Searcharxiv⌕ Search

arXiv subjects

Wenzheng Liu

Publications and source records attributed to Wenzheng Liu.

5 recordsLinked to original sources

SlopeGuard: Towards Generalizable AI-Generated Image Detection via Radial Spectral Slope Guidance

AI-generated image detectors often overfit to generator-specific artifacts, hindering generalization to unseen synthesis pipelines. Existing frequency-based methods usually learn high-dimensional spectral patterns that may preserve generator-dependent variations. We explore Radial Spectral Slope, a compact descriptor of global spectral decay that captures transferable spectral trends while suppressing fine-grained variations. Our analysis shows that real and generated images exhibit systematic but generator-dependent slope differences, indicating that slope provides useful forensic evidence but is insufficient as a standalone decision criterion. This motivates combining spectral slope with semantic representations for more robust detection. Based on this insight, we propose SlopeGuard, which couples slope-aware data processing with slope guided feature learning to construct informative patch-based views, integrate Radial Spectral Slope with semantic features, and adapt prior preservation according to spectral atypicality. Extensive experiments demonstrate that SlopeGuard achieves state-of-the-art mean accuracies of 97.5% on AIGCDetectBench, 97.8% on GenImage, and 85.1% on AIGIBench, while maintaining competitive performance under blur, cropping, JPEG compression, and additive noise.

cs.CV↗

Reconfigurable multi-tone mmWave/THz generation powered by high-efficiency Kerr parametric oscillation

Broadband and simultaneously accessible multi-tone millimeter-wave (mmWave) and terahertz (THz) signals are attractive for multi-band wireless communication and spectroscopy applications, yet their generation remains constrained by electronic bandwidth limitations and inefficient optical frequency conversion. Here, we demonstrate reconfigurable photonic generation of multi-tone mmWave/THz signals with comb spacings of 100, 200, 300, and 400 GHz, using a strongly asymmetric dual-ring silicon-nitride (SiN) microresonator and a broadband spiral plasmonic photomixer. A localized avoided-mode crossing (AMX) introduced by an auxiliary ring supports a fixed optical parametric oscillation (OPO) mode pair over an extended pump-detuning range, consistent with passive compensation of detuning-dependent phase mismatch. This enables a maximum pump-to-generated-line conversion efficiency of 67.3 % with 25.06 dB of pump depletion across different pump wavelength selections. Direct photomixing of these optical states produces simultaneous multi-frequency tones spanning 100 to 900 GHz without intermediate optical amplification. The generated carriers are further characterized through down-converted spectra, frequency-noise power spectral density (PSD), and Allan deviation measurements, yielding a \b{eta}-separation linewidth of 26 kHz at an observation time of 0.1 s. Optical modulation offers high radiated power levels exceeding 2 mW at 100 GHz at a duty cycle of 1 %, while mitigating thermal breakdown in the photomixer. High optical conversion efficiency, reconfigurable and selectable comb spacings, and broadband multi-tone generation on a chip-scale platform provide a compact route toward reconfigurable multi-band mmWave/THz sources for wireless communication, spectroscopy, and sensing.

physics.optics↗

MMReason: An Open-Ended Multi-Modal Multi-Step Reasoning Benchmark for MLLMs Toward AGI

Reasoning plays a crucial role in advancing Multimodal Large Language Models (MLLMs) toward Artificial General Intelligence. However, existing MLLM benchmarks often fall short in precisely and comprehensively evaluating long-chain reasoning abilities from three key aspects: (1) lack of difficulty and diversity, (2) susceptibility to guessability and memorization, (3) inadequate assessment of intermediate reasoning steps. To fill this gap, we introduce MMReason, a new benchmark designed to precisely and comprehensively evaluate MLLM long-chain reasoning capability with diverse, open-ended, challenging questions. First, we curate challenging questions requiring multi-step reasoning from various fields (i.e., 6 disciplines) and multiple difficulty levels (i.e., from pre-university to university, and from foundational to competition tiers). Second, these questions are reformulated into an open-ended format and filtered using a multi-model voting technique to eliminate shortcut cases related to guessing and memorization, ensuring robust reasoning evaluations. Third, we annotate the questions with detailed step-by-step solutions, and design a reference-based ternary scoring mechanism to reliably assess intermediate reasoning steps. With MMReason, we benchmark popular leading MLLMs and provide an in-depth analysis of their reasoning capabilities. We hope MMReason will serve as a valuable resource for advancing MLLM reasoning research. Code will be available at https://github.com/HJYao00/MMReason.

cs.AI↗

The mechanism of multi-anode electrode geometry applied to vacuum arc thruster

A multi-anode electrode geometry suitable for vacuum breakdown is proposed in this paper. This electrode geometry forms a unique electric field distribution based on its anode configuration which is different from the conventional electrode geometry. There exists a region A between the insulated-anode and the remote-anode where the electric field vector has opposite directions. The existence of region A directly affects the movement of electrons in the initial stage of the discharge, thereby changing the process of vacuum breakdown. The variation of discharge process affects the generation and propagation mechanism of plasma, forming a plasma plume in favor of the propulsion performance of thrusters. The photograph of plasma plume and the electron density spatial distribution of plasma plume preliminarily prove the pinch effect of multi-anode electrode geometry on the radial diffusion of plasma. In the optimization of remote-anode of multi-anode electrode geometry, the variation of the distance of cathode and remote-anode (short for D) is found to affect the pinch effect. Through the reduction of D, the directional jet effect of the multi-anode electrode geometry is enhanced under the same discharge parameters. Applying it to the VAT (multi-anode VAT) and comparing it with the VAT based on conventional electrode geometry (conventional VAT), it is found that the propulsion performance of multi-anode VAT is superior to conventional VAT, and it is further promoted with the decrease of D. Under the same discharge parameters, the impulse bit and the thrust-to-power ratio of the multi-anode VAT is able to be increased by a factor of 26 and 2.88 respectively compared with the conventional VAT. Theoretical analysis and investigation of propulsion performance validate the potential application advantages of the multi-anode VAT in the field of nanosatellite propulsion.

physics.app-ph↗

Optimization of Plasma Plume Characteristics Based on Multi-anode Coaxial Ablative Pulsed Plasma Thruster

A special surface discharge is proposed based on the multi-anode electrode geometry. Instead of the traditional surface flashover of creepage on the insulator surface between the electrodes, a surface discharge with one of the electrodes being placed far away from the insulator is achieved in this paper. The unique electric field distribution due to the multi-anode electrode geometry has a significant influence on the discharge process of the surface discharge. It changes the generation and propagation process of the plasma, forming a plasma plume contributes to the propulsion performance of the thruster. Through theoretical analysis of the obtained plume data, it is indicated that the ablative pulsed plasma thruster based on multi-anode electrode geometry (short for multi-anode APPT) promotes the internal pressure of the plasma jet during its propagation and significantly increases the density and energy of charged particles. The discharge phenomena manifest that the multi-anode APPT and the helix-coil multi-anode APPT effectively increase the intensity of the plasma plume. Through electron density spatial distribution measurement, it has been found that the helix-coil multi-anode APPT increases the density of plasma in the axial direction to more than 4 times of the conventional coaxial APPT and reduces the electron density in other directions. In the propulsion test, it has been demonstrated that the multi-anode APPT and the helix-coil multi-anode APPT have better performance in terms of the impulse bit and the thrust-to-power ratio. In addition, it is also identified that the pinch effect will be enhanced with the increase of discharge power and the propulsion performance is promoted more distinctly. The multi-anode APPT and the helixcoil multi-anode APPT have been proved to have potential application advantage in the field of micro-satellite propulsion.

physics.app-ph↗