On Adversarial Examples for Text Classification by Perturbing Latent Representations
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913343970738176 |
|---|---|
| author | Sooksatra, Korn Khanal, Bikram Rivas, Pablo |
| author_facet | Sooksatra, Korn Khanal, Bikram Rivas, Pablo |
| contents | Recently, with the advancement of deep learning, several applications in text classification have advanced significantly. However, this improvement comes with a cost because deep learning is vulnerable to adversarial examples. This weakness indicates that deep learning is not very robust. Fortunately, the input of a text classifier is discrete. Hence, it can prevent the classifier from state-of-the-art attacks. Nonetheless, previous works have generated black-box attacks that successfully manipulate the discrete values of the input to find adversarial examples. Therefore, instead of changing the discrete values, we transform the input into its embedding vector containing real values to perform the state-of-the-art white-box attacks. Then, we convert the perturbed embedding vector back into a text and name it an adversarial example. In summary, we create a framework that measures the robustness of a text classifier by using the gradients of the classifier. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_03789 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | On Adversarial Examples for Text Classification by Perturbing Latent Representations Sooksatra, Korn Khanal, Bikram Rivas, Pablo Machine Learning Artificial Intelligence Computation and Language Cryptography and Security 68T01, 68T50 I.2.7 Recently, with the advancement of deep learning, several applications in text classification have advanced significantly. However, this improvement comes with a cost because deep learning is vulnerable to adversarial examples. This weakness indicates that deep learning is not very robust. Fortunately, the input of a text classifier is discrete. Hence, it can prevent the classifier from state-of-the-art attacks. Nonetheless, previous works have generated black-box attacks that successfully manipulate the discrete values of the input to find adversarial examples. Therefore, instead of changing the discrete values, we transform the input into its embedding vector containing real values to perform the state-of-the-art white-box attacks. Then, we convert the perturbed embedding vector back into a text and name it an adversarial example. In summary, we create a framework that measures the robustness of a text classifier by using the gradients of the classifier. |
| title | On Adversarial Examples for Text Classification by Perturbing Latent Representations |
| topic | Machine Learning Artificial Intelligence Computation and Language Cryptography and Security 68T01, 68T50 I.2.7 |
| url | https://arxiv.org/abs/2405.03789 |