Research map / Probabilistic and scientific modeling / Variational and generative sampling
Sequence to Better Sequence: Continuous Revision of Combinatorial Structures
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 |
|---|---|---|---|
| Add VAE probabilistic latent model with constrained gradient-based latent optimization (REVISE) instead of deterministic autoencoder (ablation) | sequence revision via latent continuous space · Simulation study (grammar sequences) | ∆Y(x*): 0.51 vs 0.05 | Helped |
| Add VAE probabilistic latent model with constrained optimization instead of deterministic autoencoder (ablation) | beer review sentence revision · Beer reviews sentiment | ∆Y(x*): 0.52 vs 0.22 | Helped |
| Use more conservative constraint (log α=-1) instead of log α=-10000 (ablation) | REVISE · Simulation study (grammar sequences) | ∆Y(x*): 0.20 vs 0.51 | Hurt |
| Use more conservative constraint (log α=-1) instead of log α=-10000 (ablation) | REVISE · Beer reviews sentiment | ∆Y(x*): 0.31 vs 0.52 | Hurt |
| Adaptive decoder biased toward x0 instead of standard decoding (ablation) | REVISE · Simulation study (grammar sequences) | ∆Y(x*): 0.47 vs 0.51 | No clear effect |
| Adaptive decoder biased toward x0 instead of standard decoding (ablation) | REVISE · Beer reviews sentiment | ∆Y(x*): 0.52 vs 0.52 | No clear effect |
| Gradient-based latent revision (REVISE) instead of combinatorial edit search | sentence revision · Beer reviews sentiment | ∆Y(x*): 0.52 vs 0.19 | Helped |
| Gradient-based latent revision (REVISE) instead of combinatorial edit search | grammar sequence revision · Simulation study (grammar sequences) | ∆Y(x*): 0.51 vs 0.45 | Helped |
Closest papers by meaning
- Sequence-to-Sequence Learning with Latent Neural Grammars (NeurIPS 2021 · similarity 0.87)
- Learning a Generative Model for Validity in Complex Discrete Structures (ICLR 2018 · similarity 0.86)
- Generating Sequences by Learning to Self-Correct (ICLR 2023 · similarity 0.86)
- Learning to Recombine and Resample Data For Compositional Generalization (ICLR 2021 · similarity 0.86)
- Incremental Sequence Classification with Temporal Consistency (NeurIPS 2025 · similarity 0.86)
- Structured Reordering for Modeling Latent Alignments in Sequence Transduction (NeurIPS 2021 · similarity 0.86)
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.