语言大模型 / LLM

RSPrompter: Learning to Prompt for Remote Sensing Instance Segmentation based on Visual Foundation Model

Keyan Chen, Chenyang Liu, Hao Chen, Haotian Zhang, Wenyuan Li, Zhengxia Zou, Zhenwei Shi

专题命中 预训练与数据 ：foundation model(title)

Pretraining Conformer with ASR or ASV for Anti-Spoofing Countermeasure

Yikang Wang, Hiromitsu Nishizaki, Ming Li

专题命中 预训练与数据 ：pretraining(title)

Comments 7 pages, 2 figures

A Survivor in the Era of Large-Scale Pretraining: An Empirical Study of One-Stage Referring Expression Comprehension

Gen Luo, Yiyi Zhou, Jiamu Sun, Xiaoshuai Sun, Rongrong Ji

专题命中 预训练与数据 ：pretraining(title)

Gradient-Regulated Meta-Prompt Learning for Generalizable Vision-Language Models

Juncheng Li, Minghe Gao, Longhui Wei, Siliang Tang, Wenqiao Zhang, Mengze Li, Wei Ji, Qi Tian, Tat-Seng Chua, Yueting Zhuang

专题命中 预训练与数据 ：language model(title)

Comments Accepted by ICCV 2023

Channel-Aware Pretraining of Joint Encoder-Decoder Self-Supervised Model for Telephonic-Speech ASR

Vrunda N. Sukhadia, A. Arunkumar, S. Umesh

专题命中 预训练与数据 ：pretraining(title)

Comments 5 pages, 5 figures

Teaching Structured Vision&Language Concepts to Vision&Language Models

Sivan Doveh, Assaf Arbelle, Sivan Harary, Rameswar Panda, Roei Herzig, Eli Schwartz, Donghyun Kim, Raja Giryes, Rogerio Feris, Shimon Ullman, Leonid Karlinsky

专题命中 预训练与数据 ：language model(title)

Journal ref CVPR 2023

Post-training Model Quantization Using GANs for Synthetic Data Generation

Athanasios Masouris, Mansi Sharma, Adrian Boguszewski, Alexander Kozlov, Zhuo Wu, Raymond Lo

专题命中 预训练与数据 ：post-training(title)

Procedure-Aware Pretraining for Instructional Video Understanding

Honglu Zhou, Roberto Martín-Martín, Mubbasir Kapadia, Silvio Savarese, Juan Carlos Niebles

专题命中 预训练与数据 ：pretraining(title)

Comments CVPR 2023

CapDet: Unifying Dense Captioning and Open-World Detection Pretraining

Yanxin Long, Youpeng Wen, Jianhua Han, Hang Xu, Pengzhen Ren, Wei Zhang, Shen Zhao, Xiaodan Liang

专题命中 预训练与数据 ：pretraining(title)

Comments Accepted by CVPR2023

Ensemble Ranking Model with Multiple Pretraining Strategies for Web Search

Xiaojie Sun, Lulu Yu, Yiting Wang, Keping Bi, Jiafeng Guo

专题命中 预训练与数据 ：pretraining(title)

Comments 4 pages, 2 figures, WSDM Cup 2023

Generative Negative Text Replay for Continual Vision-Language Pretraining

Shipeng Yan, Lanqing Hong, Hang Xu, Jianhua Han, Tinne Tuytelaars, Zhenguo Li, Xuming He

专题命中 预训练与数据 ：pretraining(title)

Comments ECCV 2022

Depth Contrast: Self-Supervised Pretraining on 3DPM Images for Mining Material Classification

Prakash Chandra Chhipa, Richa Upadhyay, Rajkumar Saini, Lars Lindqvist, Richard Nordenskjold, Seiichi Uchida, Marcus Liwicki

专题命中 预训练与数据 ：pretraining(title)

Comments Accepted to CVF European Conference on Computer Vision Workshop(ECCVW 2022)

PreViTS: Contrastive Pretraining with Video Tracking Supervision

Brian Chen, Ramprasaath R. Selvaraju, Shih-Fu Chang, Juan Carlos Niebles, Nikhil Naik

专题命中 预训练与数据 ：pretraining(title)

Comments To be presented at WACV 2023

Generalizable Task Planning through Representation Pretraining

Chen Wang, Danfei Xu, Li Fei-Fei

专题命中 预训练与数据 ：pretraining(title)

Supervised and Self-supervised Pretraining Based COVID-19 Detection Using Acoustic Breathing/Cough/Speech Signals

Xing-Yu Chen, Qiu-Shi Zhu, Jie Zhang, Li-Rong Dai

专题命中 预训练与数据 ：pretraining(title)

Comments Accepted by ICASSP 2022

Journal ref ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022, pp. 561-565

Geometric Synthesis: A Free lunch for Large-scale Palmprint Recognition Model Pretraining

Kai Zhao, Lei Shen, Yingyi Zhang, Chuhan Zhou, Tao Wang, Ruixin Zhang, Shouhong Ding, Wei Jia, Wei Shen

专题命中 预训练与数据 ：pretraining(title)

Comments Codes are available at http://kaizhao.net/palmprint

Multiscale Convolutional Transformer with Center Mask Pretraining for Hyperspectral Image Classification

Sen Jia, Yifan Wang

专题命中 预训练与数据 ：pretraining(title)

Comments 8 pages, 26 figures, conference paper

Combining CNNs With Transformer for Multimodal 3D MRI Brain Tumor Segmentation With Self-Supervised Pretraining

Mariia Dobko, Danylo-Ivan Kolinko, Ostap Viniavskyi, Yurii Yelisieiev

专题命中 预训练与数据 ：pretraining(title)

Is Pretraining Necessary for Hyperspectral Image Classification?

Hyungtae Lee, Sungmin Eum, Heesung Kwon

专题命中 预训练与数据 ：pretraining(title)

Comments IGARSS 2019 submission

Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation

Jonas Wulff, Michael J. Black

专题命中 预训练与数据 ：pretraining(title)

Comments 16 pages, 7 figures; accepted for publication at the German Conference for Pattern Recognition (GCPR) 2018

Cross-paradigm pretraining of convolutional networks improves intracranial EEG decoding

Joos Behncke, Robin Tibor Schirrmeister, Martin Völker, Jiří Hammer, Petr Marusič, Andreas Schulze-Bonhage, Wolfram Burgard, Tonio Ball

专题命中 预训练与数据 ：pretraining(title)

Unsupervised Network Pretraining via Encoding Human Design

Ming-Yu Liu, Arun Mallya, Oncel C. Tuzel, Xi Chen

专题命中 预训练与数据 ：pretraining(title)

Comments 9 pages, 11 figures, WACV 2016: IEEE Conference on Applications of Computer Vision

Aslema at NADI 2026: Data Augmentation for Intent Recognition and Slot Filling

Aslema 参与 NADI 2026：基于少样本的 SLU 增强方法

Tajwaar Shafiq, Hunzalah Hassan Bhatti, Firoj Alam, Shammur Absar Chowdhury

机构 * Qatar Computing Research Institute（卡塔尔计算研究所） 纠错

专题命中 预训练与数据 ：LLM(abstract)；分类 cs.CL、cs.AI；large language model(comments)；language model(comments)

AI总结 Aslema 系统参与 NADI 2026 共享任务 5，通过微调模型及合成数据增强（含文化适配语句生成与语音克隆）提升性能，在官方测试集槽填充排第1、意图识别排第4，将公开脚本与合成数据集。

Comments LLMs, Native, Arabic LLMs, Augmentation, Multilingual, Multimodal, Language Diversity, Contextual Understanding, Minority Languages, Culturally Informed, Foundation Models, Large Language Models, Audio Models, Omni Models, Slot Filling

RefLAM: A Reference-Grounded Line Annotation Pipeline for Historical Arabic Manuscripts

RefLAM：一种用于历史阿拉伯文手稿的参考依据型行标注流水线

Mohamed Guechaoui, Mohamed Diaa Zellagui, Souleyman Chaib, Sahraoui Dhelim

机构 * Higher School of Computer Science (ESI-SBA)（高等计算机科学学院（ESI-SBA）） 纠错

专题命中 预训练与数据 ：large language model(abstract)；language model(abstract)；分类 cs.CL

AI总结 RefLAM是用于历史阿拉伯文手稿的参考依据型行标注流水线，结合深度学习、MLLM与模糊对齐引擎，可高效生成标注数据，吞吐量较手动标注提升75倍，发布的AraMS-28k可用于HTR模型微调。

Comments 11 pages, 6 figures, 3 tables, 2 algorithms

STORM: Stepwise Token Optimization with Reward-Guided Beam Search

STORM: 基于奖励引导束搜索的逐步令牌优化

Arthur Satouf, Giulio D'Erasmo, Yuxuan Zong, Habiboulaye Amadou Boubacar, Pablo Piantanida, Benjamin Piwowarski

机构 * MILA – Quebec AI Institute & ILLS（魁北克人工智能研究所与ILLs） ； Université Paris-Saclay & CentraleSupélec & CNRS（巴黎-萨克雷大学及CentraleSupélec与CNRS） ； Air Liquide ； Sorbonne Université & ISIR & CNRS（索邦大学及ISIR与CNRS） ； Sapienza, University of Rome（罗马大学Sapienza） 纠错

专题命中 预训练与数据 ：LLM(abstract,abstract_cn)；分类 cs.AI

AI总结 提出STORM框架，通过检索奖励引导的束搜索在每一步优化令牌选择，实现词汇检索的查询扩展，在多个基准上匹配或超越大模型重写器，并零样本迁移至18种语言。

Addressing Corpus Knowledge Poisoning Attacks on RAG Using Sparse Attention

通过稀疏注意力缓解RAG中的语料知识污染攻击

Sagie Dekel, Moshe Tennenholtz, Oren Kurland

机构 * Faculty of Data and Decision Sciences, Technion - Israel Institute of Technology, Haifa, Israel（数据与决策科学学院，技术Ion-以色列理工学院，海法，以色列） 纠错

专题命中 预训练与数据 ：LLM(abstract,abstract_cn)；分类 cs.AI

AI总结 本文提出SDAG方法，通过稀疏注意力机制有效防御RAG中的语料知识污染攻击，显著提升防御性能。

LLMTrace: A Corpus for Classification and Fine-Grained Localization of AI-Written Text

LLMTrace：用于AI生成文本分类与细粒度定位的语料库

Irina Tolstykh, Aleksandra Tsybina, Sergey Yakubson, Maksim Kuprashevich

机构 * Department of R&D, SALUTEDEV LLC（研发部，SALUTEDEV LLC） 纠错

专题命中 预训练与数据 ：large language model(abstract)；language model(abstract)；分类 cs.CL

AI总结 针对AI文本检测领域缺乏合适训练数据的问题，本文构建了双语语料库LLMTrace，支持全文二元分类与AI生成区间检测，可用于训练评估更精细的AI检测模型。

Comments EMNLP 2026 Findings

Evidence Blindness in Direct Corpus Interaction: Persistent Navigation with AtlasNav

直接语料库交互中的证据失明：基于AtlasNav的持久导航

Hongyu Guo, Zhiyu Zheng, Zhao Cao

专题命中 预训练与数据 ：large language model(abstract)；language model(abstract)；分类 cs.AI

AI总结 针对直接语料库交互中存在的证据失明问题，提出AtlasNav框架，通过一次性构建语料库图谱实现自适应导航，在BrowseComp-Plus等数据集上提升了准确率并降低了推理成本。

Comments 27 pages, 7 figures. Code, data, and trajectories will be released

ROBE: Reversed-Order-Biased-Experts for Extracting Extreme Long-tail Events from Historical Texts

ROBE：用于从历史文本中提取极长尾事件的逆序偏倚专家模型

Stella Verkijk, Piek Vossen

机构 * Vrije Universiteit Amsterdam（阿姆斯特丹自由大学） ； Huygens Instituut（惠更斯研究所） 纠错

专题命中 预训练与数据 ：large language model(abstract)；language model(abstract)；分类 cs.CL

AI总结 针对未被大模型预训练覆盖的19世纪前荷兰历史语料库，提出ROBE逆序偏倚专家模型，结合领域合成数据提升，使长尾事件提取的召回、精确率及F1值均优于简单微调编码器模型。

Comments 15 pages, 3 figures

Breaking the Assumptions: Auditing Input-Side Jailbreak Defenses Against Semantic Attacks

打破假设：针对语义攻击的输入端越狱防御审计

Aaditya Pratap, Harsh Kasyap, Somanath Tripathy

专题命中 预训练与数据 ：large language model(abstract)；language model(abstract)；分类 cs.AI

AI总结 本文针对本地部署大语言模型的输入端越狱防御开展审计，追溯防御失效的假设根源，在6款14B至35B参数的开源权重模型上，用100条来自40余公开源的越狱提示完成13800条评估记录的测试。