RepL4NLP 2024
The 9th Workshop on Representation Learning for NLP, organized by SIGREP, aims to continue the success of the first eight workshops in the series. The RepL4NLP workshop was introduced as a synthesis of several years of independent *CL workshops focusing on vector space models of meaning, compositionality, and the application of deep neural networks and spectral methods to NLP. With the widespread adoption of neural network architectures (and especially Transformer models), representation learning has become a central concept in NLP research. RepL4NLP provides a forum for discussing recent advances on these topics, ranging from the development of new representations for various tasks and applications to the analysis of existing representations, e.g., with respect to generalization and robustness.
In most NLP applications, the goal is to understand, interpret, and generate human language text extract linguistic information from raw text or to transform linguistic observations into an alternative form, e.g. from speech to text or from one language to another. Traditional statistical models relied heavily on discrete categories that are parameterized individually (e.g., each word or n- gram in a language model gets its own parameter); Representation learning—whether using spectral methods, probabilistic models, or deep neural networks—offers an alternative: by learning continuous, multidimensional representations of discrete objects (e.g., words, tags, labels, phrases, sentences), models can find representations that enable them to solve tasks more effectively. As learning representations has now been adopted as the de-facto approach to building models in NLP, their widespread use raises important questions to- wards further understanding them.
Key Dates
Direct paper submission deadline: May 17, 2024
ARR commitment deadline: June 1, 2024
Notification of acceptance: June 17, 2024
Camera-ready papers due: July 1, 2024
Workshop date: Aug 16, 2024
Keynote Speakers
TBD.
Topics
TBD.
Sponsors
TBD.