River Research and Applications, Volume 42, Issue 8, Page 1705-1719, October 2026.ABSTRACT The Marias River flows from Glacier National Park through northcentral Montana, and into the Missouri River. Annual flows gradually declined from 1902 to 2024 (~3.2%/decade) and the 1952 Tiber Dam and Lake Elwell reservoir were operated to attenuate peak flows and stabilize downstream flows year‐round.
Stewart B. Rood, Lori A. Goater
wiley +1 more source
Fed-DCR: a backdoor defense scheme for federated learning based on dual-domain synergistic detection
Tongxin xuebaoFederated learning has been shown to be vulnerable to backdoor attacks, while existing defense methods mainly rely on single statistical features in the parameter space.
Kang Jie +5 more
doaj +2 more sources
Towards Generalizable Deepfake Detection: An Inconsistency-Aware Vision-Language Model Tuning Approach. [PDF]
Sensors (Basel)Zhang L, Peng S, Xu M, Lu T.
europepmc +1 more source
LOW-RANK ADAPTATION (LoRa): REVOLUTIONIZING MODEL OPTIMIZATION IN DEEP LEARNING
This article comprehensively explores Low-Rank Adaptation (LoRa), an innovative optimization technique for deep learning models. It delves into the theoretical foundations, implementation strategies, and real-world applications of LoRa across various domains, including natural language processing, computer vision, and speech recognition.
openaire +2 more sources
A Statistical Perspective on Transformers for Small Longitudinal Cohort Data
Statistics in Medicine, Volume 45, Issue 23-24, October 2026.ABSTRACT Modeling of longitudinal cohort data typically involves complex temporal dependencies between multiple variables. There, the transformer architecture, which has been highly successful in language and vision applications, allows us to account for the fact that the most recently observed time points in an individual's history may not always be ...
Kiana Farhadyar +12 more
wiley +1 more source
A Prompt-Preserving SAM ViT-B Adaptation Framework with Historical Branch Fusion and Soft Convolutional Expert Weighting for Medical Image Segmentation. [PDF]
Bioengineering (Basel)Ping S +6 more
europepmc +1 more source
Low Rank Adaptation for Monocular Depth Estimation
This project integrates Low-Rank Adaptation (LoRA) into the LiteMono model for efficient monocular depth estimation across domains. By fine-tuning on the NYU-Depth-V2 dataset, we compare LoRA’s parameter-efficient performance against traditional fine ...
AlMusa, Hadi
core
Adapting a Pretrained LLM to Rapidly Recover Earthquake Magnitude and Location
Journal of Geophysical Research: Machine Learning and Computation, Volume 3, Issue 5, October 2026.Abstract Determining earthquake magnitude and location through fast, automated methods is fundamental for seismic monitoring. Recent studies have shown that Large Language Models (LLMs) can operate as pattern‐recognition systems capable of handling novel domains and purely numerical tasks.
Aurora Bassani +6 more
wiley +1 more source
Convolutional low-rank adaptation for efficient semantic segmentation in vision transformers. [PDF]
Sci RepSrinivasan S +4 more
europepmc +1 more source
PeriodicLoRA: Breaking the Low-Rank Bottleneck in LoRA Optimization [PDF]
Supervised fine-tuning is the most common method to adapt large language models (LLMs) to downstream tasks, but full fine-tuning LLMs requires massive computational resources.
Sui, Zhifang +9 more
core