Home Latest Articles
Latest Articles
  • Yuhong Ke, Chao Lin, Xinyi Huang, Wei Wu, Yujie Chen
    Journal of Computer Research and Development. 2024, 61(10): 2514-2525.

    In the digital era, data have become a core asset for the functioning of society, and identity authentication credentials are among the most critical and sensitive data elements. Traditional password-based authentication methods require servers to store credential information such as usernames and passwords, which poses a serious risk of data leakage. Passwordless authentication technology based on public-key cryptography replaces traditional passwords with public-private key pairs. Users employ their private keys to compute signatures for authentication information, while servers only store public information like public keys, thus eliminating the issue of servers leaking private key information. However, existing passwordless authentication systems face challenges such as incompatibility across multiple platforms, high latency in online authentication, and the difficulty of recovering private keys when devices are lost. Moreover, the transparency and auditability of these systems need improvement. To address these problems, we propose an efficient, multi-platform compatible, passwordless identity authentication scheme based on blockchain technology. The scheme combines FIDO2 passwordless authentication with blockchain, allowing users to generate and upload multiple account public keys to the blockchain network for public verification by service providers. Through optimizations such as offline account pre-registration, pre-computation of signatures, and on-chain data synchronization, the scheme achieves interoperability, low overhead, and scalability for large-scale user authentication. The scheme also incorporates an encrypted backup mechanism, enabling users to recover backup data using stored encrypted keys even if their devices are lost. Furthermore, the scheme leverages the immutable data storage provided by blockchain, allowing all participants to query the status of authentication authorizations, thus enhancing system transparency. We comprehensively evaluate the security and performance of the proposed scheme. Theoretical analysis and experiments show that the proposed scheme reduces online computational overhead by 89.09% and communication overhead by 85.57% compared with similar schemes while maintaining low-latency responses under high-load conditions.

  • Longlong Zhu, Xiang Chen, Haodong Chen, Jitang Niu, Wenjing Liu, Shengrui Lin, Dong Zhang, Chunming Wu
    Journal of Computer Research and Development. 2024, 61(10): 2526-2539.

    Blockchain suffers from network dynamics and management difficulties, making the anomalies such as DDoS attacks and account takeovers possible. Existing approaches that detect anomalies in blockchains extract features, such as historical transaction information and transaction frequencies, from blockchain accounts to identify anomalies. However, the increasing scale of blockchain data results in the challenge of high memory consumption and low detection accuracy in the feature extraction of existing approaches. To address this challenge, we propose a blockchain anomaly detection mechanism that achieves detection accuracy and reduces resource footprints. This mechanism embraces approximate sketching algorithms to transform the detection of blockchain anomalies into that of malicious accounts, including intra-block accounts and inter-block accounts. For intra-block accounts, i.e., the malicious accounts that occur inside a single block and the mechanism uses sketching algorithms to collectively filter out those accounts with high precision. For inter-block accounts, malicious accounts can be hardly detected by analyzing the information of a single block, it aggregates multi-block information to accurately detect those accounts. We evaluate our mechanism with real Ethereum block data comprising of 88847 blocks. Our results indicate that compared with typical existing approaches, our mechanism improves the recall of detecting blockchain anomalies by up to 6.3 times and the F1 score by up to 4.4 times. Therefore, our proposed blockchain anomaly detection mechanism can bring benefits to regulating blockchain transaction behaviors and maintain system security.

  • Chuan Zhang, Zihao Wang, Jinwen Liang, Mengxuan Liu, Haotian Deng, Liehuang Zhu
    Journal of Computer Research and Development. 2024, 61(10): 2540-2553.

    With the development of data assetization in enterprises, consortium blockchains have become the core infrastructure for data transactions within enterprise alliances. Constructing a decentralized, tamper-proof, and access-controlled data trading market using consortium blockchains can enhance the credibility of data transactions and reduce the risk of data leakage. However, with the increase in enterprise alliances, how to promote the circulation of data elements across different enterprise alliances and resist security threats such as data privacy leaks, data fraud, and payment repudiation have become an urgent problem in the field of data circulation. To address this, we propose a privacy-preserving data element transaction audit scheme for cross-consortium-blockchains. This scheme uses consortium blockchains for data attestation and employs smart contracts on a relay chain for data verification and payment management. By combining Pedersen commitments and Shamir’s secret sharing techniques, we construct a zero-knowledge proof mechanism to prevent data fraud without disclosing data elements. Additionally, a bilinear mapping-based data element integrity audit mechanism is introduced, allowing for the verification of data element receipts without revealing data privacy, effectively solving the problem of payment repudiation. Through theoretical analysis and experimental evaluation, the proposed scheme’s effectiveness and reliability are verified. The experimental results in a local simulation environment demonstrate that the proposed scheme is feasible and efficient.

  • Zhenyu Li, Yong Ding, Chen Yi, Hai Liang
    Journal of Computer Research and Development. 2024, 61(10): 2554-2569.

    With the advancement of the new technological revolution and industrial transformation, secure and reliable data circulation has become a pressing issue. To address this, data exchanges have begun leveraging blockchain technology to facilitate data circulation, thereby enhancing trust and regulatory capabilities during the process. However, blockchain technology still faces challenges in handling large-scale data element exchanges efficiently. To improve efficiency in the circulation of data elements via blockchain, we propose a blockchain data exchange acceleration method. Firstly, a dynamic location model for data circulation is introduced to aid data exchanges in managing the direction of data flows. Secondly, a data element aggregation upload method is proposed, enabling deep aggregation of data elements with specified circulation destinations, thus accelerating blockchain data exchange. Thirdly, an optional automatic deduplication download method is presented to address application issues encountered in blockchain-based data circulation. Finally, experimental results demonstrate that the proposed blockchain data exchange acceleration method, when combined with blockchain transaction packaging technologies, significantly enhances the efficiency of data element exchanges. Compared with the asynchronous data exchange method provided by existing blockchain frameworks, our method achieves approximately a 9.24 times improvement in data exchange performance.

  • Jiakun Hao, Peng Xiang, Yifei He, Jianbo Gao, Zhi Guan, Anming Xie, Zhong Chen
    Journal of Computer Research and Development. 2024, 61(10): 2570-2586.

    In the digital era, data, as a critical factor of production, is essential for economic growth, technological innovation, and enhancing corporate competitiveness. Traditional centralized data trading systems pose a series of data and transaction security risks. Although decentralized data trading systems based on blockchain technology have emerged, which can guarantee security for data transactions through blockchain and smart contracts, systems based on blockchain still struggle with cross-regional, cross-domain, and international data transactions. They primarily face the following three challenges: 1) Consistency of cross-domain identity resolution results. 2) Availability of cross-domain identity resolution services. 3) Security and compliance issues for cross-domain data trading systems. To address these issues, we design a secure and efficient cross-domain identity resolution mechanism based on decentralized identity technology and construct a secure, regulated decentralized data trading system based on this mechanism. The identity resolution mechanism includes a set of cross-domain protocols that ensure the consistency of resolution results and an incentive mechanism that ensures the availability of services executing these protocols. The decentralized data trading system supports granular regulation based on identity attributes and ensures the atomicity of cross-domain data transactions. Experimental verification shows that this system meets the design targets. By batch constructing cross-domain resolution requests, the identity cross-domain resolution protocol and caching mechanism can ensure the correctness of the resolution results, with the resolution latency being only 45% of the cross-domain resolution latency. Additionally, the cache can increase the hit rate by 72% through active and passive update mechanisms. The system is also compatible with multiple blockchain systems, and cross-domain data transactions can be completed within an average of 2.3 blocks, making the system efficiency within an acceptable range.

  • Biao Jin, Xiang Lin, Jinbo Xiong, Weijing You, Xuan Li, Zhiqiang Yao
    Journal of Computer Research and Development. 2024, 61(10): 2587-2606.

    Constructing an excellent deep neural network (DNN) model requires a large amount of training data, high-performance equipment, and profound expertise and knowledge. Therefore, DNN models should be regarded as the intellectual property (IP) of their owners. Protecting the IP of a DNN model also underscores the appreciation for the value of the data elements integral to its development and training process. However, DNN models are vulnerable to attacks such as theft, tampering, and illegal dissemination by malicious users. The quest for effective strategies to protect their IP has emerged as a pivotal area of academic research and an urgent challenge confronting the industry. Unlike existing related reviews, we focus on the application scenarios of DNN model watermarking. We mainly review the methodologies for DNN model IP protection based on watermarking technology from two dimensions: robust model watermarking for model copyright declaration and fragile model watermarking for model integrity verification. We discuss their characteristics, advantages, and limitations. Additionally, we elaborate on the practical application of DNN model watermarking technology. Finally, by summarizing the common technologies of various methods, we prognosticate future research directions for DNN model IP protection.

  • Jialang Liu, Yanming Guo, Mingrui Lao, Tianyuan Yu, Yulun Wu, Yunhao Feng, Jiazhuang Wu
    Journal of Computer Research and Development. 2024, 61(10): 2607-2626.

    Federated learning is designed for data privacy and data security issues, after a large number of clients are trained locally in a distributed manner, the central server then aggregates the model parameter updates provided by each local client, but the central server is unable to see how these parameters are updated, and this feature creates a serious security issue, i.e., a malicious participant can train a poisoned model and upload the parameters in the local model, and then globally model to introduce backdoor features. In this paper, we focus on the security and robustness research under the scenarios specific to federated learning, i.e., backdoor attack and defense, summarize the scenarios that generate backdoor attacks under federated learning, summarize the latest methods of backdoor attack and defense under federated learning, and compare and analyze the performance of the various attack and defense methods, revealing their advantages and limitations. Finally, we point out various potential directions and new challenges for backdoor attacks and defenses under federated learning.

  • Zhen Qin, Tianming Zhuang, Guosong Zhu, Erqiang Zhou, Yi Ding, Ji Geng
    Journal of Computer Research and Development. 2024, 61(10): 2627-2648.

    In recent years, the rapid development of artificial intelligence technology, particularly deep learning, has led to its widespread application in various fields such as computer vision and natural language processing. However, recent research indicates potential security risks associated with these advanced AI models could compromise their reliability. In light of this concern, this survey delves into cutting-edge research findings pertaining to security attacks, attack detection, and defense strategies for artificial intelligence models. Specifically regarding model security attacks, our work focuses on elucidating the principles and technical status of adversarial attacks, model inversion attacks, and model theft attacks. With regards to model attack detection methods explored in this paper, they include defensive distillation techniques, regularization approaches, outlier detection, robust statistics. As for model defense strategies examined in this study, they encompass adversarial training measures, model structure defense mechanisms, query control defenses along with other technical means. This comprehensive survey not only summarizes but also expands upon techniques and methodologies relevant to ensuring the security of artificial intelligence models thereby providing a solid theoretical foundation for their secure applications while simultaneously enabling researchers to gain a better understanding of the current state-of-the-art research within this field facilitating informed decisions when selecting future research directions.

  • Yuyu Yin, Guangqiang Wu, Youhuizi Li, Xinyu Wang, Honghao Gao
    Journal of Computer Research and Development. 2024, 61(10): 2649-2661.

    With the accelerated advancement of digitization, data elements have become the core driving force for the operation of modern society. However, at the same time, data security issues have become increasingly prominent, with frequent occurrences of data breaches and privacy violations, causing serious losses to individuals, organizations, and even countries. Against this backdrop, the security of data elements has become the focus of attention from all sectors of society, and the issue of data privacy protection in deep learning models has also attracted widespread attention. Among them, machine unlearning, as a key technology for protecting user’s privacy, aims to enable models to remove the influence of specific data while maintaining generalization performance for remaining data, providing an effective solution for protecting the security of data elements in deep learning models. Existing machine unlearning methods are mainly divided into two categories: exact unlearning and approximate unlearning. However, exact unlearning methods need to intervene in the original training process of the models, while approximate unlearning methods find it difficult to strike a balance between unlearning performance and model generalization ability. To address these issues, we propose an approximate unlearning framework based on feature constraints and adaptive loss balancing. We adopt a “forgetting-recovering” machine unlearning framework. First, for the “forgetting” process, in order to mimic the feature outputs of retrained models for the forgetting samples, we use a randomly initialized model that has not been trained on the forgetting samples to guide the feature outputs of the unlearning model, constraining forgetting at the feature level to avoid easily obtaining forgetting data information from the model. Then, a small amount of data is used for fine-tuning to “recover” the generalization performance of models on the remaining data. At the same time, we regard the above machine unlearning framework as a multi-task optimization problem and introduce adaptive loss balance to automatically balance the “forgetting” and “recovering” tasks, preventing the model from “over-forgetting” or “over-recovering”, so that the “forgetting” and “recovering” tasks can be trained relatively balanced and steadily. Extensive experiments on 3 image classification datasets show that our method can effectively forget the forgetting data and achieve optimal performance in multiple metrics.

  • Hui Nie, Ruiping Wang, Xilin Chen
    Journal of Computer Research and Development. 2024, 61(9): 2128-2141.

    We explore the transition from closed environments to open world environments and its impact on visual perception (focusing on object recognition and detection) and the field of deep learning. In open world environments, software systems need to adapt to constantly changing conditions and demands, presenting new challenges for deep learning methods. In particular, open world visual perception requires systems to understand and process environments and objects not seen during the training phase, which exceeds the capabilities of traditional closed systems. We first discuss the dynamic and adaptive system requirements brought about by technological advances, highlighting the advantages of open systems over closed systems. Then we delve into the definition of the open world and existing work, covering five dimensions of openness: open set learning, zero-shot learning, few-shot learning, long-tail learning, and incremental learning. In terms of open world recognition, we analyze the core challenges of each dimension and provide quantified evaluation metrics for each task dataset. For open world object detection, we discuss additional challenges compared with recognition, such as occlusion, scale, posture, symbiotic relationships, background interference, etc., and emphasize the importance of simulation environments in constructing open world object detection datasets. Finally, we underscore the new perspectives and opportunities that the concept of the open world brings to deep learning, acting as a catalyst for technological advancement and deeper understanding of the realistic environment challenges, offering a reference for future research.