DOE CODE   /    Search Results   /    Topological Signatures of Adversaries in Multimodal Alignments

Topological Signatures of Adversaries in Multimodal Alignments

RESOURCE

Publicly Accessible Repository
https://github.com/lanl/tda-adversarial-detection
https://doi.org/10.11578/dc.20251212.4

SAVE / SHARE



LANL Badge

Abstract

Topological Data Analysis for Adversarial Detection (LANL O4937) - Detects adversarial examples in vision-language models using persistent homology and two-sample testing. Combines TDA features from CLIP embeddings with statistical methods (ME, SCF, SAMMD, C2ST) for robust detection across ImageNet, CIFAR-10/100.
Developers:
Release Date:
2025-12-01
Project Type:
Open Source, Publicly Available Repository
Software Type:
Scientific
Licenses:
BSD 3-clause "New" or "Revised" License
Sponsoring Org.:
Code ID:
171703
Site Accession Number:
O4937
Research Org.:
Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
Country of Origin:
United States

RESOURCE

Publicly Accessible Repository
https://github.com/lanl/tda-adversarial-detection
https://doi.org/10.11578/dc.20251212.4

SAVE / SHARE



LANL Badge

Citation Formats

Bhattarai, Manish, Vu, Minh, and Zollicoffer, Geigh. Topological Signatures of Adversaries in Multimodal Alignments. Computer Software. https://github.com/lanl/tda-adversarial-detection. USDOE Laboratory Directed Research and Development (LDRD) Program. 01 Dec. 2025. Web. doi:10.11578/dc.20251212.4.
Bhattarai, Manish, Vu, Minh, & Zollicoffer, Geigh. (2025, December 01). Topological Signatures of Adversaries in Multimodal Alignments. [Computer software]. https://github.com/lanl/tda-adversarial-detection. https://doi.org/10.11578/dc.20251212.4.
Bhattarai, Manish, Vu, Minh, and Zollicoffer, Geigh. "Topological Signatures of Adversaries in Multimodal Alignments." Computer software. December 01, 2025. https://github.com/lanl/tda-adversarial-detection. https://doi.org/10.11578/dc.20251212.4.
@misc{ doecode_171703,
title = {Topological Signatures of Adversaries in Multimodal Alignments},
author = {Bhattarai, Manish and Vu, Minh and Zollicoffer, Geigh},
abstractNote = {Topological Data Analysis for Adversarial Detection (LANL O4937) - Detects adversarial examples in vision-language models using persistent homology and two-sample testing. Combines TDA features from CLIP embeddings with statistical methods (ME, SCF, SAMMD, C2ST) for robust detection across ImageNet, CIFAR-10/100.},
doi = {10.11578/dc.20251212.4},
url = {https://doi.org/10.11578/dc.20251212.4},
howpublished = {[Computer Software] \url{https://doi.org/10.11578/dc.20251212.4}},
year = {2025},
month = {dec}
}