Steganalysis of adaptive multi-rate speech streams with distributed representations of codewords

Qiu, Yiqin; Tian, Hui; Jia-liang, Xie; Mazurczyk, Wojciech; Chang, Chin‐Chen

doi:10.1016/j.jisa.2022.103250

Cited by 3 publications

(7 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We implement four types of steganalysis models based on RNN-SM [7], FCEM [11], and DRCM [12], respectively, including 1.…”

Section: Experimental Results and Performance Evaluationmentioning

confidence: 99%

“…In our experiments, we representatively port three state-of-the-art steganalysis models (i.e., RNN-SM [7], FCEM [11], and DRCM [12]) into FedSpy, although FedSpy can be extended to many other speech steganalysis methods based on deep learning. Like in RNN-SM, FCEM, and DRCM, we take the Complementary Neighbor Vertices (CNV) algorithm [27], a typical approach based on QIM [28], as our benchmark target.…”

Section: Basic Steganalysis and Target Steganography Methodsmentioning

confidence: 99%

“…In follow-up work, Yang et al proposed three more effective steganalysis methods based on the teacher-student model [9], the attention mechanism [10], and the multi-head attention mechanism [11], respectively, wherein the model in [11] (named FCEM) has the best performance in terms of both detection accuracy and detection efficiency. Recently, Qiu et al [12] proposed a novel steganalysis model with distributed representations of codewords based on codeword embedding, Bi-LSTM, and Multi-Layer Perceptron, to further improve the detection accuracy in short-length and low-embedding-rate speech streams, and achieved state-of-the-art performance. (Note that we refer to this model as DRCM in this paper.)…”

Section: Related Workmentioning

confidence: 99%

“…Therefore, speech steganalysis [5,6], whose primary purpose is to detect the existence of hidden messages in speech signals, has been attracting increasing attention in recent years. Particularly, with the development of artificial intelligence, many well-performing speech steganalysis methods based on deep learning [7][8][9][10][11][12][13][14][15][16][17][18] have been proposed, since deep learning can capture the subtle differences between the steganographic and cover samples.…”

Section: Introductionmentioning

confidence: 99%

“…FedSteg-FCEM or FedSteg-DRCM), incorporating RNN-SM (resp. FCEM or DRCM) into FedSteg [12]. In FedSteg, each client has a personalized steganalysis model after transfer learning.…”

mentioning

confidence: 99%

See 4 more Smart Citations

FedSpy: A Secure Collaborative Speech Steganalysis Framework Based on Federated Learning

et al. 2023

View full text Add to dashboard Cite

Deep learning brings the opportunity to achieve effective speech steganalysis in speech signals. However, the speech samples used to train speech steganalysis models (i.e., steganalyzers) are usually sensitive and distributed among different agencies, making it impractical to train an effective centralized steganalyzer. Therefore, in this paper, we present an effective framework, named FedSpy, using federated learning, which enables multiple agencies to securely and jointly train the speech steganalysis models without sharing their speech samples. FedSpy is a flexible and extensible framework that can work effectively in conjunction with various deep-learning-based speech steganalysis methods. We evaluate the performance of FedSpy by detecting the most widely used Quantization Index Modulation-based speech steganography with three state-of-the-art deep-learning-based steganalysis methods representatively. The results show that FedSpy significantly outperforms the local steganalyzers and achieves good detection accuracy comparable to the centralized steganalyzer.

show abstract

“…We implement four types of steganalysis models based on RNN-SM [7], FCEM [11], and DRCM [12], respectively, including 1.…”

Section: Experimental Results and Performance Evaluationmentioning

confidence: 99%

Section: Basic Steganalysis and Target Steganography Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…FedSteg-FCEM or FedSteg-DRCM), incorporating RNN-SM (resp. FCEM or DRCM) into FedSteg [12]. In FedSteg, each client has a personalized steganalysis model after transfer learning.…”

mentioning

confidence: 99%

See 3 more Smart Citations