AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Intervention-Aware Multiscale Representation Learning from Imaging Phenomics and Perturbation Transcriptomics

Jiayuan Chen, Ruoqi Liu, Zishan Gu, Ping Zhang

CVPR 2026 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervised neuroimaging representations · 0 citations (OpenAlex)

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
add transcriptome-guided intervention-aware distillation (TIDE) on top of DINODINO ViT · CPG-12 one-shot unseen compoundsTop-1 accuracy (%): 36.36 -> 42.09Improved
add TIDE to DINODINO ViT-Small · CPG-Pilot one-shot unseen genetic perturbationsTop-1 accuracy (%): 11.36 -> 16.01Improved
add TIDE to DINODINO ViT-Small · RxRx3 drug-target discoveryAP: 0.300 -> 0.317Improved
add TIDE to DINO with CA-ViTDINO CA-ViT · RxRx3 drug-target discoveryAUC: 0.545 -> 0.612Improved
add drug-molecule contrastive alignment CL(D) versus DINODINO · one-shot transfer (CPG-12)Top-1 accuracy (%): 36.36 -> 40.04Improved
CL(D) alignment instead of DINO for drug-target discoveryDINO · RxRx3 drug-target discovery (ViT)AP: 0.300 -> 0.252No clear effect
use DINO instead of MAE as SSL baselineMAE · CPG-12 one-shot unseen compoundsTop-1 accuracy (%): 4.68 -> 36.36Improved
CA-ViT backbone instead of ViT with TIDETIDE · CPG-Pilot one-shot unseen genetic perturbationsTop-1 accuracy (%): 16.01 -> 17.80Improved
add chemistry-aware codebook (FCFP, learnable) versus direct drug-identity classification (ablation)TIDE · CPG-12 one-shot unseen compoundsTop-1 accuracy (%): 39.39 -> 42.09Improved
learn the codebook end-to-end instead of freezing FCFP prototypes (ablation)TIDE · CPG-12 one-shot unseen compoundsTop-1 accuracy (%): 40.13 -> 42.09Improved
freeze FCFP codebook versus no codebook on RxRx3 (ablation)TIDE · RxRx3 one-shot unseen genetic perturbationsTop-1 accuracy (%): 5.20 -> 5.11Hurt
choose FCFP versus GIN molecular representation for the codebook (ablation)TIDE · CPG-12 one-shot unseen compoundsTop-5 accuracy (%): 69.03 -> 71.07No clear effect

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