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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fayyazi, Arya, Akrami, Haleh
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917300544733184
author Fayyazi, Arya
Akrami, Haleh
author_facet Fayyazi, Arya
Akrami, Haleh
contents 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, 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.
format Preprint
id arxiv_https___arxiv_org_abs_2603_00324
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Proof-of-Perception: Certified Tool-Using Multimodal Reasoning with Compositional Conformal Guarantees
Fayyazi, Arya
Akrami, Haleh
Computer Vision and Pattern Recognition
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, 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.
title Proof-of-Perception: Certified Tool-Using Multimodal Reasoning with Compositional Conformal Guarantees
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.00324