Gábor Melis

8 papers · 2016–2022 · 4 conferences · across top CS/AI conferences

Achievements

🗺️ Taxonomy Completionist (12) 🌉 Interdisciplinary Bridge 🧭 Keyword Pioneer 🌍 Conference Polyglot (4) 🏃 Academic Marathon (6) 🐝 Cross-Pollinator (13) 🚀 Conference Pioneer 🔥 Unstoppable (5)

Conferences

ICLR (4) EMNLP (2) JMLR (1) NAACL (1)

Top co-authors

Phil Blunsom (5) Chris Dyer (4) Wang Ling (3) Lei Yu (3) Dani Yogatama (2) Tomas Kocisky (2) Adhiguna Kuncoro (2) András György (1) Lingpeng Kong (1) Elman Mansimov (1)

Keywords

variational inference (1) neural machine translation (1) language modeling (1) document-level translation (1) mutual information (1) context modeling (1) grammar induction (1) constituency parsing (1) latent variable model (1) latent variable (1) bleu score (1) variational autoencoder (1) language model (1) posterior collapse (1) monte-carlo objective (1) recurrent neural network grammar (1) sentence-level model (1) monte carlo objective (1)

Papers

Mutual Information Constraints for Monte-Carlo Objectives to Prevent Posterior Collapse Especially in Language Modelling JMLR 2022

Capturing document context inside sentence-level neural machine translation models with self-training EMNLP 2021

Mogrifier LSTM ICLR 2020

Unsupervised Recurrent Neural Network Grammars NAACL 2019

Variational Smoothing in Recurrent Neural Network Language Models ICLR 2019

On the State of the Art of Evaluation in Neural Language Models ICLR 2018

Memory Architectures in Recurrent Neural Network Language Models ICLR 2018

Semantic Parsing with Semi-Supervised Sequential Autoencoders EMNLP 2016