AI Paper Atlas

Research map / Continual learning / Online and replay-based

Continual Learning for Named Entity Recognition

Natawut Monaikul, Giuseppe Castellucci, Simone Filice, Oleg Rokhlenko

AAAI 2021 · Continual learning › Online and replay-based · approach: Unified and unsupervised continual frameworks · 66 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add knowledge distillation from the previous model rather than training only on hard teacher labels (ablation)ExtendNER · CoNLL-03 test, all steps and permutationssentence-level F1: 0.37 pointsImproved
Add knowledge distillation from the previous model rather than training only on hard teacher labels (ablation)ExtendNER · OntoNotes test, all steps and permutationssentence-level F1: 0.81 pointsImproved
Extend the output layer instead of using separate output layers merged with heuristics (ablation)NER continual learning · CoNLL-03 and OntoNotesmacro-F1: 0.06 points; 0.04 pointsNo clear effect
Initialize each update from the previous model rather than train from scratch on the current sliceExtendNER · CoNLL-03 and OntoNotes test, all stepssentence-level F1Improved
Use partially annotated new data rather than fully annotated data for old and new entity typesExtendNER · CoNLL-03 test, all stepssentence-level F1: 0.24Hurt
Use partially annotated new data rather than fully annotated data for old and new entity typesExtendNER · OntoNotes test, all stepssentence-level F1: 0.78Hurt
Allow encoder parameters to update while training on the new entity type, while keeping the old output layer frozenTransfer learning for NER · CoNLL-03 test, step 2F1: 83.77 -> 34.95Hurt
Freeze the encoder while training the new entity output layerTransfer learning for NER · CoNLL-03 test, step 2F1: 57.49Hurt
Add new entity types sequentially with ExtendNER and KDExtendNER · CoNLL-03 test, previously learned entity typesF1: +1.308Improved
Add new entity types sequentially with ExtendNER and KDExtendNER · OntoNotes test, previously learned entity typesF1: +0.0341No clear effect
Continue adding entity types with ExtendNER and KDExtendNER · OntoNotes test, e1 from step 1 to step 6F1: 78.03 -> 77.88No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy