Papers by Joseph Le Roux
Higher-Order Dependency Parsing for Arc-Polynomial Score Functions via Gradient-Based Methods and Genetic Algorithm (2022.aacl-main)
Copied to clipboard
| Challenge: | Existing methods for higher-order dependency parsing are based on arc-polynomials . a score function is linear in arc variables, while for second-order models it is linear . |
| Approach: | They propose a method which takes advantage of arc-polynomials score functions . they use coordinate ascent and genetic search to iteratively update a candidate parse . |
| Outcome: | The proposed method matches state-of-the-art second order parsers on three standard datasets. |
Multitask Easy-First Dependency Parsing: Exploiting Complementarities of Different Dependency Representations (2020.coling-main)
Copied to clipboard
| Challenge: | Existing dependency parsing models for Arabic use complementary annotations, CATiB and UD treebanks, and partially created trees for one annotation are also available to the other as features for the score function. |
| Approach: | They propose to use Arabic dependency annotations to parse projective dependency trees using CATiB and UD treebanks. |
| Outcome: | The proposed model gives 9.9% error reduction on CATiB and 6.1% on UD compared to a strong baseline and ablation tests show that the main contribution is given by sharing tree representation between tasks, and not simply sharing biLSTM layers as is often performed in NLP multitask systems. |
Exploiting Inductive Bias in Transformers for Unsupervised Disentanglement of Syntax and Semantics with VAEs (2022.naacl-main)
Copied to clipboard
| Challenge: | Existing models for text generation do not need syntactic information such as constituency parses or semantic information such a paraphrase pairs. |
| Approach: | They propose a generative model which exhibits disentangled latent representations of syntax and semantics by using Attention in its decoder. |
| Outcome: | The proposed model outperforms supervised models on syntax/semantics transfer and shows that it can read latent variables with keys and values. |