulary speech recognition

  2015influential reference
• Grammar as a Foreign Language

  2014cited by this paper
• Towards End-To-End Speech Recognition with Recurrent Neural Networks

  2014cited by this paper
• Show and tell: A neural image caption generator

  2014cited by this paper
• Addressing the Rare Word Problem in Neural Machine Translation

  2014cited by this paper
• On Using Very Large Target Vocabulary for Neural Machine Translation

  2014cited by this paper
• Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

  2014influential reference
• Adam: A Method for Stochastic Optimization

  2014cited by this paper
• Sequence to Sequence Learning with Neural Networks

  2014cited by this paper
• Neural Machine Translation by Jointly Learning to Align and Translate

  2014influential reference
• A Clockwork RNN

  2014cited by this paper
• First-Pass Large Vocabulary Continuous Speech Recognition using Bi-Directional Recurrent DNNs

  2014cited by this paper
• Learning Lexicons From Speech Using a Pronunciation Mixture Model

  2013cited by this paper
• Acoustic data-driven pronunciation lexicon for large vocabulary speech recognition

  2013cited by this paper
• Japanese and Korean voice search

  2012cited by this paper
• Practical Variational Inference for Neural Networks

  2011cited by this paper
• The Kaldi Speech Recognition Toolkit

  2011cited by this paper
• Conference Paper

  2009cited by this paper
• Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks

  2006cited by this paper
• Automatic generation of subword units for speech recognition systems

  2002cited by this paper
• Reinforcement Learning: An Introduction

  1998cited by this paper
• Bidirectional recurrent neural networks

  1997cited by this paper
• Long Short-Term Memory

  1997influential reference
• Hierarchical Recurrent Neural Networks for Long-Term Dependencies

  1995cited by this paper

• Green Entrepreneurship for Business Sustainability: Do Environmental Dynamism and Green Structural Capital Matter?

  2024cites this paper
• SelfSeg: A Self-supervised Sub-word Segmentation Method for Neural Machine Translation

  2023cites this paper
• End-to-End Speech Recognition: A Survey

  2023cites this paper
• Prefix-Level Detection and Autocorrection of Keyboard Input Errors

  2023cites this paper
• Syntactic Inductive Biases for Natural Language Processing

  2022cites this paper
• Analyze The Factors of Organizational Citizenship Behavior on Starbucks’ Manager in Surabaya

  2021cites this paper
• Acoustic Data-Driven Subword Modeling for End-to-End Speech Recognition

  2021cites this paper
• Phonetically Induced Subwords for End-to-End Speech Recognition

  2021cites this paper
• Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLP

  2021cites this paper
• Enhancing Handwritten Text Recognition with N-gram sequence decomposition and Multitask Learning

  2020cites this paper
• Do End-to-End Speech Recognition Models Care About Context?

  2020cites this paper
• Automatic Segmented-Syllable and Deep Learning-Based Indonesian Audiovisual Speech Recognition

  2020cites this paper
• Optimizing Word Segmentation for Downstream Task

  2020cites this paper
• Imputer: Sequence Modelling via Imputation and Dynamic Programming

  2020influential citation
• Learning a Subword Inventory Jointly with End-to-End Automatic Speech Recognition

  2020influential citation
• Dynamic Programming Encoding for Subword Segmentation in Neural Machine Translation

  2020cites this paper
• Differentiable Weighted Finite-State Transducers

  2020cites this paper
• A systematic comparison of grapheme-based vs. phoneme-based label units for encoder-decoder-attention models.

  2020cites this paper
• Interacting with Smart Devices - Advancements in Gesture Recognition and Augmented Reality

  2020cites this paper
• Data-driven deep modeling and training for automatic speech recognition

  2020cites this paper
• Learning World Graphs to Accelerate Hierarchical Reinforcement Learning

  2019cites this paper
• A comprehensive analysis on attention models

  2019cites this paper
• Language Model for Query Auto-Completion

  2019cites this paper
• Integrating Source-Channel and Attention-Based Sequence-to-Sequence Models for Speech Recognition

  2019cites this paper
• Subword Language Model for Query Auto-Completion

  2019cites this paper
• An Overview of End-to-End Automatic Speech Recognition

  2019cites this paper
• Towards Using Context-Dependent Symbols in CTC Without State-Tying Decision Trees

  2019cites this paper
• Sequence-to-sequence learning for machine translation and automatic differentiation for machine learning software tools

  2019cites this paper
• Spoken command recognition for robotics

  2019cites this paper
• A Fully Differentiable Beam Search Decoder

  2019cites this paper
• Speech Recognition with Augmented Synthesized Speech

  2019cites this paper
• CompILE: Compositional Imitation Learning and Execution

  2018cites this paper
• Neural Lattice Language Models

  2018influential citation
• Advancing Acoustic-to-Word CTC Model

  2018cites this paper
• Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates

  2018cites this paper
• Hybrid CTC-Attention based End-to-End Speech Recognition using Subword Units

  2018cites this paper
• Acoustic-to-Word Recognition with Sequence-to-Sequence Models

  2018influential citation
• Speech-Transformer: A No-Recurrence Sequence-to-Sequence Model for Speech Recognition

  2018cites this paper
• Optimal Completion Distillation for Sequence Learning

  2018cites this paper
• Bytes Are All You Need: End-to-end Multilingual Speech Recognition and Synthesis with Bytes

  2018cites this paper
• Unsupervised Word Discovery with Segmental Neural Language Models

  2018cites this paper
• Compositional Imitation Learning: Explaining and executing one task at a time

  2018cites this paper
• Advancing Acoustic-to-Word CTC Model With Attention and Mixed-Units

  2018cites this paper
• Partial Greedy Algorithm to Extract a Minimum Phonetically-and-Prosodically Rich Sentence Set

  2018cites this paper
• Hierarchical Multitask Learning With CTC

  2018cites this paper
• Low-Frequency Character Clustering for End-to-End ASR System

  2018cites this paper
• Learning to Discover, Ground and Use Words with Segmental Neural Language Models

  2018cites this paper
• State-of-the-Art Speech Recognition with Sequence-to-Sequence Models

  2017cites this paper
• Building Competitive Direct Acoustics-to-Word Models for English Conversational Speech Recognition

  2017cites this paper
• Subword and Crossword Units for CTC Acoustic Models

  2017cites this paper
• Improving the Performance of Online Neural Transducer Models

  2017cites this paper
• Hybrid CTC/Attention Architecture for End-to-End Speech Recognition

  2017cites this paper
• Simplified End-to-End MMI Training and Voting for ASR

  2017cites this paper
• Advances in Joint CTC-Attention Based End-to-End Speech Recognition with a Deep CNN Encoder and RNN-LM

  2017cites this paper
• Multi-level language modeling and decoding for open vocabulary end-to-end speech recognition

  2017cites this paper
• A Comparison of Sequence-to-Sequence Models for Speech Recognition

  2017cites this paper
• Gram-CTC: Automatic Unit Selection and Target Decomposition for Sequence Labelling

  2017influential citation
• Exploring architectures, data and units for streaming end-to-end speech recognition with RNN-transducer

  2017cites this paper
• Regularizing Neural Networks by Penalizing Confident Output Distributions

  2017influential citation
• Multiscale sequence modeling with a learned dictionary

  2017cites this paper
• End-to-End Speech Recognition Models

  2016cites this paper
• Towards Better Decoding and Language Model Integration in Sequence to Sequence Models

  2016cites this paper