Research map / Probabilistic and scientific modeling / Molecules and proteins
CoSE: Compositional Stroke Embeddings
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| replace seq2seq autoencoder with CoSE transformer autoencoder (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0540 -> 0.0442 | Helped |
| replace RNN relational model with transformer CoSE relational model (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0713 -> 0.0442 | Helped |
| add positional encoding (ordered) to relational model (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0442 -> 0.0696 | Hurt |
| condition on start position in embedding prediction (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0790 -> 0.0442 | Helped |
| increase embedding dimension from 8 to 32 (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0442 -> 0.0511 | Hurt |
| add VAE regularizer (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0442 -> 0.0953 | Hurt |
| backpropagate relational model gradients into stroke encoder (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0442 -> 0.0470 | Hurt |
| use multi-component GMM (10+) for embedding prediction (ablation) | CoSE · DiDi stroke prediction | Prediction Chamfer distance | Helped |
| CoSE versus Sketch-RNN decoder | CoSE · DiDi stroke prediction | Prediction Chamfer distance: 0.0679 -> 0.0442 | Helped |
| make stroke embedding autoregressive (seq2seq-AR) (ablation) | seq2seq · DiDi stroke prediction | Prediction Chamfer distance: 0.0540 -> 0.0855 | Hurt |
Closest papers by meaning
- Modelling complex vector drawings with stroke-clouds (ICLR 2024 · similarity 0.89)
- SketchINR: A First Look into Sketches as Implicit Neural Representations (CVPR 2024 · similarity 0.85)
- Neural 3D Strokes: Creating Stylized 3D Scenes with Vectorized 3D Strokes (CVPR 2024 · similarity 0.85)
- DrawingsDreamer: A Unified Multi-View Engineering Drawings Generation Model (NeurIPS 2026 · similarity 0.84)
- SketchAgent: Language-Driven Sequential Sketch Generation (CVPR 2025 · similarity 0.84)
- Latent Representation Matters: Human-like Sketches in One-shot Drawing Tasks (NeurIPS 2024 · similarity 0.84)
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.