-
Analyzing the Impact of Low-Rank Adaptation for Cross-Domain Few-Shot Object Detection in Aerial Images
Authors:
Hicham Talaoubrid,
Anissa Mokraoui,
Ismail Ben Ayed,
Axel Prouvost,
Sonimith Hang,
Monit Korn,
Rémi Harvey
Abstract:
This paper investigates the application of Low-Rank Adaptation (LoRA) to small models for cross-domain few-shot object detection in aerial images. Originally designed for large-scale models, LoRA helps mitigate overfitting, making it a promising approach for resource-constrained settings. We integrate LoRA into DiffusionDet, and evaluate its performance on the DOTA and DIOR datasets. Our results s…
▽ More
This paper investigates the application of Low-Rank Adaptation (LoRA) to small models for cross-domain few-shot object detection in aerial images. Originally designed for large-scale models, LoRA helps mitigate overfitting, making it a promising approach for resource-constrained settings. We integrate LoRA into DiffusionDet, and evaluate its performance on the DOTA and DIOR datasets. Our results show that LoRA applied after an initial fine-tuning slightly improves performance in low-shot settings (e.g., 1-shot and 5-shot), while full fine-tuning remains more effective in higher-shot configurations. These findings highlight LoRA's potential for efficient adaptation in aerial object detection, encouraging further research into parameter-efficient fine-tuning strategies for few-shot learning. Our code is available here: https://github.com/HichTala/LoRA-DiffusionDet.
△ Less
Submitted 8 April, 2025;
originally announced April 2025.
-
The f_LT Response Function of D(e,e'p)n at Q^2=0.33(GeV/c)^2
Authors:
W. U. Boeglin,
H. Arenhövel,
K. I. Blomqvist,
R. Böhm,
M. Distler,
R. Edelhoff,
I. Ewald,
R. Florizone,
J. Friedrich,
R. Geiges,
J. Jourdan,
M. Kahrau,
M. Korn,
H. Kramer,
K. W. Krygier,
V. Kunde,
M. Kuss,
A. Liesenfeld,
K. Merle,
R. Neuhausen,
E. A. J. M. Offermann,
Th. Pospischil,
M. Potokar,
A. W. Richter,
A. Rokavec
, et al. (9 additional authors not shown)
Abstract:
The interference response function f_LT (R_LT) of the D(e,e'p)n reaction has been determined at squared four-momentum transfer Q^2 = 0.33 (GeV/c)^2 and for missing momenta up to p_miss= 0.29 (GeV/c). The results have been compared to calculations that reproduce f_LT quite well but overestimate the cross sections by 10 - 20% for missing momenta between 0.1 (GeV/c) and 0.2 (GeV/c) .
The interference response function f_LT (R_LT) of the D(e,e'p)n reaction has been determined at squared four-momentum transfer Q^2 = 0.33 (GeV/c)^2 and for missing momenta up to p_miss= 0.29 (GeV/c). The results have been compared to calculations that reproduce f_LT quite well but overestimate the cross sections by 10 - 20% for missing momenta between 0.1 (GeV/c) and 0.2 (GeV/c) .
△ Less
Submitted 23 April, 2008;
originally announced April 2008.
-
Measurements of 1/ f noise in Josephson junctions at zero voltage: Implications for decoherence in superconducting quantum bits
Authors:
Michael Mueck,
Matthias Korn,
C. G. A. Mugford,
J. B. Kycia,
John Clarke
Abstract:
Critical current fluctuations with a 1/ f spectral density (f is frequency) are potentially a limiting source of intrinsic decoherence in superconducting quantum bits (qubits) based on Josephson tunnel junctions. Prior measurements of this noise were made at nonzero voltages whereas qubits are operated in the zero voltage state. We report measurements of 1/ f noise in a dc superconducting quantu…
▽ More
Critical current fluctuations with a 1/ f spectral density (f is frequency) are potentially a limiting source of intrinsic decoherence in superconducting quantum bits (qubits) based on Josephson tunnel junctions. Prior measurements of this noise were made at nonzero voltages whereas qubits are operated in the zero voltage state. We report measurements of 1/ f noise in a dc superconducting quantum interference device first, coupled to a resonant tank circuit and operated in a dispersive mode at zero voltage, and, second, operated conventionally with a current bias in the voltage regime. Both measurements yield essentially the same magnitude of critical current 1/ f noise.
△ Less
Submitted 7 January, 2005;
originally announced January 2005.
