Existing solutions on differentially private deep learning (DPDL) either require the assumption of a trusted data server (centralized DPDL) or suffer from poor utility (local DPDL); and hence their adoptions are hampered in real-world scenarios.We present CRYPTDP, a crypto-assisted differentially private deep learning approach in the two-server model. CRYPTDP employs two non-colluding servers to collaboratively and efficiently train differentially private deep learning over the secret shares of data owners' private data while protecting the confidentiality of the data from untrusted servers. CRYPTDP is the first approach with the best of both local DPDL and centralized DPDL models, which does not resort to trusted server like local DPDL and has the utility like centralized DPDL. In particular, we also make innovations for addressing the major challenges like poor performance and security that beset CRYPTDP: We introduce a new secure computation and differential privacy friendly activation function; we propose a novel garbled-circuits-free most significant bit extraction protocol, and using the protocol we propose an efficient and secure garbled-circuits-free protocol for activation function over secret shares. Exhaustive experiments show that CRYPTDP delivers significantly better performance than the state-of-the-art local DPDL, yields higher accuracy than the state-of-the-art centralized DPDL, and can achieve two orders of magnitude faster runtime than the state-of-the-art approach.

Large-scale pre-trained vision-language models (VLMs) like CLIP show exceptional performance and zero-shot generalization. However, their reliability may be severely undermined by a critical vulnerability to subtle adversarial perturbations. Our work reveals a critical cross-modal vulnerability: visual-only perturbations induce substantial, synchronous shifts in decision attribution maps across both image and text. This phenomenon signifies a fundamental disruption of the VLM's internal logic, as it alters both the model's perceptual focus and its decision rationale. To counter this vulnerability, we introduce Cross-modal Bidirectional Attribution guided Few-shot Adversarial Prompt Tuning (CBA-FAPT), a novel method that leverages the model's internal decision rationale as a regularizer for robust learning. Our framework's core mechanism is the alignment of a novel bidirectional attribution map. This map is a unique fusion of two components. It combines forward feature attention to capture the model's perceptual focus. It also incorporates backward decision gradients to act as a proxy for the model's decision rationale, quantifying how each feature influences the final outcome. We enforce consistency on this bidirectional map between clean and adversarial examples. This approach corrects the model's internal logic on two fronts and effectively restores its adversarial robustness. Comprehensive experiments on 11 datasets demonstrate that CBA-FAPT outperforms the state-of-the-art, establishing a superior trade-off between robust and natural accuracy.

Multi-modal Time Series (MTS) is a vital ingredient to Predictive Multi-modal Artificial Intelligence (PMAI). MTS systems capture varying temporal modalities and their inherent dependencies for their accurate analytics. However, efficiently exploring these cross-modalities relationships is a challenging research due to their complexity facets and information redundancies. MTS patterns' pairwise similarity measures precede PMAI. Multi-modal Dynamic Time Warping (MDTW) is frequently explored to quantify similar MTS. Yet, it's reliant on the orthogonal conditioned local similarity measures that ignore the contributions of MTS' underlying structural relationships in the warping process and, hence, susceptible to unrealistic matching. This paper addresses the setbacks by recommending a scalable MTS recognition model, named Tensor-Slices Distance (TSD)-based MDTW (TSD-MDTW), that's subsequently advanced to two more distinct models termed Weighted modality and TSD (WmTSD-MDTW) and TSD-Mahalanobis (TSDMaha-MDTW). To quantify an alignment's cost, TSD-MDTW incorporates intrinsic spatial dependencies between modalities' coordinates, while WmTSD-MDTW relaxes information redundancies through weighing modalities based on information richness, whereas TSDMaha-MDTW embodies modalities dependencies and their coordinates' innate spatial dependencies. Besides, it proposes a scalable Tensor-based DTW (TDTW) model that re-formulates MDTW into multiple dimensions that are found paralleling warping processes. Theoretical and empirical experimental results on MTS multi-modal datasets encompassing load patterns and meteorological modalities reveal TDTW's efficiency and proposals' superior performances in terms of cluster compactness and separation over MDTW employing the state-of-the-art local similarity measures.

Secure two-party neural network (2P-NN) inference allows the server with a neural network model and the client with inputs to perform neural network inference without revealing their private data to each other. However, the state-of-the-art 2P-NN inference still suffers from large computation and communication overhead especially when used in ImageNet-scale deep neural networks. In this work, we design and build Panther, a lightweight and efficient secure 2P-NN inference system, which has great efficiency in evaluating 2P-NN inference while safeguarding the privacy of the server and the client. At the core of Panther, we have new protocols for 2P-NN inference. Firstly, we propose a customized homomorphic encryption scheme to reduce burdensome polynomial multiplications in the homomorphic encryption arithmetic circuit of linear protocols. Secondly, we present a more efficient and communication concise design for the millionaires’ protocol, which enables non-linear protocols with less communication cost. Our evaluations over three sought-after varying-scale deep neural networks show that Panther outperforms the state-of-the-art 2P-NN inference systems in terms of end-to-end runtime and communication overhead. Panther achieves state-of-the-art performance with up to 24.95× speedup for linear protocols and 6.40× speedup for non-linear protocols in WAN when compared to prior arts.

With the rapid development and widespread application of information, computer, and communication technologies, Cyber-Physical-Social Systems (CPSS) have gained increasing importance and attention. To enable intelligent applications and provide better services for CPSS users, efficient data analytical models are crucial. This paper presents a novel data analytic framework for CPSS services. First, a Tensor-Based Factorial Hidden Markov Model (T-FHMM) is introduced to comprehensively analyze multi-user activity features, enhancing CPSS activity analytics. A tensor-based Forward-Backward algorithm is then designed for T-FHMM to efficiently perform evaluation tasks using multiple probabilistic computing micro-services. Additionally, a tensor-based Baum-Welch algorithm is developed to accurately learn model parameters via parameter optimization micro-services. Furthermore, a tensor-based Viterbi algorithm is implemented with specific micro-services to improve prediction tasks. Finally, the comprehensive performance of the proposed model and algorithms is validated on three open datasets through self-comparison and other-comparison. Experimental results demonstrate that the proposed method outperforms compared methods in terms of accuracy, precision, recall, and F1-score.

Cyber-Physical-Social Systems (CPSSs) represent a transformative paradigm that integrates human, machine, and environmental interactions to support intelligent services in smart spaces. However, providing accurate and efficient user activity analysis in such environments remains challenging due to the complex, high-dimensional, and noisy nature of sensory data. Although existing tensor-based models have shown promising accuracy in user activity analysis, they often suffer from low efficiency and reduced robustness to data noise, limiting their practicality in real-time applications. To address these challenges, this study proposes a Tensor-based Hidden Semi-Markov Model (T-HSMM) designed to efficiently analyze user activity durations and their dependencies using probabilistic distributions in tensor space. The main objective is to reduce redundant tensor computations while enhancing both the accuracy and robustness of activity analysis. Moreover, to effectively address the three basic micro-services in CPSSs—evaluation, learning, and prediction—we develop tensor-based algorithms, including the Forward-Backward, Baum-Welch, and Viterbi algorithms, for the proposed T-HSMM. These algorithms facilitate three computational subtasks of activity sequence probabilities, model parameter learning, and activity prediction. We evaluated the performance of the proposed model on three widely used open datasets. The results show that T-HSMM surpasses other models in terms of accuracy, precision, recall, and F1-score while maintaining acceptable time consumption. Additionally, we discuss the impact of varying parameters on the model's performance across different daily activities.

Recurrent neural network (RNN), a branch of deep learning, is a powerful model for sequential data that has outstanding performance on a wide range of important Internet of Things (IoT) tasks. This unprecedented growth of RNN model has however encountered both heterogeneous IoT data and privacy issues. Existing RNN model can not deal with heterogeneous sequential data; often the larger datasets used in training of RNN model contain sensitive information. To tackle these challenges and for the first time, this research proposes a novel differentially private tensor-based RNN (DPTRNN) that can be applied in many challenging deep learning sequence tasks for IoT systems. Specifically, to process heterogeneous sequential data, we propose a tensor-based RNN model. To guarantee privacy, we develop a tensor-based back-propagation through time algorithm with perturbation to avoid exposing the sensitive information for training the tensor-based RNN model within the framework of differential privacy. Thorough security analysis shows that the differential private tensor-based RNN efficiently protects the confidentiality of sensitive user information for IoT. Our results from extensive experiments on two challenging large video datasets suggest that our proposed scheme is practical with guarantee of data privacy preservation and acceptable accuracy loss.

The development and applications of artificial intelligence (AI) have brought unprecedented opportunities to humans, but also brought many challenges and concerns such as unfairness, immorality, distrust, illegality, and discrimination. Responsible AI provides a new solution to effectively address these AI potential threats by integrating social/physical rules into AI systems. However, these rules are high-level regulations and ethical principles, which are difficult to be formalized. To this end, we attempt to use the data generated in various AI systems such as cyber–physical–social systems (CPSS) to discover and reflect these rules to provide more responsible services for humans. In this article, we first propose a data-driven responsible CPSS framework. Its core idea is to mine valuable rules through perception, fusion, processing, and analysis of CPSS data, and then use these rules to adaptively optimize CPSS. Based on this framework, three tensor-based couple hidden Markov models (T-CHMMs) are constructed to integrate three responsible features (i.e., timing, periodicity, and correlation) for mining potential and valuable rules. Then, the corresponding tensor-based Baum–Welch (TBW) algorithms are designed to solve their learning problems. Finally, the predictive accuracy and computational efficiency of the proposed models and algorithms are verified on three open datasets. The experimental results show that proposed methods have the best performances for various scenarios, which reflects that our methods are more promising and responsible than existing methods.

There are three key roles in Intelligent Transportation Systems (ITS): driver, vehicle and road. However, existing static interactions among Driver-Vehicle-Road (DVR) are too passive to reflect the change of driver preferences, vehicle conditions, road conditions, etc. Therefore, we provide a data-driven Cloud-Fog-Edge Collaborative Driver-Vehicle-Road (CFEC-DVR) framework. The framework could self-adaptively evolves through continuous iteration to provide better ITS services for humans. The collaboration among DVR creates a lot of relation data that construct our relation graphs. Cloud brings some privacy risks. Relation graphs have great analytic value. As DVR collaboration, privacy quality and analytic accuracy are three key issues in the framework, we propose a Relation Graph Privacy-Preserving scheme with High Accuracy in our framework, which is named as RGPP-HA. Based on machine learning, our method nearly maximizes the difficulty for attackers to know exactly how many other roles are connected to the attacked role, which enhances the privacy quality. Meanwhile, we find as much valuable information as possible from roles’ encrypted relations for more accurately analytic performance. Based on the experiments, we compare the proposed scheme RGPP-HA with existing classic and relevant schemes. The experimental results show that our scheme has the best privacy quality and analytic accuracy. This further verifies the feasibility of CFEC-DVR framework.

In the recent past, deep learning has received remarkable acceptance in real-world applications. Social computing expands the existing notion of cyber space and physical space to a more advance cyber-physical-social system (CPSS). Therefore, deep learning provides a propitious technique for accurate mining of information from CPSS, thus facilitates CPSS to offer services of exceptional quality efficiently. However, most of the current deep learning methods are struggling to keep up with the ever-increasing heterogeneous and highly nonlinear dissemination of data. Furthermore, the advancement of deep learning presents privacy concerns. This article proposes a deep private tensor autoencoder (dPTAE), where tensors are used for data representation, and differential privacy guarantees strong privacy. The core idea of our work is to enforce differential privacy through noise injection into the objective functions instead of the results they produce. In addition, the proposed method preserves the privacy of information shared amongst CPSS in smart environments. We applied dPTAE on three representative data sets. Rigorous experimental evaluations and theoretical analysis demonstrate that dPTAE is significantly effective and efficient.

With the advent of the 5G era, the Internet of Things (IoT) will flourish in the future. Millions of IoT devices will be connected by 5G networks, which will bring great challenges to network management. Software-defined network (SDN) is a novel solution for managing a large number of IoT devices over the network. In order to solve the problems of secure data analysis in SDN-based IoT, a differentially private tensor computing model (DPTCM) is proposed in this article. Our approach utilizes tensor to model and analyze the SDN-based IoT big data, and an algorithm named differentially private tensor train decomposition (DPTTD) is proposed to achieve secure computing in SDN-based IoT. The algorithm can make full use of the flexible computing power of edge-cloud computing so as to realize collaborative computing between edges, cloud, and the third party. By separating the calculation process of the private data and nonprivate data, the algorithm implements the localized calculation and preservation of the original data which can protect the data security from the source. Meanwhile, we use the differential privacy technology to protect the privacy of data transmitted to the cloud. Finally, we prove that the algorithm satisfies $\varepsilon $ -differential privacy. In the experiments, we verify our model on two real-world data sets. The experimental results show that differential privacy has a little side effect on prediction results, and our model has good performance in data prediction.

Secure cyber-physical-social big data computations are being increasingly used to protect the users' data security in cyber-physical-social systems (CPSS). Despite the increasing popularity, how to process the tasks of the secure cyber-physical-social big data computations, while taking care of the energy consumption and meeting the users' requirements, remains challenging. To address the problem, in this work, we propose a novel tensor-based optimization model for the secure sustainable cyber-physical-social big data computations. The proposed model is a general and fine-grained model, which can jointly optimize the execution time, energy consumption, reliability, and quality of experience, and can comprehensively take into account step, task, time slot, type, node, core, cryptosystem, and security level. To our knowledge, this is the first study to holistically optimize the tasks in the secure cyber-physical-social big data computations. To illustrate the proposed model, the case study of the secure high-order Lanczos in cloud-assisted CPSS is presented. Finally, the proposed model is empirically evaluated by using multi-objective optimization, and the extensive results demonstrate that from the users' perspective our proposed tensor-based optimization model is preferable for the secure sustainable cyber-physical-social big data computations.