Injecting Undetectable Backdoors in Obfuscated Neural Networks and Language Models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929490394873856 |
|---|---|
| author | Kalavasis, Alkis Karbasi, Amin Oikonomou, Argyris Sotiraki, Katerina Velegkas, Grigoris Zampetakis, Manolis |
| author_facet | Kalavasis, Alkis Karbasi, Amin Oikonomou, Argyris Sotiraki, Katerina Velegkas, Grigoris Zampetakis, Manolis |
| contents | As ML models become increasingly complex and integral to high-stakes domains such as finance and healthcare, they also become more susceptible to sophisticated adversarial attacks. We investigate the threat posed by undetectable backdoors, as defined in Goldwasser et al. (FOCS '22), in models developed by insidious external expert firms. When such backdoors exist, they allow the designer of the model to sell information on how to slightly perturb their input to change the outcome of the model.
We develop a general strategy to plant backdoors to obfuscated neural networks, that satisfy the security properties of the celebrated notion of indistinguishability obfuscation. Applying obfuscation before releasing neural networks is a strategy that is well motivated to protect sensitive information of the external expert firm. Our method to plant backdoors ensures that even if the weights and architecture of the obfuscated model are accessible, the existence of the backdoor is still undetectable.
Finally, we introduce the notion of undetectable backdoors to language models and extend our neural network backdoor attacks to such models based on the existence of steganographic functions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_05660 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Injecting Undetectable Backdoors in Obfuscated Neural Networks and Language Models Kalavasis, Alkis Karbasi, Amin Oikonomou, Argyris Sotiraki, Katerina Velegkas, Grigoris Zampetakis, Manolis Machine Learning Cryptography and Security As ML models become increasingly complex and integral to high-stakes domains such as finance and healthcare, they also become more susceptible to sophisticated adversarial attacks. We investigate the threat posed by undetectable backdoors, as defined in Goldwasser et al. (FOCS '22), in models developed by insidious external expert firms. When such backdoors exist, they allow the designer of the model to sell information on how to slightly perturb their input to change the outcome of the model. We develop a general strategy to plant backdoors to obfuscated neural networks, that satisfy the security properties of the celebrated notion of indistinguishability obfuscation. Applying obfuscation before releasing neural networks is a strategy that is well motivated to protect sensitive information of the external expert firm. Our method to plant backdoors ensures that even if the weights and architecture of the obfuscated model are accessible, the existence of the backdoor is still undetectable. Finally, we introduce the notion of undetectable backdoors to language models and extend our neural network backdoor attacks to such models based on the existence of steganographic functions. |
| title | Injecting Undetectable Backdoors in Obfuscated Neural Networks and Language Models |
| topic | Machine Learning Cryptography and Security |
| url | https://arxiv.org/abs/2406.05660 |