arXiv：cs.LG· Jaturong Kongmanee, Smile Thanapattheerakul·· 7 小时前AI 评分39
Latent Diagnostic Taxonomy：面向提示注入检测的分类器构建与决策诊断框架
The Latent Diagnostic Taxonomy: A Framework for Constructing Classifiers and Diagnosing Their Decisions, Applied to Prompt Injection Detection
AI 导读
研究者提出 Latent Diagnostic Taxonomy 框架，用于构建安全防护层分类器并诊断其哪些高置信决策可信。该框架通过交叉验证选择嵌入维度、定位约 29% 训练样本的潜在支持向量，并按安全信任、启发式偏差与覆盖、上下文不足三类进行路由。在公开提示注入数据集上，约 77% 的高置信决策对删除单个 token 不稳健，且可分为置信度校准失败与可被利用的捷径两种模式。
正文
Abstract:This paper proposes a framework for constructing a classifier as a safeguard layer, and for developing a complementary diagnostic that identifies which of the classifier's confident decisions can be trusted. This framework, the Latent Diagnostic Taxonomy, consists of (i) constructing a dimensionality-optimized classifier, in which the embedding dimensionality is empirically selected via cross-validated performance rather than fixed a priori, (ii) locating a relatively small set of latent support vectors (~ 29% of total training examples) representing influential prompts for identifying tokens that alter the classifier's predicted labels, and (iii) utilizing such tokens and their associated attack magnitudes for constructing a diagnostic taxonomy. This diagnostic taxonomy provides an end-to-end guideline for flagging prompts that require different treatments: rely Safely on the classifier's decision; flag Heuristic Bias and Heuristic Override cases; route Insufficient Context cases for further human/safety review. Applying the framework to a classifier trained on a public prompt injection dataset, we find that a substantial fraction of its confident decisions (~ 77%) are not robust to removing a single token, and that this brittleness separates into two distinct failure patterns: a confidence calibration failure and a genuinely exploitable shortcut. For each zone of the taxonomy, we also recommend strategies for remediating diagnosed prompts. We illustrate the framework as a series of steps, demonstrating how each step operates.
| Comments: | 10 pages, 5 figures |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR) |
| Cite as: | arXiv:2608.26423 [cs.LG] |
| (or arXiv:2608.26423v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2608.26423 arXiv-issued DOI via DataCite |
Submission history
From: Jaturong Kongmanee [view email]
[v1]
Wed, 26 Aug 2026 21:55:15 UTC (233 KB)
[v2]
Tue, 6 Oct 2026 15:27:36 UTC (235 KB)
来源：arXiv：cs.LG · arxiv.org