arXiv subjects

Nathaniel Robinson

Publications and source records attributed to Nathaniel Robinson.

6 recordsLinked to original sources

arXiv2024

Can Large Language Models Code Like a Linguist?: A Case Study in Low Resource Sound Law Induction

Historical linguists have long written a kind of incompletely formalized ''program'' that converts reconstructed words in an ancestor language into words in one of its attested descendants that consist of a series of ordered string rewrite functions (called sound laws). They do this by observing pairs of words in the reconstructed language (protoforms) and the descendent language (reflexes) and constructing a program that transforms protoforms into reflexes. However, writing these programs is error-prone and time-consuming. Prior work has successfully scaffolded this process computationally, but fewer researchers have tackled Sound Law Induction (SLI), which we approach in this paper by casting it as Programming by Examples. We propose a language-agnostic solution that utilizes the programming ability of Large Language Models (LLMs) by generating Python sound law programs from sound change examples. We evaluate the effectiveness of our approach for various LLMs, propose effective methods to generate additional language-agnostic synthetic data to fine-tune LLMs for SLI, and compare our method with existing automated SLI methods showing that while LLMs lag behind them they can complement some of their weaknesses.

cs.CL↗

arXiv2023

PWESuite: Phonetic Word Embeddings and Tasks They Facilitate

Mapping words into a fixed-dimensional vector space is the backbone of modern NLP. While most word embedding methods successfully encode semantic information, they overlook phonetic information that is crucial for many tasks. We develop three methods that use articulatory features to build phonetically informed word embeddings. To address the inconsistent evaluation of existing phonetic word embedding methods, we also contribute a task suite to fairly evaluate past, current, and future methods. We evaluate both (1) intrinsic aspects of phonetic word embeddings, such as word retrieval and correlation with sound similarity, and (2) extrinsic performance on tasks such as rhyme and cognate detection and sound analogies. We hope our task suite will promote reproducibility and inspire future phonetic embedding research.

cs.CL↗

arXiv2022

When Is TTS Augmentation Through a Pivot Language Useful?

Developing Automatic Speech Recognition (ASR) for low-resource languages is a challenge due to the small amount of transcribed audio data. For many such languages, audio and text are available separately, but not audio with transcriptions. Using text, speech can be synthetically produced via text-to-speech (TTS) systems. However, many low-resource languages do not have quality TTS systems either. We propose an alternative: produce synthetic audio by running text from the target language through a trained TTS system for a higher-resource pivot language. We investigate when and how this technique is most effective in low-resource settings. In our experiments, using several thousand synthetic TTS text-speech pairs and duplicating authentic data to balance yields optimal results. Our findings suggest that searching over a set of candidate pivot languages can lead to marginal improvements and that, surprisingly, ASR performance can by harmed by increases in measured TTS quality. Application of these findings improves ASR by 64.5\% and 45.0\% character error reduction rate (CERR) respectively for two low-resource languages: Guaran\'i and Suba.

cs.CL↗

arXiv2020

A Methodology to Derive Global Maps of Leaf Traits Using Remote Sensing and Climate Data

This paper introduces a modular processing chain to derive global high-resolution maps of leaf traits. In particular, we present global maps at 500 m resolution of specific leaf area, leaf dry matter content, leaf nitrogen and phosphorus content per dry mass, and leaf nitrogen/phosphorus ratio. The processing chain exploits machine learning techniques along with optical remote sensing data (MODIS/Landsat) and climate data for gap filling and up-scaling of in-situ measured leaf traits. The chain first uses random forests regression with surrogates to fill gaps in the database ($> 45 \% $ of missing entries) and maximize the global representativeness of the trait dataset. Along with the estimated global maps of leaf traits, we provide associated uncertainty estimates derived from the regression models. The process chain is modular, and can easily accommodate new traits, data streams (traits databases and remote sensing data), and methods. The machine learning techniques applied allow attribution of information gain to data input and thus provide the opportunity to understand trait-environment relationships at the plant and ecosystem scales.

stat.AP↗

arXiv2020

Interpolation and Gap Filling of Landsat Reflectance Time Series

Products derived from a single multispectral sensor are hampered by a limited spatial, spectral or temporal resolutions. Image fusion in general and downscaling/blending in particular allow to combine different multiresolution datasets. We present here an optimal interpolation approach to generate smoothed and gap-free time series of Landsat reflectance data. We fuse MODIS (moderate-resolution imaging spectroradiometer) and Landsat data globally using the Google Earth Engine (GEE) platform. The optimal interpolator exploits GEE ability to ingest large amounts of data (Landsat climatologies) and uses simple linear operations that scale easily in the cloud. The approach shows very good results in practice, as tested over five sites with different vegetation types and climatic characteristics in the contiguous US.

eess.SP↗

arXiv2020

Towards Neural Programming Interfaces

It is notoriously difficult to control the behavior of artificial neural networks such as generative neural language models. We recast the problem of controlling natural language generation as that of learning to interface with a pretrained language model, just as Application Programming Interfaces (APIs) control the behavior of programs by altering hyperparameters. In this new paradigm, a specialized neural network (called a Neural Programming Interface or NPI) learns to interface with a pretrained language model by manipulating the hidden activations of the pretrained model to produce desired outputs. Importantly, no permanent changes are made to the weights of the original model, allowing us to re-purpose pretrained models for new tasks without overwriting any aspect of the language model. We also contribute a new data set construction algorithm and GAN-inspired loss function that allows us to train NPI models to control outputs of autoregressive transformers. In experiments against other state-of-the-art approaches, we demonstrate the efficacy of our methods using OpenAI's GPT-2 model, successfully controlling noun selection, topic aversion, offensive speech filtering, and other aspects of language while largely maintaining the controlled model's fluency under deterministic settings.

cs.CL↗