Výsledky vyhledávání

Report

Towards Multimodal Video Paragraph Captioning Models Robust to Missing Modality

Autor: Chen, Sishuo, Li, Lei, Ren, Shuhuai, Gao, Rundong, Liu, Yuanxin, Bi, Xiaohan, Sun, Xu, Hou, Lu

Video paragraph captioning (VPC) involves generating detailed narratives for long videos, utilizing supportive modalities such as speech and event boundaries. However, the existing models are constrained by the assumption of constant availability of

Externí odkaz: http://arxiv.org/abs/2403.19221

Zobrazit plný text záznamu

Report

Watch Out for Your Agents! Investigating Backdoor Threats to LLM-Based Agents

Autor: Yang, Wenkai, Bi, Xiaohan, Lin, Yankai, Chen, Sishuo, Zhou, Jie, Sun, Xu

Driven by the rapid development of Large Language Models (LLMs), LLM-based agents have been developed to handle various real-world applications, including finance, healthcare, and shopping, etc. It is crucial to ensure the reliability and security of

Externí odkaz: http://arxiv.org/abs/2402.11208

Zobrazit plný text záznamu

Report

Communication Efficient Federated Learning for Multilingual Neural Machine Translation with Adapter

Autor: Liu, Yi, Bi, Xiaohan, Li, Lei, Chen, Sishuo, Yang, Wenkai, Sun, Xu

Federated Multilingual Neural Machine Translation (Fed-MNMT) has emerged as a promising paradigm for institutions with limited language resources. This approach allows multiple institutions to act as clients and train a unified model through model sy

Externí odkaz: http://arxiv.org/abs/2305.12449

Zobrazit plný text záznamu

Report

Fine-Tuning Deteriorates General Textual Out-of-Distribution Detection by Distorting Task-Agnostic Features

Autor: Chen, Sishuo, Yang, Wenkai, Bi, Xiaohan, Sun, Xu

Detecting out-of-distribution (OOD) inputs is crucial for the safe deployment of natural language processing (NLP) models. Though existing methods, especially those based on the statistics in the feature space of fine-tuned pre-trained language model

Externí odkaz: http://arxiv.org/abs/2301.12715

Zobrazit plný text záznamu

Report

Expose Backdoors on the Way: A Feature-Based Efficient Defense against Textual Backdoor Attacks

Autor: Chen, Sishuo, Yang, Wenkai, Zhang, Zhiyuan, Bi, Xiaohan, Sun, Xu

Natural language processing (NLP) models are known to be vulnerable to backdoor attacks, which poses a newly arisen threat to NLP models. Prior online backdoor defense methods for NLP models only focus on the anomalies at either the input or output l

Externí odkaz: http://arxiv.org/abs/2210.07907

Zobrazit plný text záznamu

Report

Holistic Sentence Embeddings for Better Out-of-Distribution Detection

Autor: Chen, Sishuo, Bi, Xiaohan, Gao, Rundong, Sun, Xu

Detecting out-of-distribution (OOD) instances is significant for the safe deployment of NLP models. Among recent textual OOD detection works based on pretrained language models (PLMs), distance-based methods have shown superior performance. However,

Externí odkaz: http://arxiv.org/abs/2210.07485

Zobrazit plný text záznamu