Proof-of-Perception: Certified Tool-Using Multimodal Reasoning with Compositional Conformal Guarantees

Arya Fayyazi; Haleh Akrami

CVPR 20260 citations

Proof-of-Perception: Certified Tool-Using Multimodal Reasoning with Compositional Conformal Guarantees

Abstract

We present Proof-of-Perception (PoP), a tool-using framework that casts multimodal reasoning as an executable graph with explicit reliability guarantees. Each perception or logic node outputs a conformal set \Gamma^ (t) _\delta(x), yielding calibrated, stepwise uncertainty; a lightweight controller uses these certificates to allocate compute under a budget--expanding with extra tool calls only when needed and stopping early otherwise. This grounds answers in verifiable evidence, reduces error compounding and hallucinations, and enables principled accuracy-compute trade-offs. Across document, chart, and multi-image QA benchmarks, PoP improves performance and reliability over strong chain-of-thought, ReAct-style, and program-of-thought baselines while using computation more efficiently. Code is available at \href https://github.com/AryaFayyazi/PoP https://github.com/AryaFayyazi/PoP .

BibTeX

@inproceedings{cvpr2026_proofofperceptio,
  title = {Proof-of-Perception: Certified Tool-Using Multimodal Reasoning with Compositional Conformal Guarantees},
  author = {Arya Fayyazi and Haleh Akrami},
  booktitle = {CVPR 2026},
  year = {2026}
}