The integration of Domain Knowledge Graphs (DKG) into Retrieval-Augmented Generation has emerged as a promising approach for constructing Specialized Large Language Models (spLLMs).On account of the scarcity of high-quality DKGs, existing approaches employ an evolutionary framework, wherein the DKG is continuously evolved alongside its utilization for enhancing the LLM. Yet, these methods face two key limitations: 1) heavy reliance on costly expert knowledge, and 2) neglect of the connection between the LLM's inherent knowledge and external expert knowledge. To address these issues, this paper introduces Epistemic Cognition-enhanced Specialized LLMs (EC-spLLM), a novel evolutionary framework that instills epistemic cognition into the LLM to systematically exploit both its internal knowledge and expert knowledge. At the core of EC-spLLM lies the Hallucination Detection-based Epistemic Cognition (HDEC) mechanism, which assesses the reliability of LLM-generated responses using the LLM's self cognition and hallucination detection. This assessment ability enables EC-spLLM to selectively adopt either the expert-provided golden answer or a reliable LLM-generated answer during DKG evolution, thereby reducing dependence on experts and bridging internal and external knowledge sources to enhance performance. We conducted extensive experiments on five datasets spanning five domain, e.g., emotional sociology, biology, ect. Results show that EC-spLLM reduces the usage of golden answers by an average of 67% while retaining 97.2% of the accuracy achieved by the SOTA method, and outperforms all other baselines.

Cross-view geo-localization (CVGL) establishes correspondences between ground-level and satellite images of the same geographic location, serving as a fundamental technology for smart city applications, including autonomous navigation, urban planning, and location-based services. Current CVGL approaches fall into two categories: feature-based methods achieve superior performance through 2D representation learning but lack interpretability. Spatial-based methods provide geometric understanding and interpretable matching but suffer from limited spatial modeling and weak cross-view alignment, leading to lower performance. We reformulate CVGL from a spatial perspective and propose an auxiliary task-enhanced network. The network captures spatial semantics and provides explicit alignment processes with visualizable results. We introduce an auxiliary spatial semantic alignment (SSA) task that learns spatial structure via vision foundation models (VFM) and BEV transformation to enhance the primary CVGL task. The primary task captures visual semantics, including texture and appearance. Within this unified framework with shared encoders, the primary task enriches the learned embeddings by fusing spatial structure with visual semantics, yielding spatially complete representations. Extensive experiments on three standard CVGL benchmarks demonstrate that our method significantly surpasses previous spatial-based approaches while maintaining competitive performance with state-of-the-art (SOTA) feature-based methods, achieving 98.48% R@1 on CVUSA and 71.05% R@1 on CVACT\_test. We provide comprehensive analyses through pixel-level activation maps and feature-space UMAP visualizations to validate both effectiveness and interpretability.

Betweenness centrality is one of the key centrality measures in many applications including community detections in biological networks, vulnerability detections in communication networks, misinformation filtering in social networks, etc. The top-K group betweenness centrality problem is to find a group of K nodes from a network so that the total fraction of shortest paths that pass through the K nodes is maximized. Existing studies proposed randomized sampling algorithms for the problem. We notice that the existing studies ensured that, the maximum deviation of the estimated centrality of every group from its expectation is no greater than a small given threshold for all potential groups with no more than K nodes, thereby generating too many samples, as the number of such groups is prohibitively large. In contrast, in this paper we first devise a novel algorithm that enables to estimate the centrality of a tentative group adaptively, and the algorithm immediately stops once the centrality is large enough; otherwise, the algorithm uses more samples to find a better group. We then theoretically show that, even the proposed algorithm uses much less samples, it still can find a performance-guaranteed group with high probability. Experimental results with real-world networks demonstrate that the number of samples used by the proposed algorithm is up to 36 times smaller than the state-of-the-art, while the centrality of the group found by the algorithm is no more than 4.5% smaller than the latter.

Betweenness centrality is one of the key centrality measures in many applications including community detections in biological networks, vulnerability detections in communication networks, misinformation filtering in social networks, etc. The top-$K$group betweenness centrality problem is to find a group of$K$nodes from a network so that the total fraction of shortest paths that pass through the$K$nodes is maximized. Existing studies proposed randomized sampling algorithms for the problem. We notice that the existing studies ensured that, the maximum deviation of the estimated centrality of every group from its expectation is no greater than a small given threshold for all potential groups with no more than$K$nodes, thereby generating too many samples, as the number of such groups is prohibitively large. In contrast, in this paper we first devise a novel algorithm that enables to estimate the centrality of a tentative group adaptively, and the algorithm immediately stops once the centrality is large enough; otherwise, the algorithm uses more samples to find a better group. We then theoretically show that, even the algorithm uses much less samples, it still can find a performance-guaranteed group with a large success probability. Experimental results with real-world networks demonstrate that the number of samples used by the proposed algorithm is from 2 to 18 times smaller than the state-of-the-art, while the centrality of the group found by the algorithm is no more than 4% smaller than the latter.

Large language models (LLMs) have demonstrated exceptional performance across a wide variety of domains. Nonetheless, generalist LLMs continue to fall short in reasoning tasks necessitating specialized knowledge, e.g., emotional sociology and medicine. Prior investigations into specialized LLMs focused on domain-specific training, which entails substantial efforts in domain data acquisition and model parameter fine-tuning. To address these challenges, this paper proposes the Way-to-Specialist (WTS) framework, which synergizes retrieval-augmented generation with knowledge graphs (KGs) to enhance the specialized capability of LLMs in the absence of specialized training. In distinction to existing paradigms that merely utilize external knowledge from general KGs or static domain KGs to prompt LLM for enhanced domain-specific reasoning, WTS proposes an innovative ''LLM↻KG'' paradigm, which achieves bidirectional enhancement between specialized LLM and domain knowledge graph (DKG). The proposed paradigm encompasses two closely coupled components: the DKG-Augmented LLM and the LLM-Assisted DKG Evolution. The former retrieves question-relevant domain knowledge from DKG and uses it to prompt LLM to enhance the reasoning capability for domain-specific tasks; the latter leverages LLM to generate new domain knowledge from processed tasks and use it to evolve DKG. WTS closes the loop between DKG-Augmented LLM and LLM-Assisted DKG Evolution, enabling continuous improvement in the domain specialization as it progressively answers and learns from domain-specific questions. We validate the performance of WTS on 7 datasets (e.g., TweetQA, ChatDoctor5k) spanning 6 domains, e.g., emotional sociology, medical, ect. The experimental results show that WTS surpasses the previous SOTA in 5 specialized domains, and achieves a maximum performance improvement of 11.3%.

Graph neural network (GNN), the mainstream method to learn on graph data, is vulnerable to graph evasion attacks, where an attacker slightly perturbing the graph structure can fool trained GNN models. Existing work has at least one of the following drawbacks: 1) limited to directly attack two-layer GNNs; 2) inefficient; and 3) impractical, as they need to know full or part of GNN model parameters.

This paper proposes Meta Parallel Graph Neural Network (MPGNN) to establish a scalable Network Intrusion Detection System (NIDS) for large-scale Internet of Things (IoT) networks. MPGNN leverages a meta-learning framework to optimize the parallelism of GNN-based NIDS. The core of MPGNN is a coalition formation policy that generates meta-knowledge for partitioning a massive graph into multiple coalitions/subgraphs in a way that maximizes the performance and efficiency of parallel coalitional NIDSs. We propose an offline reinforcement learning algorithm, called Graph-Embedded Adversarially Trained Actor-Critic (G-ATAC), to learn a coalition formation policy that jointly optimizes intrusion detection accuracy, communication overheads, and computational complexities of coalitional NIDSs. In particular, G-ATAC learns to capture the temporal dependencies of network states and coalition formation decisions over offline data, eliminating the need for expensive online interactions with large IoT networks. Given generated coalitions, MPGNN employs E-GraphSAGE to establish coalitional NIDSs which then collaborate via ensemble prediction to accomplish intrusion detection for the entire network. We evaluate MPGNN on two real-world datasets. The experimental results demonstrate the superiority of our method with substantial improvements in F1 score, surpassing the state-of-the-art methods by 0.38 and 0.29 for the respective datasets. Compared to the centralized NIDS, MPGNN reduces the training time of NIDS by 41.63% and 22.11%, while maintaining an intrusion detection performance comparable to centralized NIDS.

Federated learning (FL) has attracted increasing attention in recent years due to its data privacy preservation and great applicability to large-scale user scenarios. However, when FL faces numerous clients, it is inevitable to emerge the non-independent and identically distributed (non-iid) data between clients, which brings an enormous challenge for model training and performance analysis like convergence. Besides, due to the non-iid data, the participating clients of FL tend to be extremely heterogeneous so the number of samplings among clients causes a sampling variance problem, which induces a huge variation in convergence. More importantly, although FL can foster privacy security via locally retaining the training data, if local data is secret and sensitive, FL should have more powerful privacy protection to resist the cloud server or third party to infer private information from shared models or intermediate gradients. Facing the non-iid and privacy challenges, we propose a differential privacy (DP) based non-iid FL algorithm called DPNFL to jointly tackle these two issues. Specifically, motivated by the DP and its variants, we are the first to adopt the truncated concentrated differential privacy technique under the FL scenario to more tightly track end-to-end privacy loss, while requiring less noise injection for the same level of DP. To avoid the sampling variance problem, we enable the server to sample the partial clients uniformly without replacement, which also guarantees unbiased sampling. To further improve the algorithm performance, we also propose an adaptive version of DPNFL named AdDPNFL, which adopts the adaptive optimization on the server-side to simultaneously alleviate the impact of non-iid data and DP noise on model utility. Finally, we perform extensive experiments to validate the effectiveness and superiority of our algorithms.

Deep neural networks often exhibit the overconfidence issue when encountering out-of-distribution (OOD) samples. To address this, leveraging large-scale pre-trained models like CLIP has shown promise. While CLIP has the capability to encode a vast array of interconnected concepts, current OOD detection methods based on it primarily focus on ID categories and a limited set of OOD categories. In this paper, we propose a novel approach that harnesses the power of WordNet to fully exploit the rich knowledge encapsulated within CLIP, resulting in enhanced OOD detection performance. Our methodology involves constructing a word tree that includes both in-distribution (ID) words and a large set of semantically similar OOD words selected from WordNet. By matching a test image with the concepts of the words in the word tree using CLIP, we estimate the probability of the image being classified as either ID or OOD. Furthermore, we introduce a conditional random field model to effectively handle both the parent-child and the sibling-sibling conflicts in the concept matching results. Extensive experiments under various ID/OOD settings demonstrate the effectiveness of our approach and its superiority over state-of-the-art methods.

Supporting efficient and secure location-based skyline queries on encrypted data, such as private data outsourced to cloud-based systems, remains an ongoing challenge for efficiency due to significant computational costs in the ciphertext domain. To accelerate privacy-preserving skyline queries, the secure index intuitively contributes to an increase in efficiency. However, designing such a secure index is a challenge while protecting the unlinkability of queries. Meanwhile, there exist little work that can commendably assure efficiency and security. In this paper, we demonstrate SecSky, an efficient solution for supporting secure location-based skyline queries through the secure index. To support SecSky, we devise a novel unified structure, named secure R-tree (SR-tree) index, without privacy leakage (especially indirect privacy). Subsequently, we propose a novel secure location-based dominance protocol, which is utilized to calculate the dominance relationship on the SR-tree. Using this protocol as the building block, our secure location-based skyline query protocol integrates SR-tree, permutation and perturbation techniques to facilitate query processing so as to dramatically reduce the computational overhead. Meanwhile, our proposed solution avoids compromising the privacy of datasets, queries, dominance relationship and skyline results. Finally, we analyze the complexity and security of SecSky. Findings from the experimental evaluation show that our proposed scheme outperforms several other protocols by at least 3 orders of magnitude in terms of query efficiency.

Supporting secure location-based services on encrypted data that is outsourced to cloud computing platforms remains an ongoing challenge for efficiency due to expensive ciphertext calculation overhead. Furthermore, since the clouds may not be trustworthy or even malicious, data security and result authenticity has caused huge concerns. Unfortunately, little work can enable query efficiency, dataset confidentiality and result authenticity to be commendably guaranteed. In this paper, we demonstrate the potential of supporting secure and verifiable location-based skyline queries (SVLSQ). First, we devise a novel and unified structure, named semi-blind R-tree (SR-tree), which protects the query unlinkability. Based on SR-tree, we propose an authenticated data structure, named secure and verifiable scope R-tree (SVSR-tree). Then, we develop several secure protocols based on SVSR-tree to accelerate the query efficiency and reduce the size of verification objects. Our method avoids compromising the privacy of datasets, queries, results and access patterns. Meanwhile, it authenticates the soundness and completeness of the skyline results while preserving privacy. Finally, we analyze the complexity and security of SVLSQ. Findings from the performance evaluation illustrate that SVLSQ is a dramatically efficient method in terms of query (no less than 3 orders of magnitude faster than other solutions) and verification.

Triangle count is a critical parameter in mining relationships among people in social networks. However, directly publishing the findings obtained from triangle counts may bring potential privacy concern, which raises great challenges and opportunities for privacy-preserving triangle counting. In this paper, we choose to use differential privacy to protect triangle counting for large scale graphs. To reduce the large sensitivity caused in large graphs, we propose a novel graph projection method that can be used to obtain an upper bound for sensitivity in different distributions. In particular, we publish the triangle counts satisfying the node-differential privacy with two kinds of histograms: the triangle count distribution and the cumulative distribution. Moreover, we extend the research on privacy preserving triangle counting to one of its applications, the local clustering coefficient. Experimental results show that the cumulative distribution can fit the real statistical information better, and our proposed mechanism has achieved better accuracy for triangle counts while maintaining the requirement of differential privacy.

Autonomous on-demand services, such as GOGOX (formerly GoGoVan) in Hong Kong, provide a platform for users to request services and for suppliers to meet such demands. In such a platform, the suppliers have autonomy to accept or reject the demands to be dispatched to him/her, so it is challenging to make an online matching between demands and suppliers. Existing methods use round-based approaches to dispatch demands. In these works, the dispatching decision is based on the predicted response patterns of suppliers to demands in the current round, but they all fail to consider the impact of future demands and suppliers on the current dispatching decision. This could lead to taking a suboptimal dispatching decision from the future perspective. To solve this problem, we propose a novel demand dispatching model using deep reinforcement learning. In this model, we make each demand as an agent. The action of each agent, i.e., the dispatching decision of each demand, is determined by a centralized algorithm in a coordinated way. The model works in the following two steps. (1) It learns the demand’s expected value in each spatiotemporal state using historical transition data. (2) Based on the learned values, it conducts a Many-To-Many dispatching using a combinatorial optimization algorithm by considering both immediate rewards and expected values of demands in the next round. In order to get a higher total reward, the demands with a high expected value (short response time) in the future may be delayed to the next round. On the contrary, the demands with a low expected value (long response time) in the future would be dispatched immediately. Through extensive experiments using real-world datasets, we show that the proposed model outperforms the existing models in terms of Cancellation Rate and Average Response Time.

Nowadays, the booming demand of big data analytics and the constraints of computational ability and network bandwidth have made it difficult for a stand-alone agent/service provider to provide suitable information for every user from the large volume online data within the limited time. To handle this challenge, a recommender system (RS) can call in a group of agents to collaborate to learn users' preference and taste, which is known as a distributed recommender system (DRS). DRSs can improve the accuracy of a traditional RS by requesting agents to share information with each other. However, it is challenging for DRSs to make personalized recommendations for each user due to the large amount of candidates. In addition, information sharing among agents raises a privacy concern. Thus, we propose a privacy-preserving DRS in this paper, and then model each service provider as a distributed online learner with context-awareness. Service providers collaborate to make personalized recommendations by learning users' preferences according to the user context and users' history behaviors. We adopt the federated learning framework to help train a high quality privacy- preserving centralized model over a large number of distributed agents which is probably unreliable with relatively slow network connections. To handle big data scenario, we build an item-cluster tree to deal with online and increasing datasets from top to the bottom. We further consider the structure of social network and present an efficient algorithm to avoid more performance loss adaptively. Theoretical proofs show that our proposed algorithm can achieve sublinear regret and differential privacy protection simultaneously for service providers and users. Numerical results confirm that our novel framework can handle increasing big datasets and strike a trade-off between privacy-preserving level and the prediction accuracy.

The consistency of a response to a given post at the semantic level and emotional level is essential for a dialogue system to deliver humanlike interactions. However, this challenge is not well addressed in the literature, since most of the approaches neglect the emotional information conveyed by a post while generating responses. This article addresses this problem and proposes a unified end-to-end neural architecture, which is capable of simultaneously encoding the semantics and the emotions in a post and leveraging target information to generate more intelligent responses with appropriately expressed emotions. Extensive experiments on real-world data demonstrate that the proposed method outperforms the state-of-the-art methods in terms of both content coherence and emotion appropriateness.

In this paper, we study the privacy of online health data. We present a novel online health data De-Anonymization (DA) framework, named De-Health. Leveraging two real world online health datasets WebMD and HealthBoards, we validate the DA efficacy of De-Health. We also present a linkage attack framework which can link online health/medical information to real world people. Through a proof-of-concept attack, we link 347 out of 2805 WebMD users to real world people, and find the full names, medical/health information, birthdates, phone numbers, and other sensitive information for most of the re-identified users. This clearly illustrates the fragility of the privacy of those who use online health forums.

The explosive growth of e-commerce has led to the development of the recommendation system. The recommendation system aims to provide a set of items that meet users’ personalized needs through analyzing users’ consumption records. However, the timeliness of purchasing data and the implicity of feedback data pose severe challenges for the existing recommendation methods. To alleviate these challenges, we exploit the user’s consumption records from the perspectives of user and item, by modeling the data on both item and user level, where the item-level value reflects the grade of item, and the user-level value reflects the user’s purchase intention. In this article, we collect the description information and the reviews of the items from public websites, then adopt sentiment analysis techniques to model the similarities on user level and item level, respectively. In particular, we extend the traditional latent factor model and propose two novel methods— I tem L evel Similarity M atrix F actorization (ILMF) and U ser L evel Similarity M atrix F actorization (ULMF)—by introducing two novel similarity measure methods. In ILMF and ULMF, the consistency between latent factors and explicit aspects is naturally incorporated into learning latent factors of the users and items, such that we can predict the users’ preferences on different items more accurately. Moreover, we propose I tem- U ser L evel Similarity M atrix F actorization (IULMF), which combines these two methods to study their contributions on the final performance. Experimental evaluations on the real datasets show that our methods outperform the baseline approaches in terms of both the precision and NDCG.

The consistency of a response to a given post at semantic-level and emotional-level is essential for a dialogue system to deliver human-like interactions. However, this challenge is not well addressed in the literature, since most of the approaches neglect the emotional information conveyed by a post while generating responses. This article addresses this problem by proposing a unified end-to-end neural architecture, which is capable of simultaneously encoding the semantics and the emotions in a post for generating more intelligent responses with appropriately expressed emotions. Extensive experiments on real-world data demonstrate that the proposed method outperforms the state-of-the-art methods in terms of both content coherence and emotion appropriateness.

Logistic regression (LR) is the most commonly used loss function in multi-label image classification. However, it suffers from class imbalance problem caused by the huge difference in quantity between positive and negative samples as well as between different classes. First, we find that feeding randomly generated noise samples into an LR classifier is an effective way to detect class imbalances, and further define an informative imbalance metric named inference tendency based on noise sample analysis. Second, we design an efficient moving average based method for calculating inference tendency, which can be easily done during training with negligible overhead. Third, two novel rectification methods called extremum shift (ES) and tendency constraint (TC) are designed to offset or constrain inference tendency in the loss function, and mitigate class imbalances significantly. Finally, comparative experiments with Resnet on Microsoft COCO, NUS-WIDE and DeepFashion demonstrate the effectiveness of inference tendency and the superiority of our approach over the baseline LR and several state-of-the-art alternatives.

In the big data era, the generation of data presents some new characteristics, including wide distribution, high velocity, high dimensionality, and privacy concern. To address these challenges for big data analytics, we develop a privacy-preserving distributed online learning framework on the data collected from distributed data sources. Specifically, each node (i.e., data source) has the capacity of learning a model from its local dataset, and exchanges intermediate parameters with a random part of their own neighboring (logically connected) nodes. Hence, the topology of the communications in our distributed computing framework is unfixed in practice. As online learning always performs on the sensitive data, we introduce the notion of differential privacy (DP) into our distributed online learning algorithm (DOLA) to protect the data privacy during the learning, which prevents an adversary from inferring any significant sensitive information. Our model is of general value for big data analytics in the distributed setting, because it can provide rigorous and scalable privacy proof and have much less computational complexity when compared to classic schemes, e.g., secure multiparty computation (SMC). To tackle high-dimensional incoming data entries, we study a sparse version of the DOLA with novel DP techniques to save the computing resources and improve the utility. Furthermore, we present two modified private DOLAs to meet the need of practical applications. One is to convert the DOLA to distributed stochastic optimization in an offline setting, the other is to use the mini-batches approach to reduce the amount of the perturbation noise and improve the utility. We conduct experiments on real datasets in a configured distributed platform. Numerical experiment results validate the feasibility of our private DOLAs.

Nowadays, machine learning is becoming a new paradigm for mining hidden knowledge in big data. The collection and manipulation of big data not only create considerable values, but also raise serious privacy concerns. To protect the huge amount of potentially sensitive data, a straightforward approach is to encrypt data with specialized cryptographic tools. However, it is challenging to utilize or operate on encrypted data, especially to perform machine learning algorithms. In this paper, we investigate the problem of training high quality word vectors over large-scale encrypted data (from distributed data owners) with the privacy-preserving collaborative neural network learning algorithms. We leverage and also design a suite of arithmetic primitives (e.g., multiplication, fixed-point representation, sigmoid function computation, etc.) on encrypted data, served as components of our construction. We theoretically analyze the security and efficiency of our proposed construction, and conduct extensive experiments on representative real-world datasets to verify its practicality and effectiveness.