國立虎尾科技大學 |

Promises and Pitfalls of Generative AI : = An AI-Safety Centric Approach.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Promises and Pitfalls of Generative AI :/
Reminder of title:	An AI-Safety Centric Approach.
Author:	Sehwag, Vikash.
Description:	1 online resource (171 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 84-12, Section: B.
Contained By:	Dissertations Abstracts International84-12B.
Subject:	Computer engineering. -
Online resource:	click for full text (PQDT)
ISBN:	9798379717469

Promises and Pitfalls of Generative AI : = An AI-Safety Centric Approach.
Sehwag, Vikash.

Promises and Pitfalls of Generative AI :An AI-Safety Centric Approach. - 1 online resource (171 pages)

Source: Dissertations Abstracts International, Volume: 84-12, Section: B.

Thesis (Ph.D.)--Princeton University, 2023.

Includes bibliographical references

Artificial intelligence (AI) has advanced rapidly, leading to remarkable progress across numerous real-world applications. However, the prevalence of AI-enabled decisions also raises concerns about its potential safety risks, as AI systems are known to exhibit failure cases across multiple domains, such as autonomous driving, medical diagnostics, and content moderation. In this thesis, we investigate AI safety challenges through the lens of generative models, a class of machine learning models capable of approximating the underlying distribution of training datasets and synthesizing novel samples. By bridging the gap between generative models and AI safety, we reveal the immense potential of generative models in addressing safety challenges, while also identifying safety risks posed by contemporary generative models.First, we focus on improving generalization in adversarially robust learning with generative models by incorporating them into existing machine learning pipelines and distilling their knowledge by synthesizing novel synthetic images. We assess various generative models and propose a new metric (ARC), based on the indistinguishability of adversarially perturbed synthetic and real data, to accurately determine the generalization benefit of different generative models. Next, we investigate task-aware knowledge distillation from generative models, where we first demonstrate the disparate contributions of individual synthetic images in improving generalization. To adaptively sample images with the highest generalization benefit, we propose an adaptive sampling technique that guides the sampling process in diffusion models to maximize the generalization benefit of generated synthetic images.Next, we address the shortcomings of long-tailed data distributions, which underlie numerous challenges in AI safety, by using generative models to generate high-fidelity samples from low-density regions. We propose a novel low-density sampling process for diffusion models, guiding the process towards low-density regions while maintaining fidelity, and rigorously demonstrate that our process successfully generates novel high-fidelity samples from low-density regions.Finally, we demonstrate some of the key limitations of existing generative models. We first consider the outlier detection task and demonstrate the shortcomings of modern generative models in solving it. Considering our findings, we propose SSD, an unsupervised framework for outlier detection based on unlabeled in-distribution data. We further uncover that modern diffusion models, which are used by millions of users, leak the privacy of training data, where we extract a nontrivial number of training images from the pre-trained diffusion models.In summary, this thesis addresses multiple AI safety challenges and provides a comprehensive framework for the safety and reliability of AI systems under the new generative AI paradigm.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798379717469Subjects--Topical Terms:

569006
Computer engineering.
Subjects--Index Terms:

Adversarial robustnessIndex Terms--Genre/Form:

554714
Electronic books.

Promises and Pitfalls of Generative AI : = An AI-Safety Centric Approach.
LDR:04374ntm a2200409K 4500 001 1141143
005 20231130111624.5
006 m o d
007 cr mn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798379717469
035 $a (MiAaPQ)AAI30491778
035 $a AAI30491778
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Sehwag, Vikash. $3 1464765
245 1 0 $a Promises and Pitfalls of Generative AI : $b An AI-Safety Centric Approach.
264 0 $c 2023
300 $a 1 online resource (171 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 84-12, Section: B.
500 $a Advisor: Mittal, Prateek ; Chiang, Mung.
502 $a Thesis (Ph.D.)--Princeton University, 2023.
504 $a Includes bibliographical references
520 $a Artificial intelligence (AI) has advanced rapidly, leading to remarkable progress across numerous real-world applications. However, the prevalence of AI-enabled decisions also raises concerns about its potential safety risks, as AI systems are known to exhibit failure cases across multiple domains, such as autonomous driving, medical diagnostics, and content moderation. In this thesis, we investigate AI safety challenges through the lens of generative models, a class of machine learning models capable of approximating the underlying distribution of training datasets and synthesizing novel samples. By bridging the gap between generative models and AI safety, we reveal the immense potential of generative models in addressing safety challenges, while also identifying safety risks posed by contemporary generative models.First, we focus on improving generalization in adversarially robust learning with generative models by incorporating them into existing machine learning pipelines and distilling their knowledge by synthesizing novel synthetic images. We assess various generative models and propose a new metric (ARC), based on the indistinguishability of adversarially perturbed synthetic and real data, to accurately determine the generalization benefit of different generative models. Next, we investigate task-aware knowledge distillation from generative models, where we first demonstrate the disparate contributions of individual synthetic images in improving generalization. To adaptively sample images with the highest generalization benefit, we propose an adaptive sampling technique that guides the sampling process in diffusion models to maximize the generalization benefit of generated synthetic images.Next, we address the shortcomings of long-tailed data distributions, which underlie numerous challenges in AI safety, by using generative models to generate high-fidelity samples from low-density regions. We propose a novel low-density sampling process for diffusion models, guiding the process towards low-density regions while maintaining fidelity, and rigorously demonstrate that our process successfully generates novel high-fidelity samples from low-density regions.Finally, we demonstrate some of the key limitations of existing generative models. We first consider the outlier detection task and demonstrate the shortcomings of modern generative models in solving it. Considering our findings, we propose SSD, an unsupervised framework for outlier detection based on unlabeled in-distribution data. We further uncover that modern diffusion models, which are used by millions of users, leak the privacy of training data, where we extract a nontrivial number of training images from the pre-trained diffusion models.In summary, this thesis addresses multiple AI safety challenges and provides a comprehensive framework for the safety and reliability of AI systems under the new generative AI paradigm.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Computer engineering. $3 569006
650 4 $a Electrical engineering. $3 596380
650 4 $a Computer science. $3 573171
653 $a Adversarial robustness
653 $a Deep neural networks
653 $a Generative artificial intelligence
653 $a Privacy risks
653 $a Trustworthy machine learning
655 7 $a Electronic books. $2 local $3 554714
690 $a 0464
690 $a 0544
690 $a 0984
690 $a 0800
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a Princeton University. $b Electrical and Computer Engineering. $3 1413580
773 0 $t Dissertations Abstracts International $g 84-12B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30491778 $z click for full text (PQDT)