COMPASS Builds Interpretable Concept Bottleneck Transformer Using 132 Gene Signatures
marinkazitnik · x · 2026-07-03
At its core, COMPASS features a concept bottleneck design: after encoding the tumor transcriptome, it processes the data through a Transformer bottleneck layer comprising 132 gene signatures (aggregated into 44 interpretable immune and stromal concepts) to ultimately predict treatment response. By basing decisions on human-readable biological concepts rather than latent features, the model ensures strong clinical interpretability.
More from Research
- HarnessMM cuts agent tokens per task by up to 88% while raising success rates — omarsar0 · 2026-10-26
- Ben Recht: The world needs a truly open language model, and it's easier to build than you think — beenwrekt · 2026-10-26
- New Blog Breaks Black-Box Distillation Into Five Practical Steps — chrisrohlf · 2026-10-26
- DetectorScope turns black-box AI text detectors into auditable ones with sparse autoencoders — _akhaliq · 2026-10-26
- OPST paper shows LLMs self-improve with zero human data, guided only by topic descriptions — _akhaliq · 2026-10-26
- Tencent's HiSTEVE 1.0: hierarchical VLA agent plays Minecraft with native keyboard and mouse, hits 85% SOTA — _akhaliq · 2026-10-26