Research map / Image generation / Fast diffusion sampling
Multi-Objective Protein Design via Memory-Aware Test-Time Scaling in Diffusion Models
Ming Yang, Xin Zheng, Yi Li, YIZHEN ZHENG, Huan Yee Koh, Yanqing Guo, Xiaofeng Cao, Shirui Pan
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 reasoning-aware memory bank (ablation) | MOMST · Dual-objective: Globularity & pLDDT | pLDDT P50: 0.3812 -> 0.7176 | Improved |
| add reasoning-aware memory bank (ablation) | MOMST · Triple-objective: Hydrophobicity & surface exposure & pLDDT | hydrophobicity P50: -0.1961 -> -0.3808 | Hurt |
| add reasoning-aware memory bank (ablation) | MOMST · Triple-objective: Hydrophobicity & surface exposure & pLDDT | surface exposure P95: 0.7377 -> 0.8764 | Improved |
| add self-contrastive learning (ablation) | MOMST · Dual-objective: Globularity & pLDDT | globularity P50: -3.3601 -> -1.7356 | Improved |
| add self-contrastive learning (ablation) | MOMST · Dual-objective: Globularity & pLDDT | pLDDT P50: 0.8710 -> 0.7176 | Hurt |
| add inference-time Pareto alignment (ablation) | MOMST · Dual-objective: Globularity & pLDDT | pLDDT P50: 0.4824 -> 0.7176 | Improved |
| add inference-time Pareto alignment (ablation) | MOMST · Dual-objective: Globularity & pLDDT | globularity P50: -2.8370 -> -1.7356 | Improved |
| increase guidance scale lambda from 0.2 to 0.5 or more (ablation) | MOMST · Multi-objective design | functional and structural scores: 0.2 -> 0.5 | Hurt |
| use 50:50 positive/negative split of memory bank instead of 95:5 (ablation) | MOMST · Multi-objective design | reward and pLDDT | Improved |
| reduce editing budget T/L from 0.1 to 0.04 (ablation) | MOMST · Dual-objective: Hydrophobicity & pLDDT | pLDDT P50: 0.9657 -> 0.8730 | Hurt |
| relax mutation budget T/L from 0.1 to 0.2 (ablation) | MOMST · Triple-objective diversity | Dmax: 0.1667 -> 0.3133 | Improved |
| increase memory bank size from 400 to 1200 (ablation) | MOMST · Dual-objective: Globularity & pLDDT | globularity and pLDDT scores: 400 -> 1200 | Improved |
Closest papers by meaning
- Reward-Guided Iterative Refinement in Diffusion Models at Test-Time with Applications to Protein and DNA Design (ICML 2025 · similarity 0.87)
- ProteinOPD: Towards Effective and Efficient Preference Alignment for Protein Design (NeurIPS 2026 · similarity 0.86)
- Multi-Objective Molecular Design Through Learning Latent Pareto Set (AAAI 2025 · similarity 0.86)
- ProSpero: Active Learning for Robust Protein Design Beyond Wild-Type Neighborhoods (NeurIPS 2025 · similarity 0.86)
- Diffusion on Language Model Encodings for Protein Sequence Generation (ICML 2025 · similarity 0.85)
- Evolution-Inspired Loss Functions for Protein Representation Learning (ICML 2024 · similarity 0.85)
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.
