Never-Ending Language Learner

“But Watson couldn’t distinguish between polite language and profanity — which the Urban Dictionary is full of” - Eric Brown (IBM)

Subverting Machine Learning for Fun And

ProfitRam Shankar Siva Kumar, John Walton

Email: Ram.Shankar@Microsoft.com; JoWalt@Microsoft.com

Goals

• This talk:• Is a primer on Adversarial Machine Learning • Will show, through a sampling, how ML algorithms are vulnerable • Illustrates how to defend against such attacks

• This talk IS NOT • An exhaustive review of all algorithms

• End goal: Gain an intuitive understanding of ML algorithms and how to attack them

Agenda

• Motivation to Attack ML systems • Practical Attacks and Defenses • Best Practices

ML is everywhere…

“Machine Learning is shifting from an academic discipline to an industrial tool” – John Langford

In Security…!!

“The only effective approach to defending against today’s ever-increasing volume and diversity of attacks is to shift to fully automated systems capable of discovering and neutralizing attacks instantly.” - Mike Walker (on DARPA Cyber Grand Challenge)

Computer System

Data

ProgramOutput

Computer System

Data

OutputProgram

Traditional Programming

Machine Learning

Source: Lectures by Pedro Domingos

• For the program to be functional, input data must be functional • What does a program/model look like? • Literally, bunch of numbers and data points • The output model can be expressed in terms of parameters:

Things to note about

1 2 3 4 5 6 7 80

500

1,000

1,500

2,000

2,500

3,000

3,500

f(x) = 225 x + 875R² = 0.57641477465266

Linear Regression

Time

Num

ber o

f Log

ons

1 2 3 4 5 6 7 80

500

1,000

1,500

2,000

2,500

3,000

3,500

f(x) = 982.232243060313 exp( 0.130492162328839 x )R² = 0.662398041395387

Non-Linear

Number of Logons = 225 * Time + 875

Number of Logons = 982*e0.1305* (Time)

• Data and parameters define the model • By controlling the data or parameters, you can change the model• Where do you find them? • Data

• At the source • Collected in a big data store • Stored in the cloud (MLaaS)

• Parameters: • Code repository

Malicious Mindset

The mother lode

Data is collectedData is within

anomaly detector’s purview

Anomaly is significant for

detector

Anomaly is surfaced!

Source: Arun Viswanathan, Kymie Tan, and Clifford Neuman, Deconstructing the Assessment of Anomaly-based Intrusion Detectors, RAID 2013.

Putting it all together

• Opportunity = ML is/will be everywhere• Prevalence = ML is/will be widely used in security • Ease = (most) ML algorithms can be easily subverted by controlling

data/parameters• High rate of return = Once subverted, you can evade or even control

the system

Opportunity * Prevalence * Ease * High Rate of Return =

Agenda

• Motivation to Attack ML systems • Practical Attacks and Defenses • Intuitive understanding of the algorithm • How the system looks before the attack? • How the system looks after the attack?• How to defend from these attacks? • Takeaway – From Evasion to total control of the system

• Best Practices

About the dataset

• Used Enron Spam Dataset • Came out of the Federal investigation of Enron corporation• Real world corpus of spam and ham messages.

• 619,446 email messages belonging to 158 users. After cleaning it up (removing duplicate messages, discussion threads), you end up with 200,399 messages.

SPAM! HAM!

SPAM? HAM?

Word P(Word|Spam) P(Word|Ham)

Assets 0/3 2/3

Assignment 0/3 2/3

Cialis 3/3 0/3

Group 0/3 2/3

Viagra 1/3 0/3

Vallium 2/3 0/3

Naïve Bayes Algorithm

Choose whichever probability is higher:

P(Spam|M) = 0.5*(0/3)*(0/3)*(0/3) = 0 P(Ham|M) = 0.5*(2/3)*(2/3)*(2/3) = 0.14

Since 0.14 > 0 => Message is more likely to be Ham

Before Attack

• Built a vanilla Naïve Bayes classifier on Enron email dataset (with some normalizations)• Goal: Given a new subject, can I predict if it is spam or ham? • Testing on 20% of data, you get test accuracy of 62%

After the attack

• Good Word Attack: Introduce innocuous words in the messageE.g: Gas Meeting Expense Report Payroll -> Test Accuracy dropped to 52.8%

0 5 10 15 20 25 30 350

20

40

60

80

100

Number of Benign words added

Fals

e Po

sitiv

e Ra

te

Takeaway

• How to use in real-world: Spear phishing

• By manipulating the input to the algorithm, we can increase the false positive rate• Make the system unusable!

Support Vector Machines – The Ferrari of ML

• Immensely popular• Quite fast • Deliver a solid performance• Widely used in classification setting

In Security setting, beginning to gain popularity in the Malware community.• Goal: Given a piece of code, is it Malicious

or benign?

Intuition

Which is the right decision boundary?

SVM Intuition

Choose the hyperplane, that maximizes the margin between the positive and negative examples!

Those examples on the boundary are called support vectors!

}

Facts about SVMs

• Output of SVM = a set of weights + Support vectors• Once you have the support vectors {special points in the

training data}, rest of the training data can be thrown away • Takeaway: A good part of the model, is determined by

support vectors• Intuition: Controlling the support vectors, should help us to

control the model

Going after support vectors

Takeaway

• How it can be used in real-world: Fool the malware classifier• Changes to support vectors, lead to changes in decision boundary

Clustering

• Widely used learning algorithm for anomaly detection

Attack Intuition

Center

Before Attack

After Attack

Attack Point to be included

Source:Laskov, Pavel, and Marius Kloft. "A framework for quantitative security analysis of machine learning." Proceedings of the 2nd ACM workshop on Security and artificial intelligence. ACM, 2009.

Takeaway

• In order to attack the algorithm, we don’t change the parameter (centroid) -> Simply send in data as part of “normal” traffic • Increased the false negative rate

Summary of Attacks

Algorithm Result of Attack What does this mean?

Naïve Bayes Increased false positive rate You can make the system unusable

K-means clustering

Increased false negative rate You can evade detection

SVM Control of the decision boundary You have full control of what gets alerted and what doesn’t

Ensembling – You can’t fool ‘em all

- Build separate models to detect malicious activity

- The models are chosen so that they are orthogonal

- Each model independently assess for maliciousness

- Results are combining using a separate function

• Used Gaussian Naïve Bayes, linear SVM in addition to Naïve Bayes• Used a simple majority voting method, to combine the three outputs.

Using Robust Learning Methods• Intuition: Treat the tainted data points as

outliers (presumably because of noise)

Outlier?

Instead of Consider

Vanilla Naïve Bayes Multinomial Model (even better than multivariate Bernoulli model)

SVM Robust SVM (feature noise, and label noise)

K-means with finite window K-means with infinite window

Logistic Regression Robust Logistic Regression using Shift Parameters

Vanilla PCA Robust PCA with Laplcian Threshold (Antidote)

Caution!

• Pros: Well studied field with a gamut of choices• Optimization perspective • Game Theoretic perspective• Statistical perspective

• Cons: • Some of these algorithms have higher computational complexity than standard algorithms

Standard SVM: 10 minutes Robust SVM: 1 hr and 8 mins(Single node implementation, 50k data points, 20% test, no kernel )

• Requires a lot more tuning and babysitting

Agenda

• Motivation to Attack ML systems • Practical Attacks and Defenses• Best Practices

Threat Modeling

• Adversary Goal - Evasion? Poisoning? Deletion? • Adversary’s knowledge – Perfect Knowledge? Limited Knowledge?• Training set or part of it• Feature representation of each sample • Type of a learning algorithm and the form of its decision function• Parameters and hyper-parameters of the learned model • Feedback from the classifier; e.g., classifier labels for samples chosen by the

adversary.

• Attacker’s capability• Ability to modify – Complete or partial?

Source:Biggio, Battista, Blaine Nelson, and Pavel Laskov. "Poisoning attacks against support vector machines." arXiv preprint arXiv:1206.6389 (2012).

Tablestakes

• Secure log sources • Secure your storage space •Monitor data quality • Treat parameters and features as secrets • Don’t use publically available datasets to train your system •When designing the system, avoid interactive feedback

3 Key Takeaways

1) Naïve implementation of machine Learning Algorithms are vulnerable to attacks.

2) Attackers can evade detections, cause the system to be unusable or even control it.

3) Trustworthy results depend on trustworthy data.

Source: http://www.lecun.org/gallery/libpro/20011121-allyourbayes/dsc01228-02-h.jpg

Thank you! - TwC: Tim Burell - Azure Security: Ross Snider,

Shrikant Adhirkala, Sacha Faust Bourque, Bryan Smith, Marcin Olszewski, Ashish Kurmi, Lars Mohr, Ben Ridgway

- O365 Security: Dave Hull, Chetan Bhat, Jerry Cochran

- MSR: Jay Stokes, Gang Wang (intern)

- LCA: Matt Sommer