Michael Backes
YOU?
Author Swipe
View article: Prospects of the AMT single-dish science performance for monitoring gamma-ray blazars
Prospects of the AMT single-dish science performance for monitoring gamma-ray blazars Open
Blazars make up the largest fraction of Fermi-LAT-detected sources of high-energy gamma-ray emission. At the same time, their emission is known to be highly variable throughout the electromagnetic spectrum. With hints of their highest frac…
View article: Optical spectroscopy of blazars for the Cherenkov Telescope Array Observatory – IV
Optical spectroscopy of blazars for the Cherenkov Telescope Array Observatory – IV Open
Context . Blazars, comprising BL Lacertae objects (BL Lacs) and flat-spectrum radio quasars, are the most luminous extragalactic sources that dominate the γ-ray sky. They account for approximately 56% of the sources listed in the recent Fe…
View article: Prospects for dark matter observations in dwarf spheroidal galaxies with the Cherenkov Telescope Array Observatory
Prospects for dark matter observations in dwarf spheroidal galaxies with the Cherenkov Telescope Array Observatory Open
The dwarf spheroidal galaxies (dSphs) orbiting the Milky Way are widely regarded as systems supported by velocity dispersion against self-gravity, and as prime targets for the search for indirect dark matter (DM) signatures in the GeV-to-T…
View article: Bridging the Gap in Vision Language Models in Identifying Unsafe Concepts Across Modalities
Bridging the Gap in Vision Language Models in Identifying Unsafe Concepts Across Modalities Open
Vision-language models (VLMs) are increasingly applied to identify unsafe or inappropriate images due to their internal ethical standards and powerful reasoning abilities. However, it is still unclear whether they can recognize various uns…
View article: Rethinking Data Protection in the (Generative) Artificial Intelligence Era
Rethinking Data Protection in the (Generative) Artificial Intelligence Era Open
The (generative) artificial intelligence (AI) era has profoundly reshaped the meaning and value of data. No longer confined to static content, data now permeates every stage of the AI lifecycle from the training samples that shape model pa…
View article: H.E.S.S. programme searching for VHE gamma rays associated with FRBs
H.E.S.S. programme searching for VHE gamma rays associated with FRBs Open
Fast Radio Bursts (FRBs) are highly energetic, extremely short-lived bursts of radio flashes. Despite extensive research, the exact cause of these outbursts remains speculative. The high luminosity, short duration, and high dispersion meas…
View article: Watermarking LLM-Generated Datasets in Downstream Tasks
Watermarking LLM-Generated Datasets in Downstream Tasks Open
Large Language Models (LLMs) have experienced rapid advancements, with applications spanning a wide range of fields, including sentiment classification, review generation, and question answering. Due to their efficiency and versatility, re…
View article: When GPT Spills the Tea: Comprehensive Assessment of Knowledge File Leakage in GPTs
When GPT Spills the Tea: Comprehensive Assessment of Knowledge File Leakage in GPTs Open
Knowledge files have been widely used in large language model (LLM) agents, such as GPTs, to improve response quality. However, concerns about the potential leakage of knowledge files have grown significantly. Existing studies demonstrate …
View article: Galactic transient sources with the Cherenkov Telescope Array Observatory
Galactic transient sources with the Cherenkov Telescope Array Observatory Open
A wide variety of Galactic sources show transient emission at soft and hard X-ray energies: low- and high-mass X-ray binaries containing compact objects, isolated neutron stars exhibiting extreme variability as magnetars as well as pulsar-…
View article: The Challenge of Identifying the Origin of Black-Box Large Language Models
The Challenge of Identifying the Origin of Black-Box Large Language Models Open
The tremendous commercial potential of large language models (LLMs) has heightened concerns about their unauthorized use. Third parties can customize LLMs through fine-tuning and offer only black-box API access, effectively concealing unau…
View article: Fairness and/or Privacy on Social Graphs
Fairness and/or Privacy on Social Graphs Open
Graph Neural Networks (GNNs) have shown remarkable success in various graph-based learning tasks. However, recent studies have raised concerns about fairness and privacy issues in GNNs, highlighting the potential for biased or discriminato…
View article: The H.E.S.S. extragalactic sky survey with the first decade of observations
The H.E.S.S. extragalactic sky survey with the first decade of observations Open
The results of the first extragalactic gamma-ray survey by the High Energy Stereoscopic System (H.E.S.S.) are presented. The survey comprises 2720 hours of very high-energy gamma-ray observations of the extragalactic sky, recorded with H.E…
View article: On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspective
On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspective Open
Generative Foundation Models (GenFMs) have emerged as transformative tools. However, their widespread adoption raises critical concerns regarding trustworthiness across dimensions. This paper presents a comprehensive framework to address t…
View article: Captured by Captions: On Memorization and its Mitigation in CLIP Models
Captured by Captions: On Memorization and its Mitigation in CLIP Models Open
Multi-modal models, such as CLIP, have demonstrated strong performance in aligning visual and textual representations, excelling in tasks like image retrieval and zero-shot classification. Despite this success, the mechanisms by which thes…
View article: Peering Behind the Shield: Guardrail Identification in Large Language Models
Peering Behind the Shield: Guardrail Identification in Large Language Models Open
Human-AI conversations have gained increasing attention since the era of large language models. Consequently, more techniques, such as input/output guardrails and safety alignment, are proposed to prevent potential misuse of such Human-AI …
View article: Synthetic Artifact Auditing: Tracing LLM-Generated Synthetic Data Usage in Downstream Applications
Synthetic Artifact Auditing: Tracing LLM-Generated Synthetic Data Usage in Downstream Applications Open
Large language models (LLMs) have facilitated the generation of high-quality, cost-effective synthetic data for developing downstream models and conducting statistical analyses in various domains. However, the increased reliance on synthet…
View article: A comparative analysis of GNSS-inferred precipitable water vapour at the potential sites for the Africa Millimetre Telescope
A comparative analysis of GNSS-inferred precipitable water vapour at the potential sites for the Africa Millimetre Telescope Open
The Event Horizon Telescope (EHT) is a network of antennas across the globe currently used to image supermassive black holes (SMBHs) at a frequency of 230 GHz. Since the release of the image of M87$^{\ast }$ in 2019 and, subsequently, that…
View article: Analysis of the accuracy of GNSS inferred precipitable water vapour against that from a 210 GHz WVR at the H.E.S.S. site
Analysis of the accuracy of GNSS inferred precipitable water vapour against that from a 210 GHz WVR at the H.E.S.S. site Open
The High Energy Stereoscopic System (H.E.S.S.) site and the Gamsberg Mountain have been identified as potential sites for the Africa Millimetre Telescope (AMT). The AMT is poised to observe at millimetre and possibly at submillimetre wavel…
View article: Are We in the AI-Generated Text World Already? Quantifying and Monitoring AIGT on Social Media
Are We in the AI-Generated Text World Already? Quantifying and Monitoring AIGT on Social Media Open
Social media platforms are experiencing a growing presence of AI-Generated Texts (AIGTs). However, the misuse of AIGTs could have profound implications for public opinion, such as spreading misinformation and manipulating narratives. Despi…
View article: PSGraph: Differentially Private Streaming Graph Synthesis by Considering Temporal Dynamics
PSGraph: Differentially Private Streaming Graph Synthesis by Considering Temporal Dynamics Open
Streaming graphs are ubiquitous in daily life, such as evolving social networks and dynamic communication systems. Due to the sensitive information contained in the graph, directly sharing the streaming graphs poses significant privacy ris…
View article: ZeroFake: Zero-Shot Detection of Fake Images Generated and Edited by Text-to-Image Generation Models
ZeroFake: Zero-Shot Detection of Fake Images Generated and Edited by Text-to-Image Generation Models Open
The text-to-image generation model has attracted significant interest from both academic and industrial communities. These models can generate the images based on the given prompt descriptions. Their potent capabilities, while beneficial, …
View article: High-Statistics Measurement of the Cosmic-Ray Electron Spectrum with H.E.S.S.
High-Statistics Measurement of the Cosmic-Ray Electron Spectrum with H.E.S.S. Open
Owing to their rapid cooling rate and hence loss-limited propagation distance, cosmic-ray electrons and positrons (CRe) at very high energies probe local cosmic-ray accelerators and provide constraints on exotic production mechanisms such …
View article: Open LLMs are Necessary for Current Private Adaptations and Outperform their Closed Alternatives
Open LLMs are Necessary for Current Private Adaptations and Outperform their Closed Alternatives Open
While open Large Language Models (LLMs) have made significant progress, they still fall short of matching the performance of their closed, proprietary counterparts, making the latter attractive even for the use on highly private data. Rece…
View article: Hidden by a star: The redshift and the offset broad line of the flat-spectrum radio quasar PKS 0903–57
Hidden by a star: The redshift and the offset broad line of the flat-spectrum radio quasar PKS 0903–57 Open
Context. PKS 0903−57 is a little-studied γ -ray blazar that has recently attracted considerable interest due to the strong flaring episodes observed since 2020 in high energy (HE; 100 MeV ≤ E ≤ 100 GeV) and very high-energy (VHE; 100 GeV ≤…
View article: Invisibility Cloak: Disappearance under Human Pose Estimation via Backdoor Attacks
Invisibility Cloak: Disappearance under Human Pose Estimation via Backdoor Attacks Open
Despite being significant in autonomous systems, Human Pose Estimation (HPE)'s potential risks to adversarial attacks have not received comparable attention with image classification or segmentation tasks. In this paper, we study the vulne…
View article: Hidden by a star: the redshift and the offset broad line of the Flat\n Spectrum Radio Quasar PKS 0903-57
Hidden by a star: the redshift and the offset broad line of the Flat\n Spectrum Radio Quasar PKS 0903-57 Open
Context: PKS 0903-57 is a little-studied gamma-ray blazar which has recently\nattracted considerable interest due to the strong flaring episodes observed\nsince 2020 in HE (100 MeV < E < 100 GeV) and VHE (100 GeV < E < 10 TeV)\ngamma-rays.…