A Scalable Mixture Model Based Defense Against Data Poisoning Attacks on Classifiers

Mathematics

• Classifier 100%
• Mixture Model 94%
• Attack 90%
• Model-based 80%
• Training 67%
• Spam 22%
• Outsourcing 21%
• Active Learning 20%
• Naive Bayes 19%
• Data Acquisition 18%
• Corpus 17%
• Data Distribution 16%
• Support Vector Machine 15%
• Polarity 15%
• Data-driven 14%
• Subset 14%
• Hypothesis Testing 12%
• Outlier 12%
• Neural Networks 11%
• Scenarios 11%
• Skew 11%
• Experiment 9%
• Strategy 8%
• Unknown 8%
• Demonstrate 7%
• Class 3%

Engineering & Materials Science

• Classifiers 62%
• Outsourcing 52%
• Problem-Based Learning 51%
• Data acquisition 37%
• Support vector machines 35%
• Neural networks 26%
• Testing 22%
• Experiments 15%