AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

FIDIA: Function-Informed Sequence Design via Inference-Aligned Policy Optimization

Minghan Li, Fengji Li, Yilin Tao, Yue Deng

ICML 2026 · Reinforcement learning › RL from feedback · approach: RL fine-tuning of generative models

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
replace expectation objective with Best-of-N rank-based policy gradient (FIDIA) on ProteinMPNNProteinMPNN · Motif scaffolding FrameFSuccess rate (%): 26.67 -> 37.50Improved
replace expectation objective with Best-of-N rank-based policy gradient (FIDIA) on ProteinMPNNProteinMPNN · Motif scaffolding RFDSuccess rate (%): 55.00 -> 65.83Improved
use FIDIA instead of GRPO with standard expected-return objectiveProteinMPNN · Motif scaffolding FrameFSuccess rate (%): 33.33 -> 37.50Improved
BoN baseline trained on top-k candidates instead of top-1ProteinMPNN · Motif scaffolding FrameFSuccess rate (%): 34.17 -> 28.33Hurt
apply FIDIA instead of MPNN: diversity of sampled sequencesProteinMPNN · Motif scaffolding RFDDiversity: 0.31 -> 0.43Improved
cross-validate with AlphaFold2 instead of ESMFold oracleFIDIA · Motif scaffolding FrameFSuccess rate (%): 37.50 -> 36.67No clear effect
apply FIDIA to epitope scaffolding vs MPNNProteinMPNN · Epitope scaffolding (vaccine)Success rate (%): 25.45 -> 27.09Improved
fine-tune LigandMPNN with FIDIA using Vina rewardLigandMPNN · Affinity-enhancing enzyme designVina score (kcal/mol): -6.53 -> -7.21Improved
fine-tune LigandMPNN with FIDIA using Vina rewardLigandMPNN · Affinity-enhancing enzyme designSuccess rate (%): 90.74 -> 83.33Hurt
increase structure sampling budget M to 50 (ablation)FIDIA · Motif scaffolding (pipeline)Solved motifs: 21 targets at M=50Improved

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