CaltechAUTHORS
  A Caltech Library Service

Auditing AI models for Verified Deployment under Semantic Specifications

Bharadhwaj, Homanga and Huang, De-An and Xiao, Chaowei and Anandkumar, Anima and Garg, Animesh (2021) Auditing AI models for Verified Deployment under Semantic Specifications. . (Unpublished) https://resolver.caltech.edu/CaltechAUTHORS:20220714-224715197

[img] PDF - Submitted Version
Creative Commons Attribution.

1MB

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20220714-224715197

Abstract

Auditing trained deep learning (DL) models prior to deployment is vital for preventing unintended consequences. One of the biggest challenges in auditing is the lack of human-interpretable specifications for the DL models that are directly useful to the auditor. We address this challenge through a sequence of semantically-aligned unit tests, where each unit test verifies whether a predefined specification (e.g., accuracy over 95%) is satisfied with respect to controlled and semantically aligned variations in the input space (e.g., in face recognition, the angle relative to the camera). We enable such unit tests through variations in a semantically-interpretable latent space of a generative model. Further, we conduct certified training for the DL model through a shared latent space representation with the generative model. With evaluations on four different datasets, covering images of chest X-rays, human faces, ImageNet classes, and towers, we show how AuditAI allows us to obtain controlled variations for certified training. Thus, our framework, AuditAI, bridges the gap between semantically-aligned formal verification and scalability. A blog post accompanying the paper is at this link this https URL https://developer.nvidia.com/blog/nvidia-research-auditing-ai-models-for-verified-deployment-under-semantic-specifications


Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription
http://arxiv.org/abs/2109.12456arXivDiscussion Paper
https://developer.nvidia.com/blog/nvidia-research-auditing-ai-models-for-verified-deployment-under-semantic-specifications/Related ItemBlog post accompanying paper
ORCID:
AuthorORCID
Anandkumar, Anima0000-0002-6974-6797
Garg, Animesh0000-0003-0482-4296
Additional Information:Attribution 4.0 International (CC BY 4.0). Work done during Homanga’s research internship at NVIDIA. We thank everyone in the AI Algorithms team at NVIDIA Research for helpful discussions throughout the project, and for providing critical feedback during the internal reviews.
Record Number:CaltechAUTHORS:20220714-224715197
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20220714-224715197
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:115612
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:15 Jul 2022 15:09
Last Modified:15 Jul 2022 15:09

Repository Staff Only: item control page