10 papers
Prompt Compression via Activation Aggregation
Thibaud Ardoin, Semira Einsele, Evis Bregu +1
Large language models process prompts by propagating activations through dozens of layers before generating a response. We ask whether the task-relevant information contained in an…
Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings
Jonas Schäfer, Cezary Pilaszewicz, Gerhard Wunder
This work presents Dual-Embedding Watermarking (DEW), a semantic watermarking scheme for large language models (LLMs) that leverages contextual and token-level embeddings to enhanc…
LLM Self-Recognition: Steering and Retrieving Activation Signatures
Thibaud Ardoin, Jonas Schäfer, Gerhard Wunder
Recent advances in interpretability suggest that large language models (LLMs) implicitly encode signals in their generated text that enable self-recognition of their outputs. We de…
ALIGN-FL: Architecture-independent Learning through Invariant Generative component sharing in Federated Learning
Mayank Gulati, Benedikt GroÃ, Gerhard Wunder
We present ALIGN-FL, a novel approach to distributed learning that addresses the challenge of learning from highly disjoint data distributions through selective sharing of generati…
Rethinking Explanation Evaluation under the Retraining Scheme
Yi Cai, Thibaud Ardoin, Mayank Gulati +1
Feature attribution has gained prominence as a tool for explaining model decisions, yet evaluating explanation quality remains challenging due to the absence of ground-truth explan…
Machine and Deep Learning for Indoor UWB Jammer Localization
Hamed Fard, Mahsa Kholghi, Benedikt Groà +1
Ultra-wideband (UWB) localization delivers centimeter-scale accuracy but is vulnerable to jamming attacks, creating security risks for asset tracking and intrusion detection in sma…