SGSonghui Guo

Songhui Guo

*This is an algorithmically generated author record based on publicly available information.

Songhui Guo has a total of 12 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A Lightweight Cancelable Biometric Template Protection Scheme

    by Shuaichao Song, Songhui Guo, Yeming Yang, Miao Yu, Ruiyang Ding

  • Open Access

    Three-Party Password Authentication and Key Exchange Protocol Based on MLWE

    by Songhui Guo, Yunfan Song, Song Guo, Yeming Yang, Shuaichao Song

    With the rapid development of quantum theory, the discrete logarithm problem and significant integer factorization problem have polynomial solution algorithms under quantum computing, and their security is seriously threatened. Therefore, a three-party password-authenticated key agreement scheme based on module learning with errors problem was proposed, and its security was proved in the BPR model. Compared with other password-authenticated key agreement protocols, the proposed protocol has higher efficiency and a shorter key length, which can resist quantum attacks. Therefore, the protocol is efficient and secure and suitable for large-scale network communication.


  • Open Access

    Evading Logits-Based Detections to Audio Adversarial Examples by Logits-Traction Attack

    by Songshen Han, Kaiyong Xu, Songhui Guo, Miao Yu, Bo Yang

    Automatic Speech Recognition (ASR) provides a new way of human-computer interaction. However, it is vulnerable to adversarial examples, which are obtained by deliberately adding perturbations to the original audios. Thorough studies on the universal feature of adversarial examples are essential to prevent potential attacks. Previous research has shown classic adversarial examples have different logits distribution compared to normal speech. This paper proposes a logit-traction attack to eliminate this difference at the statistical level. Experiments on the LibriSpeech dataset show that the proposed attack reduces the accuracy of the LOGITS NOISE detection to 52.1%. To further verify the effectiveness of this approach in attacking detection based on logits, three different features quantifying the dispersion of logits are constructed in this paper. Furthermore, a richer target sentence is adopted for experiments. The results indicate that these features can detect baseline adversarial examples with an accuracy of about 90% but cannot effectively detect Logits-Traction adversarial examples, proving that Logits-Traction attack can evade the logits-based detection method.