About this position

Background/Motivation: Backdoor attacks are attacks on neural networks where a so-called trigger alters the decision-making behaviour of the networks, thereby creating vulnerabilities. These triggers can be injected into the training dataset or directly into the model weights. These are then called poisoned. Due to parameter-efficient fine-tuning methods, backdoor attacks on large language models (LLMs) have become s…

This is a short preview. Read the full posting on the source site. View Original ↗

View Full Posting & Apply

Opens the original posting on the institution’s portal

Similar positions

Other openings you might be interested in

Fraunhofer SIT · SIT - Secure Information Technology

Student / Thesis

Master's Thesis: Detection of Suggestive Poses and Facial Expressions

Background/Motivation: For the detection of erotic and pornographic images, models that can recognise human skin, body parts, or scenes are often used. With the help of appropriate datasets [1], classification and object detection models can be trained. However, there are also images that are obviously erotic or pornographic, but cannot be recognised by conventional methods. This applies, for example, to images in which erotic poses or partial sections of pornographic scenes can be recognised. Here, poses and facial expressions can be used for classification, among other things [2].…

🇩🇪 Darmstadt

No deadlineSource: Fraunhofer SIT

Fraunhofer SIT · SIT - Secure Information Technology

Student / Thesis

Master's Thesis: Detection and Segmentation of Suggestive Clothing

Background/Motivation: Models that can recognise human skin, body parts, or scenes are often used to detect erotic and pornographic material. With the help of appropriate datasets [1], classification and object detection models can be trained. However, there are also images that are obviously erotic or pornographic, but cannot be recognised by conventional methods. This applies, for example, to people in skin-tight latex or leather clothing. Existing approaches in the field of "Human Parsing" can already segment people and their clothing well. Additionally, datasets like Fashionpedia [2]…

🇩🇪 Darmstadt

No deadlineSource: Fraunhofer SIT

Fraunhofer SIT · SIT - Secure Information Technology

Student / Thesis

Master's Thesis: Explainability of Transformer Models in Authorship Verification

Background/Motivation: Authorship verification (AV) is used in areas such as forensics, plagiarism detection, and fake news detection to identify the true author of a text. The goal of authorship verification (AV) is to classify whether two or more texts were written by the same author (Y) or not (N). As in most AI fields today, the most powerful models are usually based on transformer architectures. While these continuously achieve new state-of-the-art results, the application of explainability methods is mostly limited to somewhat older models or architectures. This limits the applicability…

🇩🇪 Darmstadt

No deadlineSource: Fraunhofer SIT

Fraunhofer SIT · SIT - Sichere Informationstechnologie

Student / Thesis

Studentische Hilfskraft im Bereich Weiterbildungskurse in IT-Forensik und KI-Sicherheit

Im Kompetenzfeld „IT Forensics“ erforschen und entwickeln wir Methoden zur digitalen Spurensuche in IT‑Systemen und im Internet zu verschiedenen Fachthemen: Datenträgeranalyse unter Linux und Windows (Dateisysteme bspw. ext[234]/ NTFS / FAT) Erkennen von Bild-/Video-Manipulationen (z.B. Deepfakes) mittels Deeplearning-Classifiers und traditioneller Multimediaforensik Robuste Hashfunktionen zum Wiedererkennen von Bildern OSINT und Crawling, Social Media und Darknet / Kryptowährungen Machine Learning für Datenanalysen KI-Sicherheit: Angriffe auf Deeplearning (Classifiers, LLMs) und…

🇩🇪 Darmstadt

No deadlineSource: Fraunhofer SIT

Master's Thesis: Attacking Current Backdoor Detection Methods

About this position

AI Application Kit

Position Details

Similar positions

Master's Thesis: Detection of Suggestive Poses and Facial Expressions

Master's Thesis: Detection and Segmentation of Suggestive Clothing

Master's Thesis: Explainability of Transformer Models in Authorship Verification

Location

Master's Thesis: Localisation of Safety Neurones in Text2Image Models

Masterarbeit: Aktuelle Backdoor Detektionsverfahren angreifen

Studentische Hilfskraft im Bereich Weiterbildungskurse in IT-Forensik und KI-Sicherheit