Robust Text CAPTCHAs Using Adversarial Examples

Shao, Rulin; Shi, Zhouxing; Yi, Jinfeng; Chen, Pin-Yu; Hsieh, Cho‐Jui

doi:10.48550/arxiv.2101.02483

Cited by 4 publications

(8 citation statements)

References 33 publications

(50 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2021, a text-based CAPTCHA generation technique named Robust Text CAPTCHA (RTC) is proposed in [16]. The evaluation results showed that the proposed method has a high usability level and robust level against different defensive techniques.…”

Section: Techniques For Generating Adversarial Captchamentioning

confidence: 99%

“…CAPTCHA Scheme Perturbations Domain [3] Image-based Space [14] Image-based Space Text-based [15] Text-based Space [13] Image-based Space Text-based Click-based [4] Text-based Space [5] Image-based Space [18] Text-based Frequency [6] Text-based Space [16] Text-based Frequency [17] Text-image-based Space [18] Image-based Space [37] Audio-based Space [41] Image-based Frequency [38] Audio-based Space…”

Section: Studymentioning

confidence: 99%

“…SGTCS [16] Its robustness level is empirically evaluated in [16] and shows a sufficient resistance level. Also, no attack has been reported yet.…”

Section: Perturbation Techniques Recommendation Remarksmentioning

confidence: 99%

“…Most of these works utilize a deep neural network (DNN) model to directly recognize CAPTCHA characters without the segmentation step. To defend against them, several technologies have been proposed, such as in [3][4][5][6][14][15][16][17][18].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A Survey on Adversarial Perturbations and Attacks on CAPTCHAs

Alsuhibany

2023

Applied Sciences

View full text Add to dashboard Cite

The Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHA) technique has been a topic of interest for several years. The ability of computers to recognize CAPTCHA has significantly increased due to the development of deep learning techniques. To prevent this ability from being utilised, adversarial machine learning has recently been proposed by perturbing CAPTCHA images. As a result of the introduction of various removal methods, this perturbation mechanism can be removed. This paper, thus, presents the first comprehensive survey on adversarial perturbations and attacks on CAPTCHAs. In particular, the art of utilizing deep learning techniques with the aim of breaking CAPTCHAs are reviewed, and the effectiveness of adversarial CAPTCHAs is discussed. Drawing on the reviewed literature, several observations are provided as part of a broader outlook of this research direction. To emphasise adversarial CAPTCHAs as a potential solution for current attacks, a set of perturbation techniques have been suggested for application in adversarial CAPTCHAs.

show abstract

Section: Techniques For Generating Adversarial Captchamentioning

confidence: 99%

Section: Studymentioning

confidence: 99%

“…SGTCS [16] Its robustness level is empirically evaluated in [16] and shows a sufficient resistance level. Also, no attack has been reported yet.…”

Section: Perturbation Techniques Recommendation Remarksmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Survey on Adversarial Perturbations and Attacks on CAPTCHAs

Alsuhibany

2023

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…The work presented in this paper is related to defeating text-CAPTCHA schemes. As they are widely applicable across the internet, both in the commercial and non-commercial sectors [14,30]. Therefore, it is important to continue researching the security strengths of this kind of CAPTCHA scheme and bring awareness of its weaknesses to stakeholders.…”

Section: Related Workmentioning

confidence: 99%

Multiview deep learning-based attack to break text-CAPTCHAs

Yusuf

Srivastava

Singh

et al. 2022

Int. J. Mach. Learn. & Cyber.

View full text Add to dashboard Cite

Completely Automated Public Turing Test To Tell Computer and Humans Apart (CAPTCHA) is a computer program that prevents malicious computer users. Text-CAPTCHA schemes utilize less-computational costs. Hence, they are the most popularly used. This paper investigates the effectiveness of state-of-the-art (SOTA) text-CAPTCHA schemes, proposes a Multiview deep learning system to break them, and highlights their weaknesses. Rather than the usual single-view feature extraction, the proposed model explores correlational features from multiple views to increase the model’s generalization and classification accuracy. The model combines convolutional neural networks and recurrent networks to preserve the input text-CAPTCHA’s spatial and sequential order. The proposed system has successfully achieved average accuracies ranging from 93.6% to 100%, and the average time to break a text-CAPTCHA scheme ranges from 0.0032 to 0.21 seconds on eight different datasets. Furthermore, an ablation study on 71 human users was conducted to evaluate the effectiveness of the schemes. The results demonstrated that the proposed system effectively outperforms the human users whom the schemes were designed to serve. Lastly, when compared with existing systems, the proposed system outperforms existing SOTA systems with an accuracy gap of almost 40% higher.

show abstract

Defending Emotional Privacy with Adversarial Machine Learning for Social Good

Al-Maliki

Abdallah

Qadir

et al. 2023

2023 International Wireless Communications and Mobile Computing (IWCMC)

View full text Add to dashboard Cite

Protecting the privacy of personal information, including emotions, is essential, and organizations must comply with relevant regulations to ensure privacy. Unfortunately, some organizations do not respect these regulations, or they lack transparency, leaving human privacy at risk. These privacy violations often occur when unauthorized organizations misuse machine learning (ML) technology, such as facial expression recognition (FER) systems. Therefore, researchers and practitioners must take action and use ML technology for social good to protect human privacy. One emerging research area that can help address privacy violations is the use of adversarial ML for social good. Evasion attacks, which are used to fool ML systems, can be repurposed to prevent misused ML technology, such as ML-based FER, from recognizing true emotions. By leveraging adversarial ML for social good, we can prevent organizations from violating human privacy by misusing ML technology, particularly FER systems, and protect individuals' personal and emotional privacy. In this work, we propose an approach called Chaining of Adversarial ML Attacks (CAA) to create a robust attack that fools misused technology and prevents it from detecting true emotions. To validate our proposed approach, we conduct extensive experiments using various evaluation metrics and baselines. Our results show that CAA significantly contributes to emotional privacy preservation, with the fool rate of emotions increasing proportionally to the chaining length. In our experiments, the fool rate increases by 48% in each subsequent chaining stage of the chaining targeted attacks (CTA) while keeping the perturbations imperceptible (ϵ = 0.0001).

show abstract

Robust Text CAPTCHAs Using Adversarial Examples

Cited by 4 publications

References 33 publications

A Survey on Adversarial Perturbations and Attacks on CAPTCHAs

A Survey on Adversarial Perturbations and Attacks on CAPTCHAs

Multiview deep learning-based attack to break text-CAPTCHAs

Defending Emotional Privacy with Adversarial Machine Learning for Social Good

Contact Info

Product

Resources

About