David Chiang

Leveraging Translations for Speech Transcription in Low-resource Settings

2018

Recently proposed data collection frameworks for endangered language documentation aim not only to collect speech in the language of interest, but also to collect translations into a high-resource language that will render the collected resource …

Decoding with Finite-State Transducers on GPUs

2017 · arXiv (Cornell University)

Weighted finite automata and transducers (including hidden Markov models and conditional random fields) are widely used in natural language processing (NLP) to perform tasks such as morphological analysis, part-of-speech tagging, chunking, named entity recognition, speech …

An Unsupervised Probability Model for Speech-to-Translation Alignment of Low-Resource Languages

2016

For many low-resource languages, spoken language resources are more likely to be annotated with translations than with transcriptions. Translated speech data is potentially valuable for documenting endangered languages or for training speech translation systems. A …

A case study on using speech-to-translation alignments for language\n documentation

2017 · arXiv (Cornell University)

For many low-resource or endangered languages, spoken language resources are\nmore likely to be annotated with translations than with transcriptions. Recent\nwork exploits such annotations to produce speech-to-translation alignments,\nwithout access to any text transcriptions. We investigate whether …

A Continuum of Generation Tasks for Investigating Length Bias and Degenerate Repetition

2022 · arXiv (Cornell University)

Language models suffer from various degenerate behaviors. These differ between tasks: machine translation (MT) exhibits length bias, while tasks like story generation exhibit excessive repetition. Recent work has attributed the difference to task constrainedness, but …

Tighter Bounds on the Expressivity of Transformer Encoders

2023 · arXiv (Cornell University)

Characterizing neural networks in terms of better-understood formal systems has the potential to yield new insights into the power and limitations of these networks. Doing so for transformers remains an active area of research. Bhattamishra …

Theory of Neural Language Models (Dagstuhl Seminar 25282)

2026 · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics)

This report documents the program and the outcomes of Dagstuhl Seminar 25282 "Theory of Neural Language Models". The seminar aimed to bring researchers together to lay a foundation for continued work on the theory of …

An Attentional Model for Speech Translation Without Transcription

2016

Long Duong, Antonios Anastasopoulos, David Chiang, Steven Bird, Trevor Cohn. Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2016.

Transfer Learning across Low-Resource, Related Languages for Neural Machine Translation

2017 · arXiv (Cornell University)

We present a simple method to improve neural translation of a low-resource language pair using parallel data from a related, also low-resource, language pair. The method is based on the transfer method of Zoph et …

Tied Multitask Learning for Neural Speech Translation

2018

Antonios Anastasopoulos, David Chiang. Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers). 2018.