Adversarial Transferability in Wearable Sensor Systems

Introduction

This article is brief summary of our paper on β€œAdversarial Transferability in Wearable Sensor Systems.” In this work, we have explored the topic of adversarial transferability from viewpoints that we believe are novel have not been discussed yet. But before we present our results and findings, let’s first understand a few topics and the framework of the paper.

Adversarial Transferability

For the uninitiated, adversarial transferability captures the ability of adversarial examples that makes them transfer between independently trained models of different architectures. In almost all works on adversarial transferability, the discussion is usually carried out from the perspective of models. We believe in understanding adversarial transferability fully, and in uncovering the reasons behind their existence in the first place, we need to consider other avenues such as the datasets that are used to train the machine learning models. To this end in our work. We have tried to explore adversarial transferability and give a comprehensive discussion that takes into account the model and the dataset with wearable sensor systems as a case study.

Classifiers and Attack Methods

In our work we have used the following classifiers:

You can find more details about them in our paper.

Discussion

In this work, we have explored adversarial transferability from the following perspectives.

Figure 1. Accuracy of the trained classifiers.
Figure 2. Misclassification rate of different classifiers on untargeted adversarial examples computed using the DNN model.
Figure 3. Success rate of different classifiers on targeted adversarial examples computed using the DNN model.
Figure 4. Misclassification rate of even and odd models on the untargeted adversarial examples computed using the even model.
Figure 5. Success rate of even and odd models on the on targeted adversarial examples computed using the even model.
Figure 6. Misclassification rate of chest and wrist models on the untargeted adversarial examples computed using the chest model.
Figure 7. Success rate of chest and wrist models on the targeted adversarial examples computed using the chest model.
Figure 8. Misclassification rate of UCI and MHEALTH models on the untargeted adversarial examples computed using the UCI model.
Figure 9. Success rate of UCI and MHEALTH models on the targeted adversarial examples computed using the UCI model.

Conclusions

Our aim in this work was the extend the discussion of adversarial transferability beyond the current discussion of the inter-model scenario and demonstrates how adversarial transferability fares in new conditions. We think we have done what we intended to do from our results and discussion. We encourage the reader to check out our paper for more details.