实验室概况


上海交通大学人工智能安全实验室（AI Security Lab）现有教师6位、在读博士生10余位、在读硕士生20余位。 实验室聚焦人工智能技术自身的安全问题以及人工智能技术在安全领域的应用，主要研究内容安全、大模型安全、联邦学习、深度模型水印、深度伪造检测、生物特征身份认证等。 累计发表学术论文400余篇，其中多篇论文发表在IEEE TIFS / TPAMI / TFS / TCSVT / TCYB / TNNLS、ACL、ICCV、AAAI、IJCAI、EMNLP等顶级期刊和会议上。

团队成员

Avatar

刘功申

教授

人工智能安全, 自然语言处理, 大模型及其安全, 深度学习框架及其安全, 多模态大模型

Avatar

王士林

教授

多媒体信息分析与处理, 多媒体信息内容安全, 人工智能安全

Avatar

孟魁

高级实验师

内容安全, 系统安全与仿真

Avatar

张倬胜

助理教授

自然语言处理, 预训练语言模型, 自主智能体及其安全

Avatar

段苏峰

助理研究员

内容安全, 自然语言处理, 机器翻译

Avatar

黄林庆

助理研究员

低质数据挖掘与学习, 遥感影像智能解译, 人工智能安全及应用

最新成果

自然语言处理中的探针可解释方法综述
随着大规模预训练模型的广泛应用，自然语言处理的各领域（如文本分类和机器翻译）均取得了长足的发展。然而，受限于预训练模型的“黑盒”特性，其内部的决策模式以及编码的知识信息被认为是不透明的。以OpenAI发布的ChatGPT和GPT-4为代表的先进预训练模型为例，它们在各领 …
PDF Cite DOI
自然语言处理中的探针可解释方法综述
Backdoor NLP Models via AI-Generated Text
Backdoor attacks pose a critical security threat to natural language processing (NLP) models by establishing covert associations …
PDF
Backdoor NLP Models via AI-Generated Text
Multi-Grained Multimodal Interaction Network for Sentiment Analysis
Multimodal sentiment analysis aims to utilize different modalities including language, visual, and audio to identify human emotions in …
PDF
Multi-Grained Multimodal Interaction Network for Sentiment Analysis
Speaker-Adaptive Lipreading via Spatio-Temporal Information Learning
Lipreading has been rapidly developed recently with the help of large-scale datasets and big models. Despite the significant progress …
PDF
Speaker-Adaptive Lipreading via Spatio-Temporal Information Learning
Data-Free Watermark for Deep Neural Networks by Truncated Adversarial Distillation
Model watermarking secures ownership verification and copyright protection of deep neural networks. In the black-box scenario, …
PDF
Data-Free Watermark for Deep Neural Networks by Truncated Adversarial Distillation
Revisiting the Information Capacity of Neural Network Watermarks: Upper Bound Estimation and Beyond
To trace the copyright of deep neural networks, an owner can embed its identity information into its model as a watermark. The capacity …
PDF DOI
Revisiting the Information Capacity of Neural Network Watermarks: Upper Bound Estimation and Beyond
NWS: Natural Textual Backdoor Attacks via Word Substitution
Backdoor attacks pose a serious security threat for natural language processing (NLP). Backdoored NLP models perform normally on clean …
PDF
NWS: Natural Textual Backdoor Attacks via Word Substitution
SDPSAT: Syntactic Dependency Parsing Structure-Guided Semi-Autoregressive Machine Translation
The advent of non-autoregressive machine translation (NAT) accelerates the decoding superior to autoregressive machine translation (AT) …
PDF Cite DOI
SDPSAT: Syntactic Dependency Parsing Structure-Guided Semi-Autoregressive Machine Translation
Fine-Grained Lip Image Segmentation using Fuzzy Logic and Graph Reasoning
Fine-grained lip image segmentation plays a critical role in downstream tasks such as automatic lipreading, as it enables the accurate …
PDF DOI
Fine-Grained Lip Image Segmentation using Fuzzy Logic and Graph Reasoning
Is Continuous Prompt a Combination of Discrete Prompts? Towards a Novel View for Interpreting Continuous Prompts
The broad adoption of continuous prompts has brought state-of-the-art results on a diverse array of downstream natural language …
PDF Cite DOI
Is Continuous Prompt a Combination of Discrete Prompts? Towards a Novel View for Interpreting Continuous Prompts
查看全部