Global ETD Search

Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.

Results

Showing 1 to 20 of 101 for “"Backdoor"”.

  1. Backdoor detection systems for embedded devices

    A system is said to contain a backdoor when it intentionally includes a means to trigger the execution of functionality that serves to subvert its expected security. Unfortunately, such constructs are pervasive in software and systems today, particularly in the firmware of commodity embedded …

    birmingham Download PDF for Backdoor detection systems for embedded devices (opens in a new tab)

  2. Flow-Based Detection of Linux Backdoor Communication - A NetFlow Based ML-Approach to Backdoor Detection in Linux Environments

    … necessitates effective detection mechanisms for backdoor communication. This thesis explores the application of machine learning (ML) models to detect backdoor communication in Linux environments using flow-based data. Specifically, it leverages NetFlow traffic data. The study aims to determine …

    chalmers Repository record for Flow-Based Detection of Linux Backdoor Communication - A NetFlow Based ML-Approach to Backdoor Detection in Linux Environments (opens in a new tab)

  3. Backdoor in AI: Algorithms, Attacks, and Defenses

    … Among the various threats to these systems, backdoor attacks pose a particularly insidious challenge. These attacks compromise the model by embedding a hidden backdoor function, which can be triggered by specific inputs to manipulate the model's behavior. My research goal initially involves …

    rice Repository record for Backdoor in AI: Algorithms, Attacks, and Defenses (opens in a new tab)

  4. Backdoor listing in Australia: an exploratory analysis

    This thesis examines the role of backdoor listing transactions (BDL) as an alternative route for private firms to go public in the Australian market. In addition, it examines the motivations and performance of firms engaging in such activities. Based on a hand-collected sample of 200 backdoor …

    unsw Repository record for Backdoor listing in Australia: an exploratory analysis (opens in a new tab)

  5. Towards More Realistic and Practical Graph Backdoor Attacks

    … shown that they are highly vulnerable to graph backdoor attacks (GBAs), where adversaries implant malicious triggers to induce targeted misclassification during inference. Despite their effectiveness, existing GBAs are often developed under unrealistic assumptions, such as focusing exclusively …

    odu Repository record for Towards More Realistic and Practical Graph Backdoor Attacks (opens in a new tab)

  6. Synoptic investigation of east coast backdoor cold fronts

    Results are presented from a synoptic study of "back-door" cold fronts along the east coast of the united States in the years 1963 to 1972. The investigation set forth the climatology of back-door cold fronts with respect to: 1) frequency, 2) six hourly speed of movement, and 3) southern …

    nps Repository record for Synoptic investigation of east coast backdoor cold fronts (opens in a new tab)

  7. Training Phase Specific Backdoor Defences For Federated Learning

    … devices. Such manipulations can lead to backdoor attacks that embed undesirable behaviors into the joint model. This serious security issue needs to be addressed before FL systems can be deployed into critical infrastructure. This thesis proposes strategies to protect FL systems from …

    unsw Repository record for Training Phase Specific Backdoor Defences For Federated Learning (opens in a new tab)

  8. Un-fair trojan: Targeted backdoor attacks against model fairness

    … have been shown to be vulnerable against various backdoor and data poisoning attacks that adversely affect model behavior. Additionally, these attacks have been shown to make unfair predictions with respect to certain protected features. In federated learning, multiple local models contribute to a …

    njit Repository record for Un-fair trojan: Targeted backdoor attacks against model fairness (opens in a new tab)

  9. TextGuard: Provable defense against backdoor attacks on text classification

    Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2025-05-01

    uiuc Repository record for TextGuard: Provable defense against backdoor attacks on text classification (opens in a new tab)

  10. Certified and Forensic Defenses against Poisoning and Backdoor Attacks

    Data poisoning and backdoor attacks manipulate model predictions by inserting malicious instances into the training set. Most existing defenses against poisoning and backdoor attacks are empirical and easily evaded by an adaptive attacker. In addition, existing empirical defenses provide, at best, …

    oregon Repository record for Certified and Forensic Defenses against Poisoning and Backdoor Attacks (opens in a new tab)

  11. Enhancing Robustness of Graph Neural Networks Against Evasive Backdoor Attacks

    … manipulation has become a key concern. Backdoor attacks pose a significant security risk, as they can introduce subtle vulnerabilities during training that may later compromise model predictions in deployment. This thesis investigates methods for defending against such attacks through …

    brock Repository record for Enhancing Robustness of Graph Neural Networks Against Evasive Backdoor Attacks (opens in a new tab)

  12. Backdoor To Essentialism? Genetic Ancestry Testing And The Social Deconstruction Of Whiteness

    Towards the turn of the 21st century, spawned in part by the Human Genome Project, ideas about race and ethnicity shifted away from the essentialist belief that humans can be grouped into discrete, biologically relevant racial groups. More recently however, genetic ancestry testing (GAT) has …

    wayne-thes Repository record for Backdoor To Essentialism? Genetic Ancestry Testing And The Social Deconstruction Of Whiteness (opens in a new tab)

  13. Investigation of Backdoor Attacks and Design of Effective Countermeasures in Federated Learning

    … to adversarial attacks. These include backdoor attacks, which subtly alter classification in some categories, and byzantine attacks, aimed at degrading the overall model accuracy. Detecting and defending against such attacks is challenging, as adversaries can participate in the system, …

    iupui Repository record for Investigation of Backdoor Attacks and Design of Effective Countermeasures in Federated Learning (opens in a new tab)

  14. The Dual Game of Clean-Label Backdoor Attacks and Defenses in NLP

    Textual backdoor attacks are a growing security threat for natural language processing (NLP) systems. These attacks compromise language models by inserting specific ``\textit{triggers}'' into the training data. At inference time, the model misclassifies inputs that contain the trigger, …

    oregon Repository record for The Dual Game of Clean-Label Backdoor Attacks and Defenses in NLP (opens in a new tab)

  15. SEDPD: Sampling-Enhanced Differentially Private Defense against Backdoor Poisoning Attacks of Image Classification

    … DP guarantees, enhancing robustness against backdoor poisoning attacks with XAI. Results from extensive experiments on widely used datasets, such as CIFAR-10, MNIST, CIFAR-100 and Imagenette, and model architectures, including ConvMixer and ResNet-50, show that our approach effectively …

    ku Repository record for SEDPD: Sampling-Enhanced Differentially Private Defense against Backdoor Poisoning Attacks of Image Classification (opens in a new tab)

  16. Backdoor Censorship: How Media Giants' Copyright Protection Policies Collide with the First Amendment

    This thesis analyzes an alarming trend among large media companies to use copyright laws to censor news and news-related websites using takedown notices, cease-and-desist letters and other legal threats. The paper examines the origins of takedown problem and the copyright, fair use and First …

    ku Repository record for Backdoor Censorship: How Media Giants' Copyright Protection Policies Collide with the First Amendment (opens in a new tab)

  17. A Comprehensive Study on Printed Circuit Board Backdoor Coupling in High Intensity Radiated Fields Environments

    … This study provides unique insight into the backdoor coupling mechanisms associated with printed circuit boards (PCBs) as well as design techniques for reducing electromagnetic coupling in HIRF environments. Among existing literature, there is very little quantification of PCB coupling …

    lsu-thes Repository record for A Comprehensive Study on Printed Circuit Board Backdoor Coupling in High Intensity Radiated Fields Environments (opens in a new tab)

Page 1 of 6