When not to classify: Anomaly detection of attacks (ADA) on DNN classifiers at test time

Arts & Humanities

• Classifier 99%
• Anomaly 82%
• Attack 80%
• Evasion 16%
• Layer 15%
• Receiver Operating Characteristic 11%
• Scenarios 10%
• Image Database 10%
• Performance Measures 9%
• Null Hypothesis 8%
• Cache 8%
• Machine Learning 6%
• Recipient 6%
• Defending 6%
• Divergence 5%
• Statistics 5%
• Confusion 5%
• Threat 4%
• Uncertainty 3%

Medicine & Life Sciences

• Information Services 33%
• Machine Learning 31%
• Uncertainty 25%
• ROC Curve 23%
• Area Under Curve 22%
• Statistics 21%
• Learning 20%
• Joints 19%
• Weights and Measures 17%
• Databases 16%