munhouiani / trojai-literature

TrojAI Literature Review

The list below contains curated papers and arXiv articles that are related to Trojan attacks, backdoor attacks, and data poisoning on neural networks and machine learning systems. They are ordered "approximately" from most to least recent and articles denoted with a "*" mention the TrojAI program directly. Some of the particularly relevant papers include a summary that can be accessed by clicking the "Summary" drop down icon underneath the paper link. These articles were identified using variety of methods including:

• A flair embedding created from the arXiv CS subset; details will be provided later.
• A trained ASReview random forest model
• A curated manual literature review

1. A Feature Based On-Line Detector to Remove Adversarial-Backdoors by Iterative Demarcation

2. BlindNet backdoor: Attack on deep neural network using blind watermark

3. DBIA: Data-free Backdoor Injection Attack against Transformer Networks

4. Backdoor Attack through Frequency Domain

5. NTD: Non-Transferability Enabled Backdoor Detection

6. Romoa: Robust Model Aggregation for the Resistance of Federated Learning to Model Poisoning Attacks

7. Generative strategy based backdoor attacks to 3D point clouds: Work in Progress

8. Deep Neural Backdoor in Semi-Supervised Learning: Threats and Countermeasures

9. FooBaR: Fault Fooling Backdoor Attack on Neural Network Training

10. BFClass: A Backdoor-free Text Classification Framework

11. Backdoor Attacks on Federated Learning with Lottery Ticket Hypothesis

12. Data Poisoning against Differentially-Private Learners: Attacks and Defenses

13. DOES DIFFERENTIAL PRIVACY DEFEAT DATA POISONING?

14. Check Your Other Door! Establishing Backdoor Attacks in the Frequency Domain

15. HaS-Nets: A Heal and Select Mechanism to Defend DNNs Against Backdoor Attacks for Data Collection Scenarios

16. SanitAIs: Unsupervised Data Augmentation to Sanitize Trojaned Neural Networks

17. COVID-19 Diagnosis from Chest X-Ray Images Using Convolutional Neural Networks and Effects of Data Poisoning

18. Interpretability-Guided Defense against Backdoor Attacks to Deep Neural Networks

19. Trojan Signatures in DNN Weights

20. HOW TO INJECT BACKDOORS WITH BETTER CONSISTENCY: LOGIT ANCHORING ON CLEAN DATA

21. A Synergetic Attack against Neural Network Classifiers combining Backdoor and Adversarial Examples

22. Backdoor Attack and Defense for Deep Regression

23. Use Procedural Noise to Achieve Backdoor Attack

24. Excess Capacity and Backdoor Poisoning

25. BatFL: Backdoor Detection on Federated Learning in e-Health

26. Poisonous Label Attack: Black-Box Data Poisoning Attack with Enhanced Conditional DCGAN

27. Backdoor Attacks on Network Certification via Data Poisoning

28. Identifying Physically Realizable Triggers for Backdoored Face Recognition Networks

29. Simtrojan: Stealthy Backdoor Attack

30. Back to the Drawing Board: A Critical Evaluation of Poisoning Attacks on Federated Learning

31. Quantization Backdoors to Deep Learning Models

32. Multi-Target Invisibly Trojaned Networks for Visual Recognition and Detection

33. A Countermeasure Method Using Poisonous Data Against Poisoning Attacks on IoT Machine Learning

34. FederatedReverse: A Detection and Defense Method Against Backdoor Attacks in Federated Learning

35. Accumulative Poisoning Attacks on Real-time Data

36. Inaudible Manipulation of Voice-Enabled Devices Through BackDoor Using Robust Adversarial Audio Attacks

37. Stealthy Targeted Data Poisoning Attack on Knowledge Graphs

38. BinarizedAttack: Structural Poisoning Attacks to Graph-based Anomaly Detection

39. On the Effectiveness of Poisoning against Unsupervised Domain Adaptation

40. Simple, Attack-Agnostic Defense Against Targeted Training Set Attacks Using Cosine Similarity

41. Data Poisoning Attacks Against Outcome Interpretations of Predictive Models

42. BDDR: An Effective Defense Against Textual Backdoor Attacks

43. Poisoning attacks and countermeasures in intelligent networks: status quo and prospects

44. The Devil is in the GAN: Defending Deep Generative Models Against Backdoor Attacks

45. BadEncoder: Backdoor Attacks to Pre-trainedEncoders in Self-Supervised Learning

46. BadEncoder: Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning

47. Can You Hear It? Backdoor Attacks via Ultrasonic Triggers

48. Poisoning Attacks via Generative Adversarial Text to Image Synthesis

49. Ant Hole: Data Poisoning Attack Breaking out the Boundary of Face Cluster

50. Poison Ink: Robust and Invisible Backdoor Attack

51. MT-MTD: Muti-Training based Moving Target Defense Trojaning Attack in Edged-AI network

52. Text Backdoor Detection Using An Interpretable RNN Abstract Model

53. Garbage in, Garbage out: Poisoning Attacks Disguised with Plausible Mobility in Data Aggregation

54. Classification Auto-Encoder based Detector against Diverse Data Poisoning Attacks

55. Poisoning Knowledge Graph Embeddings via Relation Inference Patterns

56. Adversarial Training Time Attack Against Discriminative and Generative Convolutional Models

57. Poisoning of Online Learning Filters: DDoS Attacks and Countermeasures

58. Rethinking Stealthiness of Backdoor Attack against NLP Models

59. Robust Learning for Data Poisoning Attacks

60. SPECTRE: Defending Against Backdoor Attacks Using Robust Statistics

61. Poisoning the Search Space in Neural Architecture Search

62. Data Poisoning Won’t Save You From Facial Recognition

63. Accumulative Poisoning Attacks on Real-time Data

64. Backdoor Attack on Machine Learning Based Android Malware Detectors

65. Understanding the Limits of Unsupervised Domain Adaptation via Data Poisoning

66. Indirect Invisible Poisoning Attacks on Domain Adaptation

67. Fight Fire with Fire: Towards Robust Recommender Systems via Adversarial Poisoning Training

68. Putting words into the system’s mouth: A targeted attack on neural machine translation using monolingual data poisoning

69. SUBNET REPLACEMENT: DEPLOYMENT-STAGE BACKDOOR ATTACK AGAINST DEEP NEURAL NETWORKS IN GRAY-BOX SETTING

70. Spinning Sequence-to-Sequence Models with Meta-Backdoors

71. Sleeper Agent: Scalable Hidden Trigger Backdoors for Neural Networks Trained from Scratch

72. Poisoning and Backdooring Contrastive Learning

73. AdvDoor: Adversarial Backdoor Attack of Deep Learning System

74. Defending against Backdoor Attacks in Natural Language Generation

75. De-Pois: An Attack-Agnostic Defense against Data Poisoning Attacks

76. Poisoning MorphNet for Clean-Label Backdoor Attack to Point Clouds

77. Provable Guarantees against Data Poisoning Using Self-Expansion and Compatibility

78. MLDS: A Dataset for Weight-Space Analysis of Neural Networks

79. Poisoning the Unlabeled Dataset of Semi-Supervised Learning

80. Regularization Can Help Mitigate Poisioning Attacks. . . With The Right Hyperparameters

81. Witches' Brew: Industrial Scale Data Poisoning via Gradient Matching

82. Towards Robustness Against Natural Language Word Substitutions

83. Concealed Data Poisoning Attacks on NLP Models

84. Covert Channel Attack to Federated Learning Systems

85. Backdoor Attacks Against Deep Learning Systems in the Physical World

86. Backdoor Attacks on Self-Supervised Learning

87. Transferable Environment Poisoning: Training-time Attack on Reinforcement Learning

88. Investigation of a differential cryptanalysis inspired approach for Trojan AI detection

89. Explanation-Guided Backdoor Poisoning Attacks Against Malware Classifiers

90. Robust Backdoor Attacks against Deep Neural Networks in Real Physical World

91. The Design and Development of a Game to Study Backdoor Poisoning Attacks: The Backdoor Game

92. A Backdoor Attack against 3D Point Cloud Classifiers

93. Explainability-based Backdoor Attacks Against Graph Neural Networks

94. DeepSweep: An Evaluation Framework for Mitigating DNN Backdoor Attacks using Data Augmentation

95. Rethinking the Backdoor Attacks' Triggers: A Frequency Perspective

96. PointBA: Towards Backdoor Attacks in 3D Point Cloud

97. Online Defense of Trojaned Models using Misattributions

98. Be Careful about Poisoned Word Embeddings: Exploring the Vulnerability of the Embedding Layers in NLP Models

99. SPECTRE: Defending Against Backdoor Attacks Using Robust Covariance Estimation

100. Black-box Detection of Backdoor Attacks with Limited Information and Data

101. TOP: Backdoor Detection in Neural Networks via Transferability of Perturbation

102. T-Miner : A Generative Approach to Defend Against Trojan Attacks on DNN-based Text Classification

103. Hidden Backdoor Attack against Semantic Segmentation Models

104. What Doesn't Kill You Makes You Robust(er): Adversarial Training against Poisons and Backdoors

105. Red Alarm for Pre-trained Models: Universal Vulnerabilities by Neuron-Level Backdoor Attacks

106. Provable Defense Against Delusive Poisoning

107. An Approach for Poisoning Attacks Against RNN-Based Cyber Anomaly Detection

108. Backdoor Scanning for Deep Neural Networks through K-Arm Optimization

109. TAD: Trigger Approximation based Black-box Trojan Detection for AI*

110. WaNet - Imperceptible Warping-based Backdoor Attack

111. Data Poisoning Attack on Deep Neural Network and Some Defense Methods

112. Baseline Pruning-Based Approach to Trojan Detection in Neural Networks*

113. Covert Model Poisoning Against Federated Learning: Algorithm Design and Optimization

114. Property Inference from Poisoning

115. TROJANZOO: Everything you ever wanted to know about neural backdoors (but were afraid to ask)

116. A Master Key Backdoor for Universal Impersonation Attack against DNN-based Face Verification

117. Detecting Universal Trigger's Adversarial Attack with Honeypot

118. ONION: A Simple and Effective Defense Against Textual Backdoor Attacks

119. Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks

120. Data Poisoning Attacks to Deep Learning Based Recommender Systems

121. Backdoors hidden in facial features: a novel invisible backdoor attack against face recognition systems

122. One-to-N & N-to-One: Two Advanced Backdoor Attacks against Deep Learning Models

123. DeepPoison: Feature Transfer Based Stealthy Poisoning Attack

124. Policy Teaching via Environment Poisoning:Training-time Adversarial Attacks against Reinforcement Learning

125. Composite Backdoor Attack for Deep Neural Network by Mixing Existing Benign Features

126. SPA: Stealthy Poisoning Attack

127. Backdoor Attack with Sample-Specific Triggers

128. Explainability Matters: Backdoor Attacks on Medical Imaging

129. Escaping Backdoor Attack Detection of Deep Learning

130. Just How Toxic is Data Poisoning? A Unified Benchmark for Backdoor and Data Poisoning Attacks

131. Poisoning Attacks on Cyber Attack Detectors for Industrial Control Systems

132. Fair Detection of Poisoning Attacks in Federated Learning

133. Deep Feature Space Trojan Attack of Neural Networks by Controlled Detoxification*

134. Stealthy Poisoning Attack on Certified Robustness

135. Machine Learning with Electronic Health Records is vulnerable to Backdoor Trigger Attacks

136. Data Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses

137. Detection of Backdoors in Trained Classifiers Without Access to the Training Set

138. TROJANZOO: Everything you ever wanted to know about neural backdoors(but were afraid to ask)

139. HaS-Nets: A Heal and Select Mechanism to Defend DNNs Against Backdoor Attacks for Data Collection Scenarios

140. DeepSweep: An Evaluation Framework for Mitigating DNN Backdoor Attacks using Data Augmentation

141. Poison Attacks against Text Datasets with Conditional Adversarially Regularized Autoencoder

142. Strong Data Augmentation Sanitizes Poisoning and Backdoor Attacks Without an Accuracy Tradeoff

143. BaFFLe: Backdoor detection via Feedback-based Federated Learning

144. Detecting Backdoors in Neural Networks Using Novel Feature-Based Anomaly Detection

145. Mitigating Backdoor Attacks in Federated Learning

146. FaceHack: Triggering backdoored facial recognition systems using facial characteristics

147. Customizing Triggers with Concealed Data Poisoning

148. Backdoor Learning: A Survey

149. Rethinking the Trigger of Backdoor Attack

150. AEGIS: Exposing Backdoors in Robust Machine Learning Models

151. Weight Poisoning Attacks on Pre-trained Models

152. Poisoned classifiers are not only backdoored, they are fundamentally broken

153. Input-Aware Dynamic Backdoor Attack

154. Reverse Engineering Imperceptible Backdoor Attacks on Deep Neural Networks for Detection and Training Set Cleansing

155. BAAAN: Backdoor Attacks Against Autoencoder and GAN-Based Machine Learning Models

156. Don’t Trigger Me! A Triggerless Backdoor Attack Against Deep Neural Networks

157. Toward Robustness and Privacy in Federated Learning: Experimenting with Local and Central Differential Privacy

158. CLEANN: Accelerated Trojan Shield for Embedded Neural Networks

159. Witches’ Brew: Industrial Scale Data Poisoning via Gradient Matching

160. Intrinsic Certified Robustness of Bagging against Data Poisoning Attacks

161. Can Adversarial Weight Perturbations Inject Neural Backdoors?

162. Trojaning Language Models for Fun and Profit

163. Practical Detection of Trojan Neural Networks: Data-Limited and Data-Free Cases

164. Class-Oriented Poisoning Attack

165. Noise-response Analysis for Rapid Detection of Backdoors in Deep Neural Networks

166. Cassandra: Detecting Trojaned Networks from Adversarial Perturbations

167. Backdoor Learning: A Survey

168. Backdoor Attacks and Countermeasures on Deep Learning: A Comprehensive Review

169. Live Trojan Attacks on Deep Neural Networks

170. Odyssey: Creation, Analysis and Detection of Trojan Models

171. Data Poisoning Attacks Against Federated Learning Systems

172. Blind Backdoors in Deep Learning Models

173. Deep Learning Backdoors

174. Attack of the Tails: Yes, You Really Can Backdoor Federated Learning

175. Backdoor Attacks on Facial Recognition in the Physical World

176. Graph Backdoor

177. Backdoor Attacks to Graph Neural Networks

178. You Autocomplete Me: Poisoning Vulnerabilities in Neural Code Completion

179. Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networks

180. Trembling triggers: exploring the sensitivity of backdoors in DNN-based face recognition

181. Just How Toxic is Data Poisoning? A Unified Benchmark for Backdoor and Data Poisoning Attacks

182. Adversarial Machine Learning -- Industry Perspectives

183. ConFoc: Content-Focus Protection Against Trojan Attacks on Neural Networks

184. Model-Targeted Poisoning Attacks: Provable Convergence and Certified Bounds

185. Deep Partition Aggregation: Provable Defense against General Poisoning Attacks

186. The TrojAI Software Framework: An OpenSource tool for Embedding Trojans into Deep Learning Models*

187. Influence Function based Data Poisoning Attacks to Top-N Recommender Systems

188. BadNL: Backdoor Attacks Against NLP Models

     Summary

     • Introduces first example of backdoor attacks against NLP models using Char-level, Word-level, and Sentence-level triggers (these different triggers operate on the level of their descriptor)
       • Word-level trigger picks a word from the target model’s dictionary and uses it as a trigger
       • Char-level trigger uses insertion, deletion or replacement to modify a single character in a chosen word’s location (with respect to the sentence, for instance, at the start of each sentence) as the trigger.
       • Sentence-level trigger changes the grammar of the sentence and use this as the trigger
     • Authors impose an additional constraint that requires inserted triggers to not change the sentiment of text input
     • Proposed backdoor attack achieves 100% backdoor accuracy with only a drop of 0.18%, 1.26%, and 0.19% in the models utility, for the IMDB, Amazon, and Stanford Sentiment Treebank datasets

189. Neural Network Calculator for Designing Trojan Detectors*

190. Dynamic Backdoor Attacks Against Machine Learning Models

191. Vulnerabilities of Connectionist AI Applications: Evaluation and Defence

192. Backdoor Attacks on Federated Meta-Learning

193. Defending Support Vector Machines against Poisoning Attacks: the Hardness and Algorithm

194. Backdoors in Neural Models of Source Code

195. A new measure for overfitting and its implications for backdooring of deep learning

196. An Embarrassingly Simple Approach for Trojan Attack in Deep Neural Networks

197. MetaPoison: Practical General-purpose Clean-label Data Poisoning

198. Backdooring and Poisoning Neural Networks with Image-Scaling Attacks

199. Bullseye Polytope: A Scalable Clean-Label Poisoning Attack with Improved Transferability

200. On the Effectiveness of Mitigating Data Poisoning Attacks with Gradient Shaping

201. A Survey on Neural Trojans

202. STRIP: A Defence Against Trojan Attacks on Deep Neural Networks

     Summary

     • Authors introduce a run-time based trojan detection system called STRIP or STRong Intentional Pertubation which focuses on models in computer vision
     • STRIP works by intentionally perturbing incoming inputs (ie. by image blending) and then measuring entropy to determine whether the model is trojaned or not. Low entropy violates the input-dependance assumption for a clean model and thus indicates corruption
     • Authors validate STRIPs efficacy on MNIST,CIFAR10, and GTSRB acheiveing false acceptance rates of below 1%

203. TrojDRL: Trojan Attacks on Deep Reinforcement Learning Agents

204. Demon in the Variant: Statistical Analysis of DNNs for Robust Backdoor Contamination Detection

205. Regula Sub-rosa: Latent Backdoor Attacks on Deep Neural Networks

206. Februus: Input Purification Defense Against Trojan Attacks on Deep Neural Network Systems

207. TBT: Targeted Neural Network Attack with Bit Trojan

208. Bypassing Backdoor Detection Algorithms in Deep Learning

209. A backdoor attack against LSTM-based text classification systems

210. Invisible Backdoor Attacks Against Deep Neural Networks

211. Detecting AI Trojans Using Meta Neural Analysis

212. Label-Consistent Backdoor Attacks

213. Detection of Backdoors in Trained Classifiers Without Access to the Training Set

214. ABS: Scanning neural networks for back-doors by artificial brain stimulation

215. NeuronInspect: Detecting Backdoors in Neural Networks via Output Explanations

216. Universal Litmus Patterns: Revealing Backdoor Attacks in CNNs

217. Programmable Neural Network Trojan for Pre-Trained Feature Extractor

218. Demon in the Variant: Statistical Analysis of DNNs for Robust Backdoor Contamination Detection

219. TamperNN: Efficient Tampering Detection of Deployed Neural Nets

220. TABOR: A Highly Accurate Approach to Inspecting and Restoring Trojan Backdoors in AI Systems

221. Design of intentional backdoors in sequential models

222. Design and Evaluation of a Multi-Domain Trojan Detection Method on ins Neural Networks

223. Poison as a Cure: Detecting & Neutralizing Variable-Sized Backdoor Attacks in Deep Neural Networks

224. Data Poisoning Attacks on Stochastic Bandits

225. Hidden Trigger Backdoor Attacks

226. Deep Poisoning Functions: Towards Robust Privacy-safe Image Data Sharing

227. A new Backdoor Attack in CNNs by training set corruption without label poisoning

228. Deep k-NN Defense against Clean-label Data Poisoning Attacks

229. Transferable Clean-Label Poisoning Attacks on Deep Neural Nets

230. Revealing Backdoors, Post-Training, in DNN Classifiers via Novel Inference on Optimized Perturbations Inducing Group Misclassification

231. Explaining Vulnerabilities to Adversarial Machine Learning through Visual Analytics

232. Subpopulation Data Poisoning Attacks

233. TensorClog: An imperceptible poisoning attack on deep neural network applications

234. DeepInspect: A black-box trojan detection and mitigation framework for deep neural networks

235. Resilience of Pruned Neural Network Against Poisoning Attack

236. Spectrum Data Poisoning with Adversarial Deep Learning

237. Neural cleanse: Identifying and mitigating backdoor attacks in neural networks

238. SentiNet: Detecting Localized Universal Attacks Against Deep Learning Systems

     Summary

     • Authors develop SentiNet detection framework for locating universal attacks on neural networks
     • SentiNet is ambivalent to the attack vectors and uses model visualization / object detection techniques to extract potential attacks regions from the models input images. The potential attacks regions are identified as being the parts that influence the prediction the most. After extraction, SentiNet applies these regions to benign inputs and uses the original model to analyze the output
     • Authors stress test the SentiNet framework on three different types of attacks— data poisoning attacks, Trojan attacks, and adversarial patches. They are able to show that the framework achieves competitive metrics across all of the attacks (average true positive rate of 96.22% and an average true negative rate of 95.36%)

239. PoTrojan: powerful neural-level trojan designs in deep learning models

240. Hardware Trojan Attacks on Neural Networks

241. Spectral Signatures in Backdoor Attacks

     Summary

     • Identified a "spectral signatures" property of current backdoor attacks which allows the authors to use robust statistics to stop Trojan attacks
     • The "spectral signature" refers to a change in the covariance spectrum of learned feature representations that is left after a network is attacked. This can be detected by using singular value decomposition (SVD). SVD is used to identify which examples to remove from the training set. After these examples are removed the model is retrained on the cleaned dataset and is no longer Trojaned. The authors test this method on the CIFAR 10 image dataset.

242. Defending Neural Backdoors via Generative Distribution Modeling

243. Detecting Backdoor Attacks on Deep Neural Networks by Activation Clustering

     Summary

     • Proposes Activation Clustering approach to backdoor detection/ removal which analyzes the neural network activations for anomalies and works for both text and images
     • Activation Clustering uses dimensionality techniques (ICA, PCA) on the activations and then clusters them using k-means (k=2) along with a silhouette score metric to separate poisoned from clean clusters
     • Shows that Activation Clustering is successful on three different image/datasets (MNIST, LISA, Rotten Tomatoes) as well as in settings where multiple Trojans are inserted and classes are multi-modal

244. Model-Reuse Attacks on Deep Learning Systems

245. How To Backdoor Federated Learning

246. Trojaning Attack on Neural Networks

247. Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks

     Summary

     • Proposes neural network poisoning attack that uses "clean labels" which do not require the adversary to mislabel training inputs
     • The paper also presents a optimization based method for generating their poisoning attacks and provides a watermarking strategy for end-to-end attacks that improves the poisoning reliability
     • Authors demonstrate their method by using generated poisoned frog images from the CIFAR dataset to manipulate different kinds of image classifiers

248. Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networks

     Summary

     • Investigate two potential detection methods for backdoor attacks (Fine-tuning and pruning). They find both are insufficient on their own and thus propose a combined detection method which they call "Fine-Pruning"
     • Authors go on to show that on three backdoor techniques "Fine-Pruning" is able to eliminate or reduce Trojans on datasets in the traffic sign, speech, and face recognition domains

249. Technical Report: When Does Machine Learning FAIL? Generalized Transferability for Evasion and Poisoning Attacks

250. Backdoor Embedding in Convolutional Neural Network Models via Invisible Perturbation

251. Hu-Fu: Hardware and Software Collaborative Attack Framework against Neural Networks

252. Attack Strength vs. Detectability Dilemma in Adversarial Machine Learning

253. Data Poisoning Attacks in Contextual Bandits

254. BEBP: An Poisoning Method Against Machine Learning Based IDSs

255. Generative Poisoning Attack Method Against Neural Networks

256. BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain

     Summary

     • Introduce Trojan Attacks— a type of attack where an adversary can create a maliciously trained network (a backdoored neural network, or a BadNet) that has state-of-the-art performance on the user’s training and validation samples, but behaves badly on specific attacker-chosen inputs
     • Demonstrate backdoors in a more realistic scenario by creating a U.S. street sign classifier that identifies stop signs as speed limits when a special sticker is added to the stop sign

257. Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimization

258. Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning

259. Neural Trojans

260. Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimization

261. Certified defenses for data poisoning attacks

262. Data Poisoning Attacks on Factorization-Based Collaborative Filtering

263. Data poisoning attacks against autoregressive models

264. Using machine teaching to identify optimal training-set attacks on machine learners

265. Poisoning Attacks against Support Vector Machines

266. Backdoor Attacks against Learning Systems

267. Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning

268. Antidote: Understanding and defending against poisoning of anomaly detectors