神经网络基础：感知机、多层网络、训练方法、CNN图像分类
Perceptrons
Multi-layer Networks
Training Method
Best Practices
CNN
开始学习
现代预训练模型：Transformer架构、BERT预训练与微调、扩展模型
Transformer
Learning/Data
BERT
Fine-Tuning
Extensions
开始学习
对抗攻击与防御：白盒/黑盒攻击、模型反演、模型抽取
Understanding
Evasion
White-box
Black-box
Model Inversion
Model Extraction
开始学习
模型蒸馏与抽取：知识迁移、模型窃取、攻击与防御
Understanding
Model Distillation
Model Extraction
Attack Methods
Defense
开始学习
后门攻击与防御：数据投毒、模型投毒、Neural Cleanse检测
Backdoor Taxonomy
Outsourcing
BadNets
Trojaning
Federated Learning
Defense
Neural Cleanse
开始学习
隐私保护：成员推断、属性推断、差分隐私
Threats
Privacy Risks
Membership Inference
Attribute Inference
Differential Privacy
开始学习
深度伪造与检测：人脸合成、换脸技术、伪造检测、泛化性
Generation
Face Synthesis
Attribute Manipulation
Face Swap
Detection
Datasets
Generalizability
开始学习