Research map / Probabilistic and scientific modeling / Variational and generative sampling
Learning semantic similarity in a continuous space
Michel Deudon
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add generative pretraining (repeat/reformulate) before siamese training (ablation) | VAR-Siamese · Quora question pairs test | accuracy: 62.48 -> 88.86 | Improved |
| pretrain with repeat/reformulate instead of repeat only (ablation) | VAR-Siamese · Quora question pairs test | accuracy: 88.24 -> 88.86 | Improved |
| pretrain with repeat/reformulate instead of repeat only (AUC after siamese training) (ablation) | VAR-siamese · Quora test W2 AUC | AUC: 86.74 -> 86.27 | Hurt |
| pretrain VAE with repeat/reformulate instead of repeat only (no siamese training) (ablation) | VAE · Quora test W2 AUC | AUC: 69.91 -> 77.44 | Improved |
| add discriminative variational siamese training on top of VAE (ablation) | VAE (repeat/reformulate) · Quora test W2 AUC | AUC: 77.44 -> 86.27 | Improved |
| replace Word Mover's Distance (bag of words, W1) with VAR-siamese W2 | Quora duplicate detection · Quora test | AUC: 73.10 -> 86.74 | Improved |
| add Hadamard product to MLP input (ablation) | VAR-siamese · Quora dev | accuracy: marginal 1% | Improved |
| use Wasserstein-2 instead of other similarity metrics (Mahalanobis, JS, Euclidean, Cosine) (ablation) | VAR-siamese · Quora dev | accuracy | Improved |
Closest papers by meaning
- Supervised Word Mover's Distance (NeurIPS 2016 · similarity 0.87)
- Model Comparison for Semantic Grouping (ICML 2019 · similarity 0.86)
- Hierarchical Optimal Transport for Document Representation (NeurIPS 2019 · similarity 0.86)
- Learning to Embed Semantic Similarity for Joint Image-Text Retrieval (TPAMI 2021 · similarity 0.86)
- Siamese Recurrent Architectures for Learning Sentence Similarity (AAAI 2016 · similarity 0.86)
- Narrowing the Gap between Supervised and Unsupervised Sentence Representation Learning with Large Language Model (AAAI 2024 · similarity 0.83)
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.
