AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Back to the Future – Temporal Adaptation of Text Representations

Johannes Bjerva, Wouter Kouw, Isabelle Augenstein

AAAI 2020 · Large language models › Language, safety and interpretability · approach: Reasoning mechanisms and failure modes · 16 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 semi-supervised subspace alignment (10 labelled target points per class) to previous-year trainingBERT + SVM · PeerRead acceptanceaccuracy (avg): 53.75 -> 68.09Improved
semi-supervised SSA from past data vs training on all dataBERT + SVM · PeerRead acceptanceaccuracy (avg): 58.82 -> 68.09Improved
use unsupervised instead of semi-supervised alignment (ablation)SSA · PeerRead acceptanceaccuracy (avg): 68.09 -> 52.05Hurt
add unbounded multi-step extension to semi-supervised SSA (ablation)SSA · PeerRead acceptanceaccuracy (avg): 68.09 -> 69.44Improved
add k-means cluster IDs (instance similarity) to unbounded SSA (ablation)SSA unbounded · PeerRead acceptanceaccuracy (avg): 69.44 -> 69.56Improved
semi-supervised SSA vs previous-year trainingBERT-cased + MLP · Broad Twitter Corpus NERF1 (avg): 56.85 -> 68.75Improved
unsupervised alignment on NER (ablation)BERT-cased + MLP · Broad Twitter Corpus NERF1 (avg): 56.85 -> 46.34Hurt
add cluster IDs to unbounded SSA (ablation)SSA · Broad Twitter Corpus NERF1 (avg): 68.75 -> 71.35Improved
training on same-year in-domain data vs semi-supervised SSABERT-cased + MLP · Broad Twitter Corpus NERF1 (avg): 60.00 vs 68.75Mixed
semi-supervised SSA vs previous rumour trainingBERT + SVM · RumourEval-2019 SDQCF1 (avg): 37.00 -> 38.65Improved
unsupervised alignment vs previous rumour training (ablation)BERT + SVM · RumourEval-2019 SDQCF1 (avg): 37.00 -> 37.00No clear effect
add unbounded extension (ablation)semi-supervised SSA · RumourEval-2019 SDQCF1 (avg): 38.65 -> 37.90No 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