花了半年的时间调研深度学习与迁移学习各个方向的文章,终于在过年前将 Transferability in Deep Learning: A Survey投了出去(~▽~)~*。当下大部分的深度学习应用(尤其是CV和NLP)或多或少都会用到预训练模型,…
To foster the verifiability and testability of Deep Neural Networks (DNN), an increasing number of methods for test case generation techniques are being developed. When confronted with testing DNN models, the user can apply any existing test generation technique. However, it needs to do so for ...
Adversarial attack transferability is well-recognized in deep learning. Prior work has partially explained transferability by recognizing common adversarial subspaces and correlations between decision boundaries, but little is known beyond this. We propose that transferability between seemingly different models ...
Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples Lustin 2 人赞同了该文章 对抗样本不仅可以影响一个模型上产生作用,也可以影响其他模型,即使这两个模型含有不同的架构,而且这两个模型有不一样训练集,只要所两个模型经过有目的的训练就都可以达到预期的效果。一个...
A Survey on Transferability of Adversarial Examples across Deep Neural Networks. Jindong Gu, Xiaojun Jia, Pau de Jorge, Wenqain Yu, Xinwei Liu, Avery Ma, Yuan Xun, Anjun Hu, Ashkan Khakzar, Zhijiang Li, Xiaochun Cao, Philip Torr. Preprint 2023. [pdf] If you find our paper and repo...
The best performance is marked in bold. PMTrans takes advantage of a different data partitioning strategy. Our FFTAT adheres to the same data partitioning strategy as SSRT and CDTrans.ResNet101 [he2016deep] clp inf pnt qdr rel skt Avg. clp - 19.3 37.5 11.1 52.2 41.1 32.2 inf 30.2 - ...
Here we address these issues by 1) Using a radial basis function network (RBFN) in place of the KNN allowing far fewer images to be used in deployment and 2) Using an autoencoder network for explainability. In addition to these techniques, we examine the effects of transfer learning to ...
Robust physical-world attacks on deep learning visual classification. In CVPR, 2018. 2 [11] Leon A Gatys, Alexander S Ecker, and Matthias Bethge. Im- age style transfer using convolutional neural networks. In CVPR, 2016. 2 [12] Ian J. Goodfellow, Jonathon Shlens, a...
(4) 利用基于模糊规则的深度神经网络实现无源多域自适应Source-Free Multidomain Adaptation With Fuzzy Rule-Based Deep Neural Networks (5) 通过模糊强化学习实现一类多代理系统的规定时间编队控制Prescribed-Time Formation Control for a Class of Multiagent Systems via Fuzzy Reinforcement Learning ...
Empirical Findings of MLP in SL-MLP 主要观点: MLP projector avoids transferability drop at stage 5 in supervised pretraining. MLP projector enlarges the intra-class variation of features. MLP projector reduces feature distribution distance between pre-D and eval-D. ...