Remove Computational Linguistics Remove Computer Vision Remove Metadata Remove Natural Language Processing
article thumbnail

All Languages Are NOT Created (Tokenized) Equal

Topbots

Language Disparity in Natural Language Processing This digital divide in natural language processing (NLP) is an active area of research. 70% of research papers published in a computational linguistics conference only evaluated English.[ Association for Computational Linguistics.

article thumbnail

Modular Deep Learning

Sebastian Ruder

d) Hypernetwork: A small separate neural network generates modular parameters conditioned on metadata.  Instead of learning module parameters directly, they can be generated using an auxiliary model (a hypernetwork) conditioned on additional information and metadata. Computer vision and cross-modal learning.

article thumbnail

The State of Multilingual AI

Sebastian Ruder

Developing models that work for more languages is important in order to offset the existing language divide and to ensure that speakers of non-English languages are not left behind, among many other reasons. Writing System and Speaker Metadata for 2,800+ Language Varieties. Lucassen, T., 2340–2354).