Semantic Anomaly

Semantic anomaly detection focuses on identifying deviations in meaning or structure within data, going beyond simple statistical outliers. Current research emphasizes the use of large language models (LLMs) and vision transformers, leveraging their contextual understanding and ability to generate interpretable anomaly maps, for applications ranging from business process monitoring to autonomous driving safety. This field is crucial for improving the robustness and reliability of AI systems across diverse domains, offering insights into both system vulnerabilities and the underlying cognitive processes involved in anomaly recognition. The development of generalizable anomaly detection models, capable of handling diverse data types and anomaly types without extensive retraining, is a key area of ongoing investigation.

Papers

July 17, 2024

GeneralAD: Anomaly Detection Across Domains by Attending to Distorted Features
Luc P. J. Sträter, Mohammadreza Salehi, Efstratios Gavves, Cees G. M. Snoek, Yuki M. Asano
Anomaly Detection Self Supervised Anomaly Anomaly Detection Framework Distortion Realism Semantic Anomaly Temporal Anomaly

June 22, 2024

DABL: Detecting Semantic Anomalies in Business Processes Using Large Language Models
Wei Guan, Jian Cao, Jianqi Gao, Haiyan Zhao, Shiyou Qian
Business Process Detecting Anomaly Semantic Anomaly Real World Process Semantic Anomaly Detection

June 5, 2024

Too Big to Fail: Larger Language Models are Disproportionately Resilient to Induction of Dementia-Related Linguistic Anomalies
Changye Li, Zhecheng Sheng, Trevor Cohen, Serguei Pakhomov
Attention Mechanism Attention Layer Larger Language Model Dementia Related Linguistic Anomaly Automatic Induction Semantic Anomaly

March 11, 2024

Toward Generalist Anomaly Detection via In-context Residual Learning with Few-shot Sample Prompts
Jiawen Zhu, Guansong Pang
Anomaly Detection Semantic Anomaly

December 5, 2023

Weakly Supervised Detection of Hallucinations in LLM Activations
Miriam Rateike, Celia Cintas, John Wamburu, Tanya Akumu, Skyler Speakman
Large Language Model Pre Trained Model Weakly Supervised Model Hallucination Weakly Supervised Anomaly Detection Semantic Anomaly

October 2, 2023

Naming Practices of Pre-Trained Models in Hugging Face
Wenxin Jiang, Chingwo Cheung, Mingyu Kim, Heesoo Kim, George K. Thiruvathukal, James C. Davis
Pre Trained Model Human Face DNN Architecture Model Reuse Semantic Anomaly Model Discovery Naming Variation

May 18, 2023

Semantic Anomaly Detection with Large Language Models
Amine Elhafsi, Rohan Sinha, Christopher Agia, Edward Schmerling, Issa Nesnas, Marco Pavone
Autonomous Driving Semantic Anomaly Semantic Anomaly Detection Semantic Reasoning

April 12, 2022

Self-Supervised Losses for One-Class Textual Anomaly Detection
Kimberly T. Mai, Toby Davies, Lewis D. Griffin
Anomaly Detection Anomaly Score Self Supervised Objective Self Supervised Loss Semantic Anomaly

March 25, 2022

GPT-D: Inducing Dementia-related Linguistic Anomalies by Deliberate Degradation of Artificial Neural Language Models
Changye Li, David Knopman, Weizhe Xu, Trevor Cohen, Serguei Pakhomov
Deep Learning GPT Neo Transformer Based Deep Human Speech Dementia Related Linguistic Anomaly Document Degradation Semantic Anomaly Neural Network Language Model

November 15, 2021

Triggerless Backdoor Attack for NLP Tasks with Clean Labels
Leilei Gan, Jiwei Li, Tianwei Zhang, Xiaoya Li, Yuxian Meng, Fei Wu, Yi Yang, Shangwei Guo, Chun Fan
Backdoor Attack NLP Field NLP Task NLP Model Textual Backdoor Attack Multimodal Backdoor Semantic Anomaly Clean Label

November 12, 2021

Variation and generality in encoding of syntactic anomaly information in sentence embeddings
Qinxuan Wu, Allyson Ettinger
Sentence Embeddings Data Encoding Category Wise Variation Anomaly Type Semantic Anomaly Syntactic Pattern Anomaly Representation