Defenses Against Adversarial Attacks - Revision history

BPeat at 19:12, 16 September 2023

2023-09-16T19:12:42Z

BPeat at 19:09, 16 September 2023

2023-09-16T19:09:56Z

BPeat: Text replacement - "http:" to "https:"

2023-03-28T13:26:04Z

Text replacement - "http:" to "https:"

BPeat at 21:29, 14 September 2020

2020-09-14T21:29:17Z

BPeat: BPeat moved page Defenses Against Adversarial Examples for Deep Neural Networks to Defenses Against Adversarial Attacks without leaving a redirect

2019-10-02T02:27:27Z

BPeat moved page Defenses Against Adversarial Examples for Deep Neural Networks to Defenses Against Adversarial Attacks without leaving a redirect

BPeat at 02:26, 2 October 2019

2019-10-02T02:26:06Z

BPeat at 02:20, 2 October 2019

2019-10-02T02:20:28Z

BPeat at 01:13, 21 May 2019

2019-05-21T01:13:35Z

BPeat at 23:19, 2 February 2019

2019-02-02T23:19:49Z

BPeat at 12:59, 29 July 2018

2018-07-29T12:59:09Z

@@ Line 20: / Line 20: @@
 <youtube>sh6OS6Lssv4</youtube>
-Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy! [https://colab.research.google.com/drive/19N9VWTukXTPUj9eukeie55XIu3HKR5TT Defense Against Adversarial Attacks |] [[Creatives#Siraj Raval|Siraj Raval]]
+Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy!
 <youtube>JjlV62_kGUc</youtube>

@@ Line 25: / Line 25: @@
 == MagNet ==
-* MagNet includes one or more separate detector networks and a reformer network. The detector networks learn to differentiate between normal and adversarial examples by approximating the manifold of normal examples. Since they assume no specific process for generating adversarial examples, they generalize well. The reformer network moves adversarial examples towards the manifold of normal examples, which is effective for correctly classifying adversarial examples with small perturbation.  We discuss the intrinsic difficulties in defending against whitebox attack and propose a mechanism to defend against graybox attack. Inspired by the use of randomness in cryptography, we use diversity to strengthen MagNet. We show empirically that MagNet is effective against the most advanced state-of-the-art attacks in blackbox and graybox scenarios without sacrificing false positive rate on normal examples.[https://arxiv.org/pdf/1705.09064.pdf MagNet: a Two-Pronged Defense against Adversarial Examples | Meng, D., Chen, H.], 11 Sep 2017
+* MagNet includes one or more separate detector networks and a reformer network. The detector networks learn to differentiate between normal and adversarial examples by approximating the [[Manifold Hypothesis|manifold]] of normal examples. Since they assume no specific process for generating adversarial examples, they generalize well. The reformer network moves adversarial examples towards the [[Manifold Hypothesis|manifold]] of normal examples, which is effective for correctly classifying adversarial examples with small perturbation.  We discuss the intrinsic difficulties in defending against whitebox attack and propose a mechanism to defend against graybox attack. Inspired by the use of randomness in cryptography, we use diversity to strengthen MagNet. We show empirically that MagNet is effective against the most advanced state-of-the-art attacks in blackbox and graybox scenarios without sacrificing false positive rate on normal examples.[https://arxiv.org/pdf/1705.09064.pdf MagNet: a Two-Pronged Defense against Adversarial Examples | Meng, D., Chen, H.], 11 Sep 2017
 * [https://web.cs.ucdavis.edu/~hchen/paper/meng2017-slides.pdf MagNet: a Two-Pronged Defense against Adversarial Examples - Presentation | Meng, D., Chen, H.]

@@ Line 5: / Line 5: @@
 |description=Helpful resources for your journey with artificial intelligence; videos, articles, techniques, courses, profiles, and tools
 }}
-[http://www.youtube.com/results?search_query=Adversarial+threat+attack+Deep+Learning+Artificial+Intelligence Youtube search...]
+[https://www.youtube.com/results?search_query=Adversarial+threat+attack+Deep+Learning+Artificial+Intelligence Youtube search...]
-[http://www.google.com/search?q=Adversarial+threat+attack+deep+machine+learning+ML+artificial+intelligence ...Google search]
+[https://www.google.com/search?q=Adversarial+threat+attack+deep+machine+learning+ML+artificial+intelligence ...Google search]
 * [[Cybersecurity]]
 ** [[Cybersecurity References]]
 ** [[Defense]]
-*** [http://gcn.com/articles/2019/01/25/darpa-adversarial-ai-defense.aspx DARPA outlines adversarial AI defense | Susan Miller - GCN]
+*** [https://gcn.com/articles/2019/01/25/darpa-adversarial-ai-defense.aspx DARPA outlines adversarial AI defense | Susan Miller - GCN]
 ** [[Offense - Adversarial Threats/Attacks]]
 ______________________________________________________
-* A method based on a semidefinite relaxation that outputs a certificate that for a given network and test input, no attack can force the error to exceed a certain value. Second, as this certificate is differentiable, we jointly optimize it with the network parameters, providing an adaptive regularizer that encourages robustness against all attacks. On MNIST, our approach produces a network and a certificate that no attack that perturbs each pixel by at most � = 0.1 can cause more than 35 percent test error.[http://arxiv.org/pdf/1801.09344.pdf Certified Defenses against Adversarial Examples | Raghunathan, A., Steinhardt, J., Liang, P.], 29 Jan 2018
+* A method based on a semidefinite relaxation that outputs a certificate that for a given network and test input, no attack can force the error to exceed a certain value. Second, as this certificate is differentiable, we jointly optimize it with the network parameters, providing an adaptive regularizer that encourages robustness against all attacks. On MNIST, our approach produces a network and a certificate that no attack that perturbs each pixel by at most � = 0.1 can cause more than 35 percent test error.[https://arxiv.org/pdf/1801.09344.pdf Certified Defenses against Adversarial Examples | Raghunathan, A., Steinhardt, J., Liang, P.], 29 Jan 2018
 <youtube>sh6OS6Lssv4</youtube>
-Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy! [http://colab.research.google.com/drive/19N9VWTukXTPUj9eukeie55XIu3HKR5TT Defense Against Adversarial Attacks |] [[Creatives#Siraj Raval|Siraj Raval]]
+Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy! [https://colab.research.google.com/drive/19N9VWTukXTPUj9eukeie55XIu3HKR5TT Defense Against Adversarial Attacks |] [[Creatives#Siraj Raval|Siraj Raval]]
 <youtube>JjlV62_kGUc</youtube>
 == MagNet ==
-* MagNet includes one or more separate detector networks and a reformer network. The detector networks learn to differentiate between normal and adversarial examples by approximating the manifold of normal examples. Since they assume no specific process for generating adversarial examples, they generalize well. The reformer network moves adversarial examples towards the manifold of normal examples, which is effective for correctly classifying adversarial examples with small perturbation.  We discuss the intrinsic difficulties in defending against whitebox attack and propose a mechanism to defend against graybox attack. Inspired by the use of randomness in cryptography, we use diversity to strengthen MagNet. We show empirically that MagNet is effective against the most advanced state-of-the-art attacks in blackbox and graybox scenarios without sacrificing false positive rate on normal examples.[http://arxiv.org/pdf/1705.09064.pdf MagNet: a Two-Pronged Defense against Adversarial Examples | Meng, D., Chen, H.], 11 Sep 2017
+* MagNet includes one or more separate detector networks and a reformer network. The detector networks learn to differentiate between normal and adversarial examples by approximating the manifold of normal examples. Since they assume no specific process for generating adversarial examples, they generalize well. The reformer network moves adversarial examples towards the manifold of normal examples, which is effective for correctly classifying adversarial examples with small perturbation.  We discuss the intrinsic difficulties in defending against whitebox attack and propose a mechanism to defend against graybox attack. Inspired by the use of randomness in cryptography, we use diversity to strengthen MagNet. We show empirically that MagNet is effective against the most advanced state-of-the-art attacks in blackbox and graybox scenarios without sacrificing false positive rate on normal examples.[https://arxiv.org/pdf/1705.09064.pdf MagNet: a Two-Pronged Defense against Adversarial Examples | Meng, D., Chen, H.], 11 Sep 2017
-* [http://web.cs.ucdavis.edu/~hchen/paper/meng2017-slides.pdf MagNet: a Two-Pronged Defense against Adversarial Examples - Presentation | Meng, D., Chen, H.]
+* [https://web.cs.ucdavis.edu/~hchen/paper/meng2017-slides.pdf MagNet: a Two-Pronged Defense against Adversarial Examples - Presentation | Meng, D., Chen, H.]
 <youtube>wZ-wIdAcWQE</youtube>
-* ...we show that adversarial examples crafted based on the L1 distortion metric can easily bypass MagNet [http://arxiv.org/pdf/1805.00310.pdf On the Limitation of MagNet Defense against L1-based Adversarial Examples | Lu, P., Chen, P., Chen, K., Yu, C.], 9 May 2018
+* ...we show that adversarial examples crafted based on the L1 distortion metric can easily bypass MagNet [https://arxiv.org/pdf/1805.00310.pdf On the Limitation of MagNet Defense against L1-based Adversarial Examples | Lu, P., Chen, P., Chen, K., Yu, C.], 9 May 2018
-* MagNet and "Efficient Defenses..." were recently proposed as a defense to adversarial examples. We find that we can construct adversarial examples that defeat these defenses with only a slight increase in distortion. [http://arxiv.org/pdf/1711.08478.pdf MagNet and "Efficient Defenses Against Adversarial Attacks" are Not Robust to Adversarial Examples | Carlini, N., Wagner, D.], 22 Nov 2017
+* MagNet and "Efficient Defenses..." were recently proposed as a defense to adversarial examples. We find that we can construct adversarial examples that defeat these defenses with only a slight increase in distortion. [https://arxiv.org/pdf/1711.08478.pdf MagNet and "Efficient Defenses Against Adversarial Attacks" are Not Robust to Adversarial Examples | Carlini, N., Wagner, D.], 22 Nov 2017

@@ Line 20: / Line 20: @@
 <youtube>sh6OS6Lssv4</youtube>
-Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy! [http://colab.research.google.com/drive/19N9VWTukXTPUj9eukeie55XIu3HKR5TT Defense Against Adversarial Attacks | Siraj Raval]
+Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy! [http://colab.research.google.com/drive/19N9VWTukXTPUj9eukeie55XIu3HKR5TT Defense Against Adversarial Attacks |] [[Creatives#Siraj Raval|Siraj Raval]]
 <youtube>JjlV62_kGUc</youtube>

← Older revision		Revision as of 02:26, 2 October 2019
Line 19:		Line 19:

	<youtube>sh6OS6Lssv4</youtube>		<youtube>sh6OS6Lssv4</youtube>
		+
		+	Machine Learning technology isn't perfect, it's vulnerable to many different types of attacks! In this episode, I'll explain 2 common types of attacks and 2 common types of defenses using various code demos from across the Web. There's some really dope mathematics involved with adversarial attacks, and it was a lot of fun reading about the 'cat and mouse' game between new attack techniques, followed by new defense techniques. I encourage anyone new to the field who finds this stuff interesting to learn more about it. I definitely plan to. Let's look into some math, code, and examples. Enjoy! [http://colab.research.google.com/drive/19N9VWTukXTPUj9eukeie55XIu3HKR5TT Defense Against Adversarial Attacks \| Siraj Raval]
		+
	<youtube>JjlV62_kGUc</youtube>		<youtube>JjlV62_kGUc</youtube>

← Older revision	Revision as of 02:27, 2 October 2019
(No difference)

← Older revision		Revision as of 23:19, 2 February 2019
Line 1:		Line 1:
		+	{{#seo:
		+	\|title=PRIMO.ai
		+	\|titlemode=append
		+	\|keywords=artificial, intelligence, machine, learning, models, algorithms, data, singularity, moonshot, Tensorflow, Google, Nvidia, Microsoft, Azure, Amazon, AWS
		+	\|description=Helpful resources for your journey with artificial intelligence; videos, articles, techniques, courses, profiles, and tools
		+	}}
	[http://www.youtube.com/results?search_query=Adversarial+threat+attack+Deep+Learning+Artificial+Intelligence Youtube search...]		[http://www.youtube.com/results?search_query=Adversarial+threat+attack+Deep+Learning+Artificial+Intelligence Youtube search...]
		+	[http://www.google.com/search?q=Adversarial+threat+attack+deep+machine+learning+ML+artificial+intelligence ...Google search]

	* [[Cybersecurity]]		* [[Cybersecurity]]