SlideShare a Scribd company logo
1 of 65
Robust Filtering Schemes for Machine
Learning Systems to Defend Adversarial
Attacks
Presented by :
Kishor Datta Gupta
kgupta1@memphis.edu
1
2
Presentation Outline
• Adversarial Attacks(AA)
• Adversarial Defenses
Introduction
Nature of Adversarial Attacks
• Input Filtering scheme
• Output Filtering Scheme
• End-to-End Protection Scheme
Our Solution
Conclusion
Publications
Adversarial Attack (AA) on AI/ML
Types:
• Poisoning Attack : Manipulate training data
“Manipulation of training data, Machine Learning (ML) model architecture, or manipulate
testing data in a way that will result in wrong output from ML”
3
Reference[1]
Adversarial Attack (AA) on AI/ML
Types:
• Poisoning Attack : Manipulate training data
• Evasion Attack: Manipulate input data
“Manipulation of training data, Machine Learning (ML) model architecture, or manipulate
testing data in a way that will result in wrong output from ML”
4
Reference[2]
Adversarial Attack (AA) on AI/ML
Types:
• Poisoning Attack : Manipulate training data
• Evasion Attack: Manipulate input data
• Trojan AI : Manipulate AI Architecture (example: Changes weights value)
“Manipulation of training data, Machine Learning (ML) model architecture, or manipulate
testing data in a way that will result in wrong output from ML”
5
Ref: Dasgupta et el 2020
Type of Evasion-Based Attacks
• One pixel Attack (Not practical)
Score based attack
• Lavan
• Dpatch
Patch Attack ( Human can identify)
• Basic : FGSM, BIM
• Saliency Map attack: JSMA
• Advanced low perturb attacks: CW
Gradient Attack:
• Hopskipjump attack
• Deepfool attack
Decision Attack:
• BPDA
Adaptive Attack:
6
Uses the gradients of the loss with respect to the input image to create a new
image that maximizes the loss. This new image is called the adversarial image.
(Ref: Dasgupta et el 2020)
Defense Strategies for AA
• Generate Adversarial Example and
Retrain the model
• Limitations: Reduce the accuracy of
learning model
Retrain:
• Using PCA, low-pass filtering, JPEG
compression, soft thresholding techniques
as pre-processing technique.
• Limitation: Vulnerable to adaptive attack.
Input Reconstruction or
Transformation:
• Modifying the ML architecture to detect
adversarial attack
• Limitations: Require Modification of
learning models.
Model Modification:
Reference[5]
7
Nature of AAs
Reference[5]
8
Adversarial Input has Noise
Different Attack method has different types of noise/manipulation style
9
Generalized
Example 1
Real Example 2
Filters which can detect some noises
Observation: Clean and adversarial images have quantifiable noise difference
10
Example 1
Example 2
Low Noise AAs are not effective in
Physical world
Percentage of adversarial samples getting ineffective due to environmental factors
11
Attack Types
Minimum Adversarial noise:
Print/Screen
Key Research Focus
12
AA are transferable, and there are numerous ways to formulate AAs in different ML models.
AA have an additional noise signature which is detectable by some filters.
Low noise AAs are not effective in the physical world.
Counter physical world AA and identifying TrojAI can be sufficient if other security policies are effective.
Our initial approach is to focus on adversarial noise detection in input data only,
instead of study how these attack formulated in ML model or how the ML model
behaves. (consider ML models are inaccessible Blackbox)
DF Scheme Architecture
13
Proposed Solution
Input Filter Scheme
Reference[5]
14
DF Scheme Architecture
15
Input Filter Scheme
Detection of Adversarial Traits(3)
16
Detection of Adversarial Traits(2)
17
Average Histogram for white color.
Filters Metrics For Adversarial
Detection
18
Adversarial
Clean
Histogram's representation is dependent of the color of the object being studied, ignoring its shape and texture.
Filters Metrics For Adversarial
Detection
19
Clean
Adversarial
Filters For Adversarial Detection
20
Experimental Setup
Filters Class Filter ID Filter Name Filter ID Filter Name Filter ID Filter Name
Noise Reduction FT1 Sharp FT7 Thickening FT13 Shrink
Noise Addition FT2 blur FT8
Additive
noise
FT14 Dither
Texture Based FT3 Gabor FT9 Census FT15 wavelet
Transformation FT4 Fourier FT10 Laplace FT16 LogPolar
Analytical FT5 Distance FT11 Morph FT17
Gaussian
edge
Edge Detection FT6 Sobel FT12 Canny
21
Different filter types has different accuracy for different attack family.
Relation Between AAs and Filters
• Sharpening, Thickening,
Shrinking
Noise Reduction Filters
•Blur, Additive noise, Dithering
Noise Addition Filters
•Fourier, Laplace, Log polar
Transformation Filters
•Gabor, Census
Texture based Filters
• Sobel, Canny
Edge Detection Filters
• Distance filter, Morph
Analytical Filter
• FGSM,
BIM,MBIM,PGD
Gradient
Attack:
• Hopskipjump,
Deepfool
Decision
Attack:
22
Attack Family
Filter types
Darker means Better Accuracy
Relation Between AAs and Filters (2)
• Sharpening, Thickening,
Shrinking
Noise Reduction Filters
•Blur, Additive noise, Dithering,
Additive Gaussian Noise
Noise Addition Filters
•Fourier, Laplace, Log polar
Transformation Filters
•Gabor, Census
Texture based Filters
• Sobel, Canny
Edge Detection Filters
• Distance filter, Morph
Analytical Filter
• FGSM,
BIM,MBIM,PGD
Gradient
Attack:
• Hopskipjump,
Deepfool
Decision
Attack:
23
Filter from same family has similar performance
AA Attack Family
Filter types
GN
Blur
AN
Dither
ing
Gabor
Census
Darker means Better Accuracy
Genetic Algorithm
24
A small set of filter can show same performance.
Our Goal:
• We need to find an optimal sequence of filters that can detect most of the attack types using
SNR and Histogram values.
• We need multiple set of filter sequences so for each input we can use different sequence of
filter to make the defense system dynamic.
Problem: There can be billion possible combination of filter set can exist.
Determining Filter Sequence to detect AAs
25
Exhaustive search will be computationally costly.
Genetic algorithm can be used for multiple filter sequence set search.
Use of GA is to find optimal sequence of filters with fitness satisfying three objectives: Accuracy ,
Time cost and Diversity of filter family
Population initialized with randomly created filter sequence of variable length, As Example
Individual 1: FT2-FT6-FT7,
Individual 2: FT8-FT3-FT10-FT2
Individual 3: FT7-FT5
..
..
Use of Genetic Algorithm(GA)
• Based on these 3 objective, a variable length multi-objective weighted fitness function
with penalty for longer sequence length shorted the individuals based on their optimality.
• An elitist strategy employed to keep best and used steady-state GA (replacing half of the
population. A random mutation also done to avoid local optima.
• GA evolution terminated after an optimum where best fitness does not change for a period.
Each individual's accuracy calculated based on their accuracy to detect adversarial example
Time cost is based on time consumption filter sequence take to process 100 image.
How many filter family is represented in sequence counted as diversity
26
GA Implementation (Flow Diagram)
27
GA Encoding: Variable Length & different order
28
Illustration of GA Encoding Performance Metrics
29
Maintaining Diversity in GA Population
30
GA run Progression & Dropoffs
GA Performance Measure (Fitness Function)
31
Use of Penalty Function
32
GA Performances with Varying Parameters
33
GA Performances (Cont..)
34
Analysis: GA is better than Random and Brute search
35
MOGA: Input Scheme Experimental Results
Multi-objective GA algorithm
Objective 1: Accuracy (blue)
Objective 2: Time (pink)
Objective 3: Filter Family Diversity (Black)
Red: Each sequence position
Green: Pareto optimal sets
36
Accuracy Against FGSM, JSMA and CW attack samples
MOGA: Experimental Results and Comparison (2)
Adversarial Detection CNN EAF
Accuracy 96% 100% *
Training Time 1200s 90s
Test 100input 0.75s 0.09
Dataset: MNIST Adversarial (600 images)
CNN: 28 convolutional (kernel size 3)-> 2 MaxPool (kernel size 2) ->28 convolutional ->Flatten->relu->128
dense (25 epoch)
EAF: 16 filter
1 GPU, Ram 32GB
* Used all 60000 images to generate range.
37
Limitations of only Input Filter based Defense
• Need to generate adversarial inputs.
• Not effective against TrojAI/Backdoors.
• Vulnerable to adaptive attacks as filter numbers are
finite.
38
Output Filter Scheme
39
DF Scheme Architecture
40
We need an adaptive defense strategy which don’t modify
the learning model and don’t require the adversarial
knowledge.
Output filtering scheme
Detect adversarial input using only the knowledge of
non-adversarial data,
Converting it as an Outlier detection problem
41
Effect of Filters
MNIST dataset: AA and Clean data in 2D space
42
Representation in 2D space
43
Performance in Binary Classification
44
Negative Selection Algorithm for Outlier Detection
• Define Self as a normal pattern of activity or stable behavior of a system/process as a collection of logically split segments (equal-size) of
pattern sequence and represent the collection as a multiset S of strings of length lover a finite alphabet.
• Generate a set R of detectors, each of which fails to match any string in S.
• Monitor new observations (of S) for changes by continually testing the detectors matching against representatives of S. If any detector
ever matches, a change ( or deviation) must have occurred in system behavior.
45
V-detector Negative Selection
Algorithm
Main idea of V-detector By allowing the detectors to have some variable
properties, V-detector enhances negative selection algorithm from
several aspects: It takes fewer large detectors to cover non-self region –
saving time and space Small detector covers holes better. Coverage is
estimated when the detector set is generated. The shapes of detectors
or even the types of matching rules can be extended to be variable too.
46
(Reference: JI and Dasgupta 2005)
Generating (Negative) Detector set
47
Use of Detectors
48
Testing Phase Flow Diagram
49
Experimental Results
Detection accuracy for different attack type for different class of CIFAR and MNIST dataset
50
Detection accuracy for binary classification of clean and adversarial input(all) MNIST dataset
Adversarial attacks on CFIAR and
IMAGENET* detection rate
51
Outlier Detection models
Type Abbr Algorithm
Linear Model
MCD Minimum Covariance Determinant (use the mahalanobis distances as the outlier scores)
OCSVM One-Class Support Vector Machines
LMDD Deviation-based Outlier Detection (LMDD)
Proximity-Based
LOF Local Outlier Factor
COF Connectivity-Based Outlier Factor
CBLOF Clustering-Based Local Outlier Factor
LOCI LOCI: Fast outlier detection using the local correlation integral
HBOS Histogram-based Outlier Score
SOD Subspace Outlier Detection
ROD Rotation-based Outlier Detection
Probabilistic
ABOD Angle-Based Outlier Detection
COPOD COPOD: Copula-Based Outlier Detection
FastABOD Fast Angle-Based Outlier Detection using approximation
MAD Median Absolute Deviation (MAD)
SOS Stochastic Outlier Selection
Outlier Ensembles
IForest Isolation Forest
FB Feature Bagging
LSCP LSCP: Locally Selective Combination of Parallel Outlier Ensembles
XGBOD Extreme Boosting Based Outlier Detection (Supervised)
LODA Lightweight On-line Detector of Anomalies
Neural Networks
AutoEncoder Fully connected AutoEncoder (use reconstruction error as the outlier score)
VAE Variational AutoEncoder (use reconstruction error as the outlier score)
Beta-VAE Variational AutoEncoder (all customized loss term by varying gamma and capacity)
SO_GAAL Single-Objective Generative Adversarial Active Learning
MO_GAAL Multiple-Objective Generative Adversarial Active Learning
52
Comparison of
results with different
outlier detection
models to compare
V-detector NSA
performance with
other OCC methods.
Comparison with different outlier methods
53
Limitation
54
• ML model are processing all the input.
• Detection process is longer for trivial adversarial examples.
End-to-End Scheme
55
DF Scheme Architecture
56
57
Basic Workflow of End-to-End scheme
Detail components of DF scheme 58
Comparison with other methods
F1 score comparison of different detection method with our proposed method
Advantage our proposed method over other detection methods
59
• No attack Sample Generation Needed.
• No ML Model Modification.
• Protection against preprocess based adaptive attacks.
• Independent of ML model architecture and transferable for similar dataset
Summary
• This research conducted extensive investigation to develop end-to-
end protection mechanism for Learning Systems.
• For a given problem/dataset, we used a collection of filters (having
varying degree of discriminatory abilities) to find a robust ensemble of
filters using a genetic algorithm (GA) for AA detection .
• Variable Length MOGA for searching set of Filters that are effective
against different type of AAs.
• We devised an adaptive negative filtering methodology to detect
adversarial attacks that does not modify the ML model or information
about the ML model.
• Our strategy can be implemented in most of the ML-based system
without expensive retraining.
• Current Adaptive attacks are ineffective in our negative filtering
approach as they are regenerating for each/or batch of input. 60
Publications
Patent under submission: System for Dual-Filtering for Learning Systems to Prevent Adversarial Attacks. (APP no: 63/022,323)
by Dipankar Dasgupta, Kishor Datta Gupta
Conference:
•Gupta, Kishor Datta, Dipankar Dasgupta, and Zahid Akhtar. "Adversarial Input Detection Using Image Processing Techniques (IPT)." In 2020 11th
IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON), pp. 0309-0315. IEEE, 2020.
https://doi.org/10.1109/UEMCON51285.2020.9298060
•Gupta, Kishor Datta, Dipankar Dasgupta, and Zahid Akhtar. "Applicability issues of evasion-based adversarial attacks and mitigation techniques." In
2020 IEEE Symposium Series on Computational Intelligence (SSCI), pp. 1506-1515. IEEE, 2020. https://doi.org/10.1109/SSCI47803.2020.9308589
•Gupta, Kishor Datta, and Dipankar Dasgupta. "Using Negative Detectors for Identifying Adversarial Data Manipulation in Machine Learning" In 2021
International Joint Conference on Neural Networks (IJCNN), Shenzhen, China, July18–22, 2021.
Journal:
•Gupta, Kishor Datta and Dipankar Dasgupta. “Dual-Filtering (DF) Schemes for Learning Systems to prevent Adversarial Attacks” Journal: Springer
Complex & Intelligent Systems, Manuscript ID: CAIS-D-21-00347, Submission date: March 2021. (under review)
•Gupta, Kishor Datta, and Dipankar Dasgupta. “Adaptive Ensemble of Filters (AEF) to Detect Adversarial Inputs” Journal: ACM Transactions on
Evolutionary Learning and Optimization, Manuscript ID: TELO-2020-45, Submission date: December 2020. (Second Review)
•Gupta, Kishor Datta, Dipankar Dasgupta, and Zahid Akhtar. “Determining Sequence of Image Processing Technique (IPT) to Detect Adversarial
Attacks” Journal: Springer Nature Computer Science, Manuscript ID: SNCS-D-20-01775, Submission date: October 2020. (Minor Revision),
61
Direction for future researchers
• Can explore use of deep learning method to generate filters
instead of searching using GA.
• Can explore explainable AI method to make the system more
reliable.
• Explore zero-shot learning method to detect adversarial input
using only self data.
62
Different Adversarial Attack Points
in deployed system
63
Effective operating system and communication channel security is a prerequisite.
Q/A
64
References:
1.Machine learning in cyber security: Survey, D Dasgupta, Z Akhtar and Sajib Sen
2.https://medium.com/onfido-tech/adversarial-attacks-and-defences-for-convolutional-neural-networks-66915ece52e7
3.“Poisoning Attacks against Support Vector Machines”, Biggio et al. 2013.[https://arxiv.org/abs/1206.6389]
4.“Intriguing properties of neural networks”, Szegedy et al. 2014. [https://arxiv.org/abs/1312.6199]
5.“Explaining and Harnessing Adversarial Examples”, Goodfellow et al. 2014. [https://arxiv.org/abs/1412.6572]
6.“Towards Evaluating the Robustness of Neural Networks”, Carlini and Wagner 2017b. [https://arxiv.org/abs/1608.04644]
7.“Practical Black-Box Attacks against Machine Learning”, Papernot et al. 2017. [https://arxiv.org/abs/1602.02697]
8.“Attacking Machine Learning with Adversarial Examples”, Goodfellow, 2017. [https://openai.com/blog/adversarial-example-research/]
9.https://medium.com/@ODSC/adversarial-attacks-on-deep-neural-networks-ca847ab1063
10.Tianyu Gu, Brendan Dolan-Gavitt, and Siddharth Garg. Badnets: Identifying vulnerabilities in the machine learning model supply chain. arXiv preprint
arXiv:1708.06733, 2017.
11.A brief survey of Adversarial Machine Learning and Defense Strategies.Z Akhtar, D Dasgupta Technical Report, The University of Memphis,
12.Determining Sequence of Image Processing Technique (IPT) to Detect Adversarial AttacksKD Gupta, D Dasgupta, Z Akhtar, arXiv preprint arXiv:2007.00337
13.Aaditya Prakash, Nick Moran, Solomon Garber, Antonella DiLillo, and James Storer. Deflecting adversarial attacks with pixel deflection. In Proceedings of the IEEE
conference on computer vision and pattern recognition, pages 8571–8580, 2018.
14.Nicholas Carlini. Lessons learned from evaluating the robustness of defenses to adversarial examples. 2019.
15.Nicholas Carlini, Anish Athalye, Nicolas Papernot, Wieland Brendel, Jonas Rauber, Dimitris Tsipras, Ian Goodfellow, Aleksander Madry, and Alexey Kurakin. On
evaluating adversarial robustness. arXiv preprint arXiv:1902.06705, 2019.
16.Nicholas Carlini and DavidWagner. Defensive distillation is not robust to adversarial examples. arXiv preprint arXiv:1607.04311, 2016.
17.Nicholas Carlini and David Wagner. Adversarial examples are not easily detected: Bypassing ten detection methods. In Proceedings of the 10th ACM Workshop on
Artificial Intelligence and Security, pages 3–14, 2017.
18.Nicholas Carlini and DavidWagner. Magnet and" efficient defenses against adversarial attacks“ are not robust to adversarial examples. arXiv preprint
arXiv:1711.08478, 2017.
65

More Related Content

What's hot

Machine Learning Course | Edureka
Machine Learning Course | EdurekaMachine Learning Course | Edureka
Machine Learning Course | Edureka
Edureka!
 
Explainable AI in Industry (AAAI 2020 Tutorial)
Explainable AI in Industry (AAAI 2020 Tutorial)Explainable AI in Industry (AAAI 2020 Tutorial)
Explainable AI in Industry (AAAI 2020 Tutorial)
Krishnaram Kenthapadi
 

What's hot (20)

Artificial Intelligence with Python | Edureka
Artificial Intelligence with Python | EdurekaArtificial Intelligence with Python | Edureka
Artificial Intelligence with Python | Edureka
 
Machine Learning
Machine LearningMachine Learning
Machine Learning
 
Introduction to Machine Learning
Introduction to Machine LearningIntroduction to Machine Learning
Introduction to Machine Learning
 
Robustness in deep learning
Robustness in deep learningRobustness in deep learning
Robustness in deep learning
 
Artificial Intelligence and Machine Learning in Research
Artificial Intelligence and Machine Learning in ResearchArtificial Intelligence and Machine Learning in Research
Artificial Intelligence and Machine Learning in Research
 
Machine learning
Machine learningMachine learning
Machine learning
 
An Introduction to XAI! Towards Trusting Your ML Models!
An Introduction to XAI! Towards Trusting Your ML Models!An Introduction to XAI! Towards Trusting Your ML Models!
An Introduction to XAI! Towards Trusting Your ML Models!
 
WILD PATTERNS - Introduction to Adversarial Machine Learning - ITASEC 2019
WILD PATTERNS - Introduction to Adversarial Machine Learning - ITASEC 2019WILD PATTERNS - Introduction to Adversarial Machine Learning - ITASEC 2019
WILD PATTERNS - Introduction to Adversarial Machine Learning - ITASEC 2019
 
Machine Learning and Real-World Applications
Machine Learning and Real-World ApplicationsMachine Learning and Real-World Applications
Machine Learning and Real-World Applications
 
Adversarial Attacks and Defenses in Deep Learning.pdf
Adversarial Attacks and Defenses in Deep Learning.pdfAdversarial Attacks and Defenses in Deep Learning.pdf
Adversarial Attacks and Defenses in Deep Learning.pdf
 
Machine learning
Machine learningMachine learning
Machine learning
 
Machine Learning Course | Edureka
Machine Learning Course | EdurekaMachine Learning Course | Edureka
Machine Learning Course | Edureka
 
Introduction to AI
Introduction to AIIntroduction to AI
Introduction to AI
 
CounterFactual Explanations.pdf
CounterFactual Explanations.pdfCounterFactual Explanations.pdf
CounterFactual Explanations.pdf
 
Introduction to Artificial Intelligence and Machine Learning
Introduction to Artificial Intelligence and Machine Learning Introduction to Artificial Intelligence and Machine Learning
Introduction to Artificial Intelligence and Machine Learning
 
"An Introduction to Machine Learning and How to Teach Machines to See," a Pre...
"An Introduction to Machine Learning and How to Teach Machines to See," a Pre..."An Introduction to Machine Learning and How to Teach Machines to See," a Pre...
"An Introduction to Machine Learning and How to Teach Machines to See," a Pre...
 
Introduction to Machine Learning
Introduction to Machine LearningIntroduction to Machine Learning
Introduction to Machine Learning
 
Tutorial on Deep Learning
Tutorial on Deep LearningTutorial on Deep Learning
Tutorial on Deep Learning
 
Explainable AI in Industry (AAAI 2020 Tutorial)
Explainable AI in Industry (AAAI 2020 Tutorial)Explainable AI in Industry (AAAI 2020 Tutorial)
Explainable AI in Industry (AAAI 2020 Tutorial)
 
Machine Learning
Machine LearningMachine Learning
Machine Learning
 

Similar to Robust Filtering Schemes for Machine Learning Systems to Defend Adversarial Attacks

A Sparse-Coding Based Approach for Class-Specific Feature Selection
A Sparse-Coding Based Approach for Class-Specific Feature SelectionA Sparse-Coding Based Approach for Class-Specific Feature Selection
A Sparse-Coding Based Approach for Class-Specific Feature Selection
Davide Nardone
 

Similar to Robust Filtering Schemes for Machine Learning Systems to Defend Adversarial Attacks (20)

Injection Attack detection using ML for
Injection Attack detection using ML  forInjection Attack detection using ML  for
Injection Attack detection using ML for
 
Ppt on speech processing by ranbeer
Ppt on speech processing by ranbeerPpt on speech processing by ranbeer
Ppt on speech processing by ranbeer
 
IDS for IoT.pptx
IDS for IoT.pptxIDS for IoT.pptx
IDS for IoT.pptx
 
Ah34207211
Ah34207211Ah34207211
Ah34207211
 
A Sparse-Coding Based Approach for Class-Specific Feature Selection
A Sparse-Coding Based Approach for Class-Specific Feature SelectionA Sparse-Coding Based Approach for Class-Specific Feature Selection
A Sparse-Coding Based Approach for Class-Specific Feature Selection
 
Simulation of EMI Filters Using Matlab
Simulation of EMI Filters Using MatlabSimulation of EMI Filters Using Matlab
Simulation of EMI Filters Using Matlab
 
The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)The International Journal of Engineering and Science (The IJES)
The International Journal of Engineering and Science (The IJES)
 
ASP UNIT 1 QUESTIONBANK ANSWERS.pdf
ASP UNIT 1 QUESTIONBANK ANSWERS.pdfASP UNIT 1 QUESTIONBANK ANSWERS.pdf
ASP UNIT 1 QUESTIONBANK ANSWERS.pdf
 
ASP UNIT 1 QUESTIONBANK ANSWERS (1).pdf
ASP UNIT 1 QUESTIONBANK ANSWERS (1).pdfASP UNIT 1 QUESTIONBANK ANSWERS (1).pdf
ASP UNIT 1 QUESTIONBANK ANSWERS (1).pdf
 
A Threshold fuzzy entropy based feature selection method applied in various b...
A Threshold fuzzy entropy based feature selection method applied in various b...A Threshold fuzzy entropy based feature selection method applied in various b...
A Threshold fuzzy entropy based feature selection method applied in various b...
 
Machine learning and linear regression programming
Machine learning and linear regression programmingMachine learning and linear regression programming
Machine learning and linear regression programming
 
A Tale of Experiments on Bug Prediction
A Tale of Experiments on Bug PredictionA Tale of Experiments on Bug Prediction
A Tale of Experiments on Bug Prediction
 
Subverting Machine Learning Detections for fun and profit
Subverting Machine Learning Detections for fun and profitSubverting Machine Learning Detections for fun and profit
Subverting Machine Learning Detections for fun and profit
 
Testing of Cyber-Physical Systems: Diversity-driven Strategies
Testing of Cyber-Physical Systems: Diversity-driven StrategiesTesting of Cyber-Physical Systems: Diversity-driven Strategies
Testing of Cyber-Physical Systems: Diversity-driven Strategies
 
Feature Selection Techniques for Software Fault Prediction (Summary)
Feature Selection Techniques for Software Fault Prediction (Summary)Feature Selection Techniques for Software Fault Prediction (Summary)
Feature Selection Techniques for Software Fault Prediction (Summary)
 
Kaggle Gold Medal Case Study
Kaggle Gold Medal Case StudyKaggle Gold Medal Case Study
Kaggle Gold Medal Case Study
 
Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...
Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...
Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...
 
Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...
Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...
Modeling the Effect of Packet Loss on Speech Quality: Genetic Programming Bas...
 
digital filters on open-loop system.pptx
digital filters on open-loop system.pptxdigital filters on open-loop system.pptx
digital filters on open-loop system.pptx
 
Deep Recurrent Neural Network for Multi-target Filtering
Deep Recurrent Neural Network for Multi-target FilteringDeep Recurrent Neural Network for Multi-target Filtering
Deep Recurrent Neural Network for Multi-target Filtering
 

More from Kishor Datta Gupta

Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...
Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...
Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...
Kishor Datta Gupta
 
A safer approach to build recommendation systems on unidentifiable data
A safer approach to build recommendation systems on unidentifiable dataA safer approach to build recommendation systems on unidentifiable data
A safer approach to build recommendation systems on unidentifiable data
Kishor Datta Gupta
 
Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...
Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...
Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...
Kishor Datta Gupta
 

More from Kishor Datta Gupta (20)

GAN introduction.pptx
GAN introduction.pptxGAN introduction.pptx
GAN introduction.pptx
 
Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...
Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...
Interpretable Learning Model for Lower Dimensional Feature Space: A Case stud...
 
A safer approach to build recommendation systems on unidentifiable data
A safer approach to build recommendation systems on unidentifiable dataA safer approach to build recommendation systems on unidentifiable data
A safer approach to build recommendation systems on unidentifiable data
 
Adversarial Attacks and Defense
Adversarial Attacks and DefenseAdversarial Attacks and Defense
Adversarial Attacks and Defense
 
Who is responsible for adversarial defense
Who is responsible for adversarial defenseWho is responsible for adversarial defense
Who is responsible for adversarial defense
 
Zero shot learning
Zero shot learning Zero shot learning
Zero shot learning
 
Using Negative Detectors for Identifying Adversarial Data Manipulation in Mac...
Using Negative Detectors for Identifying Adversarial Data Manipulation in Mac...Using Negative Detectors for Identifying Adversarial Data Manipulation in Mac...
Using Negative Detectors for Identifying Adversarial Data Manipulation in Mac...
 
Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...
Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...
Deep Reinforcement Learning based Recommendation with Explicit User-ItemInter...
 
Machine learning in computer security
Machine learning in computer securityMachine learning in computer security
Machine learning in computer security
 
Policy Based reinforcement Learning for time series Anomaly detection
Policy Based reinforcement Learning for time series Anomaly detectionPolicy Based reinforcement Learning for time series Anomaly detection
Policy Based reinforcement Learning for time series Anomaly detection
 
Cyber intrusion
Cyber intrusionCyber intrusion
Cyber intrusion
 
understanding the pandemic through mining covid news using natural language p...
understanding the pandemic through mining covid news using natural language p...understanding the pandemic through mining covid news using natural language p...
understanding the pandemic through mining covid news using natural language p...
 
Different representation space for MNIST digit
Different representation space for MNIST digitDifferent representation space for MNIST digit
Different representation space for MNIST digit
 
"Can NLP techniques be utilized as a reliable tool for medical science?" -Bui...
"Can NLP techniques be utilized as a reliable tool for medical science?" -Bui..."Can NLP techniques be utilized as a reliable tool for medical science?" -Bui...
"Can NLP techniques be utilized as a reliable tool for medical science?" -Bui...
 
Applicability issues of Evasion-Based Adversarial Attacks and Mitigation Tech...
Applicability issues of Evasion-Based Adversarial Attacks and Mitigation Tech...Applicability issues of Evasion-Based Adversarial Attacks and Mitigation Tech...
Applicability issues of Evasion-Based Adversarial Attacks and Mitigation Tech...
 
Adversarial Input Detection Using Image Processing Techniques (IPT)
Adversarial Input Detection Using Image Processing Techniques (IPT)Adversarial Input Detection Using Image Processing Techniques (IPT)
Adversarial Input Detection Using Image Processing Techniques (IPT)
 
Clustering report
Clustering reportClustering report
Clustering report
 
Basic digital image concept
Basic digital image conceptBasic digital image concept
Basic digital image concept
 
An empirical study on algorithmic bias (aiml compsac2020)
An empirical study on algorithmic bias (aiml compsac2020)An empirical study on algorithmic bias (aiml compsac2020)
An empirical study on algorithmic bias (aiml compsac2020)
 
Hybrid pow-pos-based-system against majority attack-in-cryptocurrency system ...
Hybrid pow-pos-based-system against majority attack-in-cryptocurrency system ...Hybrid pow-pos-based-system against majority attack-in-cryptocurrency system ...
Hybrid pow-pos-based-system against majority attack-in-cryptocurrency system ...
 

Recently uploaded

Kuwait City MTP kit ((+919101817206)) Buy Abortion Pills Kuwait
Kuwait City MTP kit ((+919101817206)) Buy Abortion Pills KuwaitKuwait City MTP kit ((+919101817206)) Buy Abortion Pills Kuwait
Kuwait City MTP kit ((+919101817206)) Buy Abortion Pills Kuwait
jaanualu31
 
Integrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - NeometrixIntegrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - Neometrix
Neometrix_Engineering_Pvt_Ltd
 
+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...
+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...
+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...
Health
 
DeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakesDeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakes
MayuraD1
 
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments""Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
mphochane1998
 
Standard vs Custom Battery Packs - Decoding the Power Play
Standard vs Custom Battery Packs - Decoding the Power PlayStandard vs Custom Battery Packs - Decoding the Power Play
Standard vs Custom Battery Packs - Decoding the Power Play
Epec Engineered Technologies
 
notes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptnotes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.ppt
MsecMca
 
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
ssuser89054b
 

Recently uploaded (20)

Kuwait City MTP kit ((+919101817206)) Buy Abortion Pills Kuwait
Kuwait City MTP kit ((+919101817206)) Buy Abortion Pills KuwaitKuwait City MTP kit ((+919101817206)) Buy Abortion Pills Kuwait
Kuwait City MTP kit ((+919101817206)) Buy Abortion Pills Kuwait
 
Hostel management system project report..pdf
Hostel management system project report..pdfHostel management system project report..pdf
Hostel management system project report..pdf
 
Learn the concepts of Thermodynamics on Magic Marks
Learn the concepts of Thermodynamics on Magic MarksLearn the concepts of Thermodynamics on Magic Marks
Learn the concepts of Thermodynamics on Magic Marks
 
Integrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - NeometrixIntegrated Test Rig For HTFE-25 - Neometrix
Integrated Test Rig For HTFE-25 - Neometrix
 
+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...
+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...
+97470301568>> buy weed in qatar,buy thc oil qatar,buy weed and vape oil in d...
 
Block diagram reduction techniques in control systems.ppt
Block diagram reduction techniques in control systems.pptBlock diagram reduction techniques in control systems.ppt
Block diagram reduction techniques in control systems.ppt
 
A Study of Urban Area Plan for Pabna Municipality
A Study of Urban Area Plan for Pabna MunicipalityA Study of Urban Area Plan for Pabna Municipality
A Study of Urban Area Plan for Pabna Municipality
 
DeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakesDeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakes
 
Design For Accessibility: Getting it right from the start
Design For Accessibility: Getting it right from the startDesign For Accessibility: Getting it right from the start
Design For Accessibility: Getting it right from the start
 
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments""Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
 
Standard vs Custom Battery Packs - Decoding the Power Play
Standard vs Custom Battery Packs - Decoding the Power PlayStandard vs Custom Battery Packs - Decoding the Power Play
Standard vs Custom Battery Packs - Decoding the Power Play
 
Generative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPTGenerative AI or GenAI technology based PPT
Generative AI or GenAI technology based PPT
 
data_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfdata_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdf
 
notes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.pptnotes on Evolution Of Analytic Scalability.ppt
notes on Evolution Of Analytic Scalability.ppt
 
Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...
Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...
Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...
 
S1S2 B.Arch MGU - HOA1&2 Module 3 -Temple Architecture of Kerala.pptx
S1S2 B.Arch MGU - HOA1&2 Module 3 -Temple Architecture of Kerala.pptxS1S2 B.Arch MGU - HOA1&2 Module 3 -Temple Architecture of Kerala.pptx
S1S2 B.Arch MGU - HOA1&2 Module 3 -Temple Architecture of Kerala.pptx
 
Air Compressor reciprocating single stage
Air Compressor reciprocating single stageAir Compressor reciprocating single stage
Air Compressor reciprocating single stage
 
Thermal Engineering-R & A / C - unit - V
Thermal Engineering-R & A / C - unit - VThermal Engineering-R & A / C - unit - V
Thermal Engineering-R & A / C - unit - V
 
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
 
Bridge Jacking Design Sample Calculation.pptx
Bridge Jacking Design Sample Calculation.pptxBridge Jacking Design Sample Calculation.pptx
Bridge Jacking Design Sample Calculation.pptx
 

Robust Filtering Schemes for Machine Learning Systems to Defend Adversarial Attacks

  • 1. Robust Filtering Schemes for Machine Learning Systems to Defend Adversarial Attacks Presented by : Kishor Datta Gupta kgupta1@memphis.edu 1
  • 2. 2 Presentation Outline • Adversarial Attacks(AA) • Adversarial Defenses Introduction Nature of Adversarial Attacks • Input Filtering scheme • Output Filtering Scheme • End-to-End Protection Scheme Our Solution Conclusion Publications
  • 3. Adversarial Attack (AA) on AI/ML Types: • Poisoning Attack : Manipulate training data “Manipulation of training data, Machine Learning (ML) model architecture, or manipulate testing data in a way that will result in wrong output from ML” 3 Reference[1]
  • 4. Adversarial Attack (AA) on AI/ML Types: • Poisoning Attack : Manipulate training data • Evasion Attack: Manipulate input data “Manipulation of training data, Machine Learning (ML) model architecture, or manipulate testing data in a way that will result in wrong output from ML” 4 Reference[2]
  • 5. Adversarial Attack (AA) on AI/ML Types: • Poisoning Attack : Manipulate training data • Evasion Attack: Manipulate input data • Trojan AI : Manipulate AI Architecture (example: Changes weights value) “Manipulation of training data, Machine Learning (ML) model architecture, or manipulate testing data in a way that will result in wrong output from ML” 5 Ref: Dasgupta et el 2020
  • 6. Type of Evasion-Based Attacks • One pixel Attack (Not practical) Score based attack • Lavan • Dpatch Patch Attack ( Human can identify) • Basic : FGSM, BIM • Saliency Map attack: JSMA • Advanced low perturb attacks: CW Gradient Attack: • Hopskipjump attack • Deepfool attack Decision Attack: • BPDA Adaptive Attack: 6 Uses the gradients of the loss with respect to the input image to create a new image that maximizes the loss. This new image is called the adversarial image. (Ref: Dasgupta et el 2020)
  • 7. Defense Strategies for AA • Generate Adversarial Example and Retrain the model • Limitations: Reduce the accuracy of learning model Retrain: • Using PCA, low-pass filtering, JPEG compression, soft thresholding techniques as pre-processing technique. • Limitation: Vulnerable to adaptive attack. Input Reconstruction or Transformation: • Modifying the ML architecture to detect adversarial attack • Limitations: Require Modification of learning models. Model Modification: Reference[5] 7
  • 9. Adversarial Input has Noise Different Attack method has different types of noise/manipulation style 9 Generalized Example 1 Real Example 2
  • 10. Filters which can detect some noises Observation: Clean and adversarial images have quantifiable noise difference 10 Example 1 Example 2
  • 11. Low Noise AAs are not effective in Physical world Percentage of adversarial samples getting ineffective due to environmental factors 11 Attack Types Minimum Adversarial noise: Print/Screen
  • 12. Key Research Focus 12 AA are transferable, and there are numerous ways to formulate AAs in different ML models. AA have an additional noise signature which is detectable by some filters. Low noise AAs are not effective in the physical world. Counter physical world AA and identifying TrojAI can be sufficient if other security policies are effective. Our initial approach is to focus on adversarial noise detection in input data only, instead of study how these attack formulated in ML model or how the ML model behaves. (consider ML models are inaccessible Blackbox)
  • 16. Detection of Adversarial Traits(3) 16
  • 17. Detection of Adversarial Traits(2) 17
  • 18. Average Histogram for white color. Filters Metrics For Adversarial Detection 18 Adversarial Clean Histogram's representation is dependent of the color of the object being studied, ignoring its shape and texture.
  • 19. Filters Metrics For Adversarial Detection 19 Clean Adversarial
  • 20. Filters For Adversarial Detection 20
  • 21. Experimental Setup Filters Class Filter ID Filter Name Filter ID Filter Name Filter ID Filter Name Noise Reduction FT1 Sharp FT7 Thickening FT13 Shrink Noise Addition FT2 blur FT8 Additive noise FT14 Dither Texture Based FT3 Gabor FT9 Census FT15 wavelet Transformation FT4 Fourier FT10 Laplace FT16 LogPolar Analytical FT5 Distance FT11 Morph FT17 Gaussian edge Edge Detection FT6 Sobel FT12 Canny 21
  • 22. Different filter types has different accuracy for different attack family. Relation Between AAs and Filters • Sharpening, Thickening, Shrinking Noise Reduction Filters •Blur, Additive noise, Dithering Noise Addition Filters •Fourier, Laplace, Log polar Transformation Filters •Gabor, Census Texture based Filters • Sobel, Canny Edge Detection Filters • Distance filter, Morph Analytical Filter • FGSM, BIM,MBIM,PGD Gradient Attack: • Hopskipjump, Deepfool Decision Attack: 22 Attack Family Filter types Darker means Better Accuracy
  • 23. Relation Between AAs and Filters (2) • Sharpening, Thickening, Shrinking Noise Reduction Filters •Blur, Additive noise, Dithering, Additive Gaussian Noise Noise Addition Filters •Fourier, Laplace, Log polar Transformation Filters •Gabor, Census Texture based Filters • Sobel, Canny Edge Detection Filters • Distance filter, Morph Analytical Filter • FGSM, BIM,MBIM,PGD Gradient Attack: • Hopskipjump, Deepfool Decision Attack: 23 Filter from same family has similar performance AA Attack Family Filter types GN Blur AN Dither ing Gabor Census Darker means Better Accuracy
  • 24. Genetic Algorithm 24 A small set of filter can show same performance.
  • 25. Our Goal: • We need to find an optimal sequence of filters that can detect most of the attack types using SNR and Histogram values. • We need multiple set of filter sequences so for each input we can use different sequence of filter to make the defense system dynamic. Problem: There can be billion possible combination of filter set can exist. Determining Filter Sequence to detect AAs 25 Exhaustive search will be computationally costly. Genetic algorithm can be used for multiple filter sequence set search.
  • 26. Use of GA is to find optimal sequence of filters with fitness satisfying three objectives: Accuracy , Time cost and Diversity of filter family Population initialized with randomly created filter sequence of variable length, As Example Individual 1: FT2-FT6-FT7, Individual 2: FT8-FT3-FT10-FT2 Individual 3: FT7-FT5 .. .. Use of Genetic Algorithm(GA) • Based on these 3 objective, a variable length multi-objective weighted fitness function with penalty for longer sequence length shorted the individuals based on their optimality. • An elitist strategy employed to keep best and used steady-state GA (replacing half of the population. A random mutation also done to avoid local optima. • GA evolution terminated after an optimum where best fitness does not change for a period. Each individual's accuracy calculated based on their accuracy to detect adversarial example Time cost is based on time consumption filter sequence take to process 100 image. How many filter family is represented in sequence counted as diversity 26
  • 27. GA Implementation (Flow Diagram) 27
  • 28. GA Encoding: Variable Length & different order 28 Illustration of GA Encoding Performance Metrics
  • 30. 30 GA run Progression & Dropoffs
  • 31. GA Performance Measure (Fitness Function) 31
  • 32. Use of Penalty Function 32
  • 33. GA Performances with Varying Parameters 33
  • 35. Analysis: GA is better than Random and Brute search 35
  • 36. MOGA: Input Scheme Experimental Results Multi-objective GA algorithm Objective 1: Accuracy (blue) Objective 2: Time (pink) Objective 3: Filter Family Diversity (Black) Red: Each sequence position Green: Pareto optimal sets 36 Accuracy Against FGSM, JSMA and CW attack samples
  • 37. MOGA: Experimental Results and Comparison (2) Adversarial Detection CNN EAF Accuracy 96% 100% * Training Time 1200s 90s Test 100input 0.75s 0.09 Dataset: MNIST Adversarial (600 images) CNN: 28 convolutional (kernel size 3)-> 2 MaxPool (kernel size 2) ->28 convolutional ->Flatten->relu->128 dense (25 epoch) EAF: 16 filter 1 GPU, Ram 32GB * Used all 60000 images to generate range. 37
  • 38. Limitations of only Input Filter based Defense • Need to generate adversarial inputs. • Not effective against TrojAI/Backdoors. • Vulnerable to adaptive attacks as filter numbers are finite. 38
  • 41. We need an adaptive defense strategy which don’t modify the learning model and don’t require the adversarial knowledge. Output filtering scheme Detect adversarial input using only the knowledge of non-adversarial data, Converting it as an Outlier detection problem 41
  • 42. Effect of Filters MNIST dataset: AA and Clean data in 2D space 42
  • 44. Performance in Binary Classification 44
  • 45. Negative Selection Algorithm for Outlier Detection • Define Self as a normal pattern of activity or stable behavior of a system/process as a collection of logically split segments (equal-size) of pattern sequence and represent the collection as a multiset S of strings of length lover a finite alphabet. • Generate a set R of detectors, each of which fails to match any string in S. • Monitor new observations (of S) for changes by continually testing the detectors matching against representatives of S. If any detector ever matches, a change ( or deviation) must have occurred in system behavior. 45
  • 46. V-detector Negative Selection Algorithm Main idea of V-detector By allowing the detectors to have some variable properties, V-detector enhances negative selection algorithm from several aspects: It takes fewer large detectors to cover non-self region – saving time and space Small detector covers holes better. Coverage is estimated when the detector set is generated. The shapes of detectors or even the types of matching rules can be extended to be variable too. 46 (Reference: JI and Dasgupta 2005)
  • 49. Testing Phase Flow Diagram 49
  • 50. Experimental Results Detection accuracy for different attack type for different class of CIFAR and MNIST dataset 50 Detection accuracy for binary classification of clean and adversarial input(all) MNIST dataset
  • 51. Adversarial attacks on CFIAR and IMAGENET* detection rate 51
  • 52. Outlier Detection models Type Abbr Algorithm Linear Model MCD Minimum Covariance Determinant (use the mahalanobis distances as the outlier scores) OCSVM One-Class Support Vector Machines LMDD Deviation-based Outlier Detection (LMDD) Proximity-Based LOF Local Outlier Factor COF Connectivity-Based Outlier Factor CBLOF Clustering-Based Local Outlier Factor LOCI LOCI: Fast outlier detection using the local correlation integral HBOS Histogram-based Outlier Score SOD Subspace Outlier Detection ROD Rotation-based Outlier Detection Probabilistic ABOD Angle-Based Outlier Detection COPOD COPOD: Copula-Based Outlier Detection FastABOD Fast Angle-Based Outlier Detection using approximation MAD Median Absolute Deviation (MAD) SOS Stochastic Outlier Selection Outlier Ensembles IForest Isolation Forest FB Feature Bagging LSCP LSCP: Locally Selective Combination of Parallel Outlier Ensembles XGBOD Extreme Boosting Based Outlier Detection (Supervised) LODA Lightweight On-line Detector of Anomalies Neural Networks AutoEncoder Fully connected AutoEncoder (use reconstruction error as the outlier score) VAE Variational AutoEncoder (use reconstruction error as the outlier score) Beta-VAE Variational AutoEncoder (all customized loss term by varying gamma and capacity) SO_GAAL Single-Objective Generative Adversarial Active Learning MO_GAAL Multiple-Objective Generative Adversarial Active Learning 52
  • 53. Comparison of results with different outlier detection models to compare V-detector NSA performance with other OCC methods. Comparison with different outlier methods 53
  • 54. Limitation 54 • ML model are processing all the input. • Detection process is longer for trivial adversarial examples.
  • 57. 57 Basic Workflow of End-to-End scheme
  • 58. Detail components of DF scheme 58
  • 59. Comparison with other methods F1 score comparison of different detection method with our proposed method Advantage our proposed method over other detection methods 59 • No attack Sample Generation Needed. • No ML Model Modification. • Protection against preprocess based adaptive attacks. • Independent of ML model architecture and transferable for similar dataset
  • 60. Summary • This research conducted extensive investigation to develop end-to- end protection mechanism for Learning Systems. • For a given problem/dataset, we used a collection of filters (having varying degree of discriminatory abilities) to find a robust ensemble of filters using a genetic algorithm (GA) for AA detection . • Variable Length MOGA for searching set of Filters that are effective against different type of AAs. • We devised an adaptive negative filtering methodology to detect adversarial attacks that does not modify the ML model or information about the ML model. • Our strategy can be implemented in most of the ML-based system without expensive retraining. • Current Adaptive attacks are ineffective in our negative filtering approach as they are regenerating for each/or batch of input. 60
  • 61. Publications Patent under submission: System for Dual-Filtering for Learning Systems to Prevent Adversarial Attacks. (APP no: 63/022,323) by Dipankar Dasgupta, Kishor Datta Gupta Conference: •Gupta, Kishor Datta, Dipankar Dasgupta, and Zahid Akhtar. "Adversarial Input Detection Using Image Processing Techniques (IPT)." In 2020 11th IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON), pp. 0309-0315. IEEE, 2020. https://doi.org/10.1109/UEMCON51285.2020.9298060 •Gupta, Kishor Datta, Dipankar Dasgupta, and Zahid Akhtar. "Applicability issues of evasion-based adversarial attacks and mitigation techniques." In 2020 IEEE Symposium Series on Computational Intelligence (SSCI), pp. 1506-1515. IEEE, 2020. https://doi.org/10.1109/SSCI47803.2020.9308589 •Gupta, Kishor Datta, and Dipankar Dasgupta. "Using Negative Detectors for Identifying Adversarial Data Manipulation in Machine Learning" In 2021 International Joint Conference on Neural Networks (IJCNN), Shenzhen, China, July18–22, 2021. Journal: •Gupta, Kishor Datta and Dipankar Dasgupta. “Dual-Filtering (DF) Schemes for Learning Systems to prevent Adversarial Attacks” Journal: Springer Complex & Intelligent Systems, Manuscript ID: CAIS-D-21-00347, Submission date: March 2021. (under review) •Gupta, Kishor Datta, and Dipankar Dasgupta. “Adaptive Ensemble of Filters (AEF) to Detect Adversarial Inputs” Journal: ACM Transactions on Evolutionary Learning and Optimization, Manuscript ID: TELO-2020-45, Submission date: December 2020. (Second Review) •Gupta, Kishor Datta, Dipankar Dasgupta, and Zahid Akhtar. “Determining Sequence of Image Processing Technique (IPT) to Detect Adversarial Attacks” Journal: Springer Nature Computer Science, Manuscript ID: SNCS-D-20-01775, Submission date: October 2020. (Minor Revision), 61
  • 62. Direction for future researchers • Can explore use of deep learning method to generate filters instead of searching using GA. • Can explore explainable AI method to make the system more reliable. • Explore zero-shot learning method to detect adversarial input using only self data. 62
  • 63. Different Adversarial Attack Points in deployed system 63 Effective operating system and communication channel security is a prerequisite.
  • 65. References: 1.Machine learning in cyber security: Survey, D Dasgupta, Z Akhtar and Sajib Sen 2.https://medium.com/onfido-tech/adversarial-attacks-and-defences-for-convolutional-neural-networks-66915ece52e7 3.“Poisoning Attacks against Support Vector Machines”, Biggio et al. 2013.[https://arxiv.org/abs/1206.6389] 4.“Intriguing properties of neural networks”, Szegedy et al. 2014. [https://arxiv.org/abs/1312.6199] 5.“Explaining and Harnessing Adversarial Examples”, Goodfellow et al. 2014. [https://arxiv.org/abs/1412.6572] 6.“Towards Evaluating the Robustness of Neural Networks”, Carlini and Wagner 2017b. [https://arxiv.org/abs/1608.04644] 7.“Practical Black-Box Attacks against Machine Learning”, Papernot et al. 2017. [https://arxiv.org/abs/1602.02697] 8.“Attacking Machine Learning with Adversarial Examples”, Goodfellow, 2017. [https://openai.com/blog/adversarial-example-research/] 9.https://medium.com/@ODSC/adversarial-attacks-on-deep-neural-networks-ca847ab1063 10.Tianyu Gu, Brendan Dolan-Gavitt, and Siddharth Garg. Badnets: Identifying vulnerabilities in the machine learning model supply chain. arXiv preprint arXiv:1708.06733, 2017. 11.A brief survey of Adversarial Machine Learning and Defense Strategies.Z Akhtar, D Dasgupta Technical Report, The University of Memphis, 12.Determining Sequence of Image Processing Technique (IPT) to Detect Adversarial AttacksKD Gupta, D Dasgupta, Z Akhtar, arXiv preprint arXiv:2007.00337 13.Aaditya Prakash, Nick Moran, Solomon Garber, Antonella DiLillo, and James Storer. Deflecting adversarial attacks with pixel deflection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 8571–8580, 2018. 14.Nicholas Carlini. Lessons learned from evaluating the robustness of defenses to adversarial examples. 2019. 15.Nicholas Carlini, Anish Athalye, Nicolas Papernot, Wieland Brendel, Jonas Rauber, Dimitris Tsipras, Ian Goodfellow, Aleksander Madry, and Alexey Kurakin. On evaluating adversarial robustness. arXiv preprint arXiv:1902.06705, 2019. 16.Nicholas Carlini and DavidWagner. Defensive distillation is not robust to adversarial examples. arXiv preprint arXiv:1607.04311, 2016. 17.Nicholas Carlini and David Wagner. Adversarial examples are not easily detected: Bypassing ten detection methods. In Proceedings of the 10th ACM Workshop on Artificial Intelligence and Security, pages 3–14, 2017. 18.Nicholas Carlini and DavidWagner. Magnet and" efficient defenses against adversarial attacks“ are not robust to adversarial examples. arXiv preprint arXiv:1711.08478, 2017. 65