Federated learning is a promising approach for training machine learning models while preserving data privacy. However, its distributed nature makes it vulnerable to backdoor attacks, particularly in NLP tasks, where related research remains limited. This paper introduces SDBA, a novel backdoor attack mechanism designed for NLP tasks in federated learning environments. Through a systematic analysis across LSTM and GPT-2 models, we identify the most vulnerable layers for backdoor injection and achieve both stealth and long-lasting durability by applying layer-wise gradient masking and top-k% gradient masking. Also, to evaluate the task generalizability of SDBA, we additionally conduct experiments on the T5 model. Experiments on next-token prediction, sentiment analysis, and question answering tasks show that SDBA outperforms existing backdoors in terms of durability and effectively bypasses representative defense mechanisms, demonstrating notable performance in transformer-based models such as GPT-2. These results highlight the urgent need for robust defense strategies in NLP-based federated learning systems.

As communication technology advances, various and heterogeneous data are communicated in distributed environments through network systems. Meanwhile, along with the development of communication technology, the attack surface has expanded, and concerns regarding network security have increased. Accordingly, to deal with potential threats, research on network intrusion detection systems (NIDSs) has been actively conducted. Among the various NIDS technologies, recent interest is focused on artificial intelligence (AI)-based anomaly detection systems, and various models have been proposed to improve the performance of NIDS. However, there still exists the problem of data imbalance, in which AI models cannot sufficiently learn malicious behavior and thus fail to detect network threats accurately. In this study, we propose a novel AI-based NIDS that can efficiently resolve the data imbalance problem and improve the performance of the previous systems. To address the aforementioned problem, we leveraged a state-of-the-art generative model that could generate plausible synthetic data for minor attack traffic. In particular, we focused on the reconstruction error and Wasserstein distance-based generative adversarial networks, and autoencoder-driven deep learning models. To demonstrate the effectiveness of our system, we performed comprehensive evaluations over various data sets and demonstrated that the proposed systems significantly outperformed the previous AI-based NIDS.

Traditional public key exchange protocols are based on algebraic number theory. In another perspective, neural cryptography, which is based on neural networks, has been emerging. It has been reported that two parties can exchange secret key pairs with the synchronization phenomenon in neural networks. Although there are various models of neural cryptography, called Tree Parity Machine (TPM), many of them are not suitable for practical use, considering efficiency and security. In this paper, we propose a Vector-Valued Tree Parity Machine (VVTPM), which is a generalized architecture of TPM models and can be more efficient and secure for real-life systems. In terms of efficiency and security, we show that the synchronization time of the VVTPM has the same order as the basic TPM model, and it can be more secure than previous results with the same synaptic depth.

Over the recent years, cloud storage services have become increasingly popular, where users can outsource data and access the outsourced data anywhere, anytime. Accordingly, the data in the cloud is growing explosively. Among the outsourced data, most of them are duplicated. Cloud storage service providers can save huge amounts of resources via client-side deduplication. On the other hand, for safe outsourcing, clients who use the cloud storage service desire data integrity and confidentiality of the outsourced data. However, ensuring confidentiality and integrity in the cloud storage environment can be difficult. Recently, in order to achieve integrity with deduplication, the notion of deduplicatable proof of storage has emerged, and various schemes have been proposed. However, previous schemes are still inefficient and insecure. In this paper, we propose a symmetric key based deduplicatable proof of storage scheme, which ensures confidentiality with dictionary attack resilience and supports integrity auditing based on symmetric key cryptography. In our proposal, we introduce a bit-level challenge in a deduplicatable proof of storage protocol to minimize data access. In addition, we prove the security of our proposal in the random oracle model with information theory. Implementation results show that our scheme has the best performance.