Federated Class-Incremental Learning (FCIL) enables distributed models to continuously learn new categories while preserving privacy, which suffers from the problem of catastrophic forgetting. To address this issue, generative replay has emerged as a mainstream solution, yet its performance is hampered by two fundamental bottlenecks: (1) low-fidelity synthesis, where generated visual samples fail to effectively represent historical knowledge, and (2) class imbalance in FCIL, which undermines fair learning across classes. In this paper, we propose a novel generative replay framework called FedART (Federated Adaptive Replay with Text-anchors). To combat low-fidelity synthesis, FedART employs a text-anchored initialization strategy. Instead of optimizing from a random start, this approach provides strong semantic priors to guide the generation process. To tackle class imbalance, we design a dual adaptive aggregation mechanism. This mechanism applies tailored weighting strategies at both the generator and classifier levels, leveraging local training dynamics to ensure both the quality of generative knowledge and the fairness of classifier aggregation. Extensive experiments on CIFAR-100 and Tiny-ImageNet demonstrate that FedART significantly outperforms state-of-the-art methods, achieving an accuracy of up to 43.62% and establishing a new and robust benchmark for enhancing the effectiveness of generative replay in FCIL.

Neural Architecture Search (NAS) has been extensively studied due to its ability in automatic architecture engineering. Existing NAS methods rely heavily on the gradients and data labels, which either incur immense computational costs or suffer from discretization discrepancy due to the supernet structure. Moreover, the majority of them are limited in generating diverse architectures. To alleviate these issues, in this paper, we propose a novel zero-cost proxy called $\mathsf {MeCo}$MeCo based on the Pearson correlation matrix of the feature maps. Unlike the previous work, the computation of $\mathsf {MeCo}$MeCo as well as its variant $\mathsf {MeCo_{opt}}$MeCoopt requires only one random data for a single forward pass. Based on the proposed zero-cost proxy, we further craft a new zero-shot NAS scheme called $\mathsf {FLASH}$FLASH, which harnesses a new proxy-based operation scoring function and a greedy heuristic. Compared to the existing methods, $\mathsf {FLASH}$FLASH is highly efficient and can construct diverse model architectures instead of repeated cells. We design comprehensive experiments and extensively evaluate our designs on multiple benchmarks and datasets. The experimental results show that our method is one to six orders of magnitudes more efficient than the state-of-the-art baselines with the highest model accuracy.

Machine Learning as a Service (MLaaS) offers powerful data analytics services to clients with limited resources. However, it still raises concerns about the integrity of delegated computation and the privacy of the server's model parameters. To address these issues, zero-knowledge Machine Learning (zkML) has been suggested for computation verifiability with privacy guarantee for ML models. Nevertheless, the existing zkML schemes focus on only one classical ML classification algorithm or deep neural networks, which may not achieve satisfactory accuracy or require large-scale training data and model parameters, thus limiting their usefulness in certain applications. In this article, we propose ezDPS, an efficient and zero-knowledge scheme for classical ML inference that processes data in multiple stages for improved accuracy. Unlike prior works, each stage of the ezDPS pipeline is based on a well-established classical ML algorithm, including Discrete Wavelet Transformation, Zero-Score Normalization, Principal Components Analysis, and Support Vector Machine. We design new gadgets to prove various ML operations effectively. Our implementation of ezDPS has been fully tested on real datasets, and experimental results show that it is up to three orders of magnitude more efficient than generic circuit-based approaches, while also maintaining greater accuracy than single ML classification approaches.

Multimodal deep learning technologies have advanced significantly, which brings extensive applications in diverse fields. The substantial computational demands of training and prediction in multimodal deep learning have made the End-Edge-Cloud (EEC) framework popular. It is essential to protect multimodal data and model privacy in such a framework. However, traditional cryptographic methods, though secure for data and models at edge nodes, cause efficiency limitations. In this paper, we propose SecMdp, an SGX-assisted secure computational framework for multimodal data in the EEC architecture. Edge nodes are equipped with the trusted execution environment (e.g., Intel SGX) to run multimodal algorithms. Additionally, to address the side-channel attacks of SGX, we present an enhanced PathORAM algorithm, MM_PathORAM, for the multimodal training and prediction processes, which are tailored for multimodal deep learning scenarios. It accelerates multimodal data access while protecting data privacy and model security. Experimental evaluation supports the effectiveness of our design in preserving edge computing efficiency. It demonstrates negligible impact on the speed of multimodal data loading, the configuration of model parameters during training, or the accuracy of predictions.

Federated learning has emerged as a promising paradigm for large-scale collaborative training tasks, harnessing diverse local datasets from different clients to jointly train global models. In real-world implementations, client data could have label noise, causing the quality of the global model to be influenced. Existing label-correction solutions assume all the clients are discreet and fail to consider detecting the malicious clients, thus are not practical or privacy-preserving. In this paper, we present zkCor, an efficient and reliable label noise correction scheme with zero-knowledge confidentiality. Our method is designed upon FedCorr [1], but with more relaxed security assumptions. zkCor is established from the ingenious synergy of the label noise correction protocol and the zero-knowledge proof (ZKP), requiring each client to provide a computation integrity proof to the aggregator in each iteration. Thus, clients are forced to jointly guarantee label-correction reliability. We further devise a batch ZKP that is efficient and more suitable for federated learning settings. We rigorously illustrate the building blocks of zkCor and complete the prototype implementation. The extensive experiments demonstrate that zkCor can gain at least 2 to 30 times better performance than the baseline approach on verification workloads with nearly no extra proof time cost from clients.

IOTA has emerged as a promising blockchain platform specially designed for the Internet of Things (IoT). Its distributed ledger, called tangle, adopts a directed acyclic graph (DAG) structure to achieve fast transaction confirmation and high scalability. While the tangle tremendously mitigates blockchain performance concerns relative to a traditional single chain, it simultaneously increases the potential risk of double-spending attacks. Utilizing constructing illegal tangle branches to substitute for legitimate ones, attackers inside IOTA can launch double-spending attacks and seriously compromise the tangle security. In this work, we take the first step toward investigating the problem of tangle vulnerability by leveraging the large deviation theory. The proposed scheme, called SecTangle, can assist IOTA in effectively reducing the tangle vulnerability to resist double-spending attacks. The core idea is to explore the security threshold defined and deduced to affect the robustness of the tangle by evaluating the probability of tangle vulnerability. By adjusting the critical factors of the security threshold, fake tangle branches can be found by IOTA efficiently to prevent double-spending attacks. Besides, we further devise a transaction recovery algorithm to recover time-sensitive legitimate transaction branches. This paper validates that the proposed scheme is efficient with comprehensive theoretical analysis and simulation experiments.

Blockchain-based data storage has become an emerging paradigm, providing a fair and transparent data platform for decentralized applications. However, how to achieve secure on-chain verification for arbitrary SQL queries in such a decentralized storage remains under-explored. Due to the limitations of authenticated data structure (ADS), existing works either do not consider arbitrary query verification issue or fail to achieve practical gas consumption efficiency. In this paper, we present a novel arbitrary query verification scheme for decentralized storage. The proposed scheme, named$\mathsf {zkQuery}$, enables efficient public verification for arbitrary queries with zero-knowledge confidentiality.$\mathsf {zkQuery}$is built from the ingenious synergy of techniques from both zero-knowledge proof and smart contract technology. The core idea is to delegate smart contracts to fairly execute results verification and utilize our tailored zero-knowledge proof protocol to facilitate arbitrary computation in a privacy-preserving manner. The verification protocols of$\mathsf {zkQuery}$are highly customized for decentralized storage, where the complexity of on-chain verification can be completed in logarithmic time, significantly decreasing gas consumption. We rigorously provide security analysis and complete the prototype implementation. The extensive experiments over the NEAR blockchain show that$\mathsf {zkQuery}$can gain at least$2\times $better performance than the baseline approach on all metrics.

Pre-trained encoders in computer vision have recently received great attention from both research and industry communities. Among others, a promising paradigm is to utilize self-supervised learning (SSL) to train image encoders with massive unlabeled samples, thereby endowing encoders with the capability to embed abundant knowledge into the feature representations. Backdoor attacks on SSL disrupt the encoder's feature extraction capabilities, causing downstream classifiers to inherit backdoor behavior and leading to misclassification. Existing backdoor defense methods primarily focus on supervised learning scenarios and cannot be effectively migrated to SSL pre-trained encoders. In this article, we present a backdoor defense scheme based on self-supervised knowledge distillation. Our approach aims to eliminate backdoors while preserving the feature extraction capability using the downstream dataset. We incorporate the benefits of contrastive and non-contrastive SSL methods for knowledge distillation, ensuring differentiation between the representations of various classes and the consistency of representations within the same class. Consequently, the extraction capability of pre-trained encoders is preserved. Extensive experiments against multiple attacks demonstrate that the proposed scheme outperforms the state-of-the-art solutions.

Federated learning has recently emerged as a popular learning paradigm that enables multiple clients to jointly train a high-quality model without sharing their local training datasets. Each client will train its local model by its local dataset, and the global model will be aggregated by local models. However, training datasets from all clients are usually heterogeneous, because they are chosen by the clients themselves. This leads to over-fitting in the local models, thus affecting the performance of the global model. There are existing methods such as regularization in local optimization and improving model aggregation. However, they introduce additional computing or storage overhead.In this paper, we present a novel system design FedIR to eliminate the impact of data heterogeneity. FedIR introduces the idea of learning invariant features in domain adaptation so that the aggregated global model can handle the data heterogeneity well. We refer to the Optimal Path Search to assist model training in obtaining better invariant representations. The modifications to local model structures are very small, with little impact on local training and server aggregation. Extensive experiments demonstrate that FedIR achieves state-of-the-art performance on popular federated learning benchmarks including CIFAR-10 and CIFAR-100, with less computation cost and communication rounds.

Neural Architecture Search (NAS) is a promising paradigm in automatic architecture engineering. Zero-shot NAS can evaluate the network without training via some specific metrics called zero-cost proxies. Though effective, the existing zero-cost proxies either invoke at least one backpropagation or depend highly on the data and labels. To alleviate the above issues, in this paper, we first reveal how the Pearson correlation matrix of the feature maps impacts the convergence rate and the generalization capacity of an over-parameterized neural network. Enlightened by the theoretical analysis, we propose a novel zero-cost proxy called $\mathsf{MeCo}$, which requires only one random data for a single forward pass. We further propose an optimization approach $\mathsf{MeCo_{opt}}$ to improve the performance of our method. We design comprehensive experiments and extensively evaluate $\mathsf{MeCo}$ on multiple popular benchmarks. $\mathsf{MeCo}$ achieves the highest correlation with the ground truth (e.g., 0.89 on NATS-Bench-TSS with CIFAR-10) among all the state-of-the-art proxies, which is also fully independent of the data and labels. Moreover, we integrate $\mathsf{MeCo}$ with the existing generation method to comprise a complete NAS. The experimental results illustrate that $\mathsf{MeCo}$-based NAS can select the architecture with the highest accuracy and a low search cost. For instance, the best network searched by $\mathsf{MeCo}$-based NAS achieves 97.31% on CIFAR-10, which is 0.04% higher than the baselines under the same settings. Our code is available at https://github.com/HamsterMimi/MeCo

IOTA is an emerging blockchain platform specially designed for the Internet of Things (IoT). It adopts a distributed ledger based on a directed acyclic graph (DAG) structure called the Tangle, achieving high concurrency, incremental scalability, and zero transaction fees. While the DAG-based ledger structure can promise many benefits for IOTA, it also exposes a wide attacking surface for the double-spending attack. By creating fake branches to replace the legitimate ones in the Tangle, the attackers can achieve the purpose of double-spending. Therefore, there is a great need for securing IOTA with preventive strategies. In this work, we present the design of an effective preventive strategy for IOTA that can reduce the Tangle vulnerability against the double-spending attack. Our main idea is to in-vestigate the Tangle vulnerability problem by using the large deviation theory. We provide the Tangle vulnerability probability and the decay speed of the probability. By evaluating the decay speed, we find the key factor related to the robustness level against double-spending attacks. Besides, we define and deduce the security threshold based on the evaluation results that can affect the probability decay speed of the Tangle vulnerability. With the security threshold, IOTA can efficiently inspect illegal Tangle branches to resist double-spending attacks. Finally, the comprehensive theoretical analysis and simulation experiments demonstrate that the proposed strategy is efficient and practical.

IOTA has emerged as a promising and feeless decentralized computing paradigm for developing blockchain-based Internet of Things (IoT) applications with high-performance transaction rates and incremental scalability. To support micro-payments of IoT devices, IOTA has abandoned the original blockchain reward mechanism while IOTA nodes voluntarily contribute resources to maintain network stability. However, removing the mining rewards results in the resource cost of generating IOTA ledgers (known as the Tangle) being borne only by IOTA nodes. With the continuous expansion of the IOTA network, cost consumption is increasing. Thus, the inability to effectively reduce the cost of Tangle generation would lead to people being reluctant to dedicate resources to IOTA for maintaining the network robustness. In this paper, for the first time, we present a full-fledged transaction cost optimization scheme for IOTA, called Tangless, which can assist IOTA nodes in effectively reducing the Tangle generation cost while maintaining the strong robustness of the IOTA network. By using our proposed scheme, each IOTA node can effectively formulate the threshold of transaction approval rate in real time, maintaining the stability of the IOTA network with the optimal computational cost. We harness Lyapunov optimization theory to design a computational optimization algorithm for minimizing the total cost of nodes in IOTA. Then, we resort to large deviations theory to devise an optimized transaction rate control algorithm to further eliminate orphan Tangles that waste computational costs. Comprehensive theoretical analysis and simulation experiments confirm the effectiveness and practicability of our proposed scheme.

Spectrum auction is one of the most effective ways to achieve dynamic spectrum allocation in cognitive radio networks , and it provides one effective way to manage the spectrum demands of IoT devices with limited resources. Most spectrum auctions focus on protecting bidder privacy and achieving excellent social efficiency, but few tackles the verification of auction results that are controlled by the auctioneer. In this paper, we propose a cloud-based framework for verifiable privacy-preserving spectrum auctions. Our framework adopts a modified AFGH re-encryption algorithm that achieves both bid privacy protection and auction results verification at the same time. The cloud server helps to compute auction results based on homomorphic encryption , and an auctioneer decrypts the encrypted data from the server to obtain auction results. Meanwhile, the property of re-encryption makes it possible for any bidder to verify the auction results without compromising other bidders’ privacy.

Summary Mobile ad hoc Network (MANET) is a cluster of moveable devices connected through a wireless medium to design network with rapidly changing topologies due to mobility. MANETs are applicable in variety of innovative application scenarios where smart devices exchange data among each other. In this case, security of data is the major concern to provide dependable solution to users. This article presents a secure mechanism for data transfer where sender splits the data into fragments and receiver gets the actual data by assimilating the data fragments. We have presented an Enhanced Secured Lempel‐Ziv‐Welch (ES‐LZW) algorithm that provides cryptographic operations for secure data transfer. In proposed model, we have utilized the disjoint paths to transfer the data fragments from sender side and assimilate these fragments at receiver to get the original data. The messages containing data fragments are compressed and encrypted as well. Our scheme ensures confidentiality, integrity, efficient memory utilization, and resilience against node compromising attacks. We have validated our work through extensive simulations in NS‐2.35 using TCL and C language. Results prove that our scheme reduces memory consumption along with less encryption and decryption cost as compared to blowfish especially when plaintext has more repetitive data. We have also analyzed the impact of creating data fragments, fraction of communication compromised, and probability to compromise the data fragments by subverting intermediaries.

Realizing the full potential of the Internet of Things (IoT) requires solving technical and business challenges including the identification of things, their organization, and integration. The subsequent management of large data volumes that are generated from such systems, and the effective use of knowledge-based decision systems that can make use of IoT resources remains a challenge at present. Various representation formats already exist for specifying sensors and devices that are part of the IoT ecosystem. However, many of these are either specific to use within a particular application area (e.g., environmental monitoring), or specific to a middleware platform. Overcoming device, firmware, and data format heterogeneity remains a significant challenge in real world IoT systems. Consequently, dealing with data that are generated from such systems and reasoning with these data is constrained due to these limitations. IoT-based platforms also offer a variety of different communication protocols (for both long range [at low data rates] and short range [at high data rates]), such as SigFox, LoRaWAN, NB-IoT, Wifi Direct, and so on. These protocols generally offer different decision points around energy used, distance covered, and data rates observed. Another aspect of heterogeneity in IoT systems therefore relates to dealing and switching between these protocols based on context of use and application requirements. We received 13 papers aligned with the theme of this special issue. In particular, the benefit of using deep learning to solve “traditional” problems is being recognized by a number of researchers. All papers were initially screened by the editors to ensure alignment with the theme of this special issue, and high-quality papers were sent to reviewers. In total, seven papers were selected for inclusion in the special issue. The submitted papers combine the use of novel methods and demonstrate effective use of experimental techniques. The papers included in this special issue are: A physiological data-driven model for learners' cognitive load detection using HRV-PRV feature fusion and optimized XGBoost classification

Crowdsourcing is a promising technology to accomplish a complex task via eliciting services from a large group of contributors. Recent observations indicate that the success of crowdsourcing has been threatened by the malicious behaviors of the contributors. In this paper, we analyze the attack problem using an iterated prisoner's dilemma (IPD) game and propose a reward-penalty expected payoff algorithm based on zero-determinant (ZD) strategies to reward a worker's cooperation or penalize its defection in order to entice the final cooperation. Both theoretical analysis and simulation studies are performed, and the results indicate that the proposed algorithm has the following two attractive characteristics: 1) the requestor can incentivize the worker to become cooperative without any long-term extra cost; and 2) the proposed algorithm is fair so that the requestor cannot arbitrarily penalize an innocent worker to increase its payoff even though it can dominate the game. To the best of our knowledge, we are the first to adopt the ZD strategies to stimulate both players to cooperate in an IPD game. Moreover, our proposed algorithm is not restricted to solve only the problem of crowdsourcing dilemma - it can be employed to tackle any problem that can be formulated into an IPD game.

Quality control in crowdsourcing is challenging due to the heterogeneous nature of the workers. The state-of-the-art solutions attempt to address the issue from the technical perspective, which may be costly because they function as an additional procedure in crowdsourcing. In this paper, an economics based idea is adopted to embed quality control into the crowdsourcing process, where the requestor can take advantage of the market power to stimulate the workers for submitting high-quality jobs. Specifically, we employ two sequential games to model the interactions between the requestor and the workers, with one considering binary strategies while the other taking continuous strategies. Accordingly, two incentive algorithms for improving the job quality are proposed to tackle the sequential crowdsourcing dilemma problem. Both algorithms are based on a sequential zero-determinant (ZD) strategy modified from the classical ZD strategy. Such a revision not only provides a theoretical basis for designing our incentive algorithms, but also enlarges the application space of the classical ZD strategy itself. Our incentive algorithms have the following desired features: 1) they do not depend on any specific crowdsourcing scenario; 2) they leverage economics theory to train the workers to behave nicely for better job quality instead of filtering out the unprofessional workers; 3) no extra costs are incurred in a long run of crowdsourcing; and 4) fairness is realized as even the requestor (the ZD player), who dominates the game, cannot increase her utility by arbitrarily penalizing any innocent worker.

In recent years, smart sports equipment and body sensor systems have become popular in professional and amateur sports. One of a few remaining problems in real-time applications is the discovery of knowledge from the embedded sensors data. In sports training, such knowledge helps accelerated motor learning. The authors start with exploring the possibilities of the classification of golf swing performance with the 1-D convolutional neural network (CNN) in real-time. They thoroughly investigate multiple golf swing data classifiers based on CNNs fed with multi-sensor signals. The authors test the possibilities of real-time performance of CNN methods on the multi-length sequences. In addition, they thoroughly evaluate the performance of their well-trained CNN-based classifier on the aforementioned test set in terms of common indicators. Experiments and corresponding results show that the authors' models can satisfy the real-time requirement of the accuracy of the classification and outperform support vector machine (SVM).

Due to the complexity and volume, outsourcing ciphertexts to a cloud is deemed to be one of the most effective approaches for big data storage and access. Nevertheless, verifying the access legitimacy of a user and securely updating a ciphertext in the cloud based on a new access policy designated by the data owner are two critical challenges to make cloud-based big data storage practical and effective. Traditional approaches either completely ignore the issue of access policy update or delegate the update to a third party authority; but in practice, access policy update is important for enhancing security and dealing with the dynamism caused by user join and leave activities. In this paper, we propose a secure and verifiable access control scheme based on the NTRU cryptosystem for big data storage in clouds. We first propose a new NTRU decryption algorithm to overcome the decryption failures of the original NTRU, and then detail our scheme and analyze its correctness, security strengths, and computational efficiency. Our scheme allows the cloud server to efficiently update the ciphertext when a new access policy is specified by the data owner, who is also able to validate the update to counter against cheating behaviors of the cloud. It also enables (i) the data owner and eligible users to effectively verify the legitimacy of a user for accessing the data, and (ii) a user to validate the information provided by other users for correct plaintext recovery. Rigorous analysis indicates that our scheme can prevent eligible users from cheating and resist various attacks such as the collusion attack.

Crowdsourcing is a promising paradigm to accomplish a complex task via eliciting services from a large group of contributors. However, recent observations indicate that the success of crowdsourcing is being threatened by the malicious behaviors of the contributors. In this paper, we analyze the malicious attack problem using an iterated prisoner's dilemma (IPD) game and propose a zero-determinant (ZD) strategy based scheme by rewarding a worker's cooperation or penalizing the defection for enticing his final cooperation. Both theoretical analysis and simulation study indicate that the proposed algorithm has two attractive characteristics: 1) the requestor can incentivize the worker to keep on cooperating by only increasing the short-term payment; and 2) the proposed algorithm is fair, so the requestor cannot arbitrarily penalize an innocent worker to increase her payoff even though she can dominate the game. To the best of our knowledge, we are the first to use the ZD strategy to stimulate both players to cooperate in an IPD game. Moreover, our proposed algorithm is not restricted to solve the problem of the malicious crowdsourcing - it can be employed to tackle any problem that can be formulated by an IPD game.

Air pollution is becoming a serious issue that threatens everyone's daily life. Accordingly how to measure the local air quality easily and quickly becomes an urgent problem. Sensor-based methods are relatively expensive, and image based air quality measurement is a promising direction as it is often less cost. Meanwhile web service of precise air quality measurement is of great importance as it allows to timely monitor the air pollution and can provide recommendations for decision makers. This paper devises an effective web service to address this challenging problem. Specifically we offer a service letting mobile device users to upload photos taken outdoor with meta information. Once the background web service computation is finished, we return the air quality level at their location to the users. In our service system, it includes three basic modules: 1) using GPS location information to get the basic air quality value from the nearest official air quality station, 2) using photo by our air quality assessment based on dictionary learning for image representation to compute air quality. In this method we add ℓ21norm to the target function to promote the nonzero coefficients of the words aggregate within similar quality level, 3) using photo uploaded by user by our CNN based photo air pollution estimation method to obtain the air quality estimation. Finally we combine the above three estimations to reach the final estimation to the end users. Empirical experiments are conducted on the real-world dataset that collaborates the efficacy of our method.

Air quality has become a major global concern for human beings involving all social stratums, for both developing and developed countries. Web service of precise and early air pollution forecasting is of great importance as it allows people to pro-actively take preventative and protective measurements. As an endeavor on the course of machine learning based air quality forecasting, this paper presents an initiative and its technological details in solving this challenging problem. Specifically, this work involves three major highlights regarding with both algorithmic innovation and deployment with its impact: 1) We propose a multi-channel ensemble learning framework, 2) We propose a new supervised feature learning and extraction method, i.e. sufficient statistics feature mapping based on Deep Boltzman Machine, which serves as a building block for our learning system, 3) We target our air pollution prediction method to the city of Beijing, China as it is at the forefront for battling against air pollution, which is embodied as a web service for prediction. Extensive experiments of real time air pollution forecasting on the real-world data demonstrates the effectiveness of the proposed method and value of the deployed web service system.

Crowdsourcing applications are vulnerable to malicious behaviors, posing serious threats to their adoption and large deployment. Based on the notion that the requestor (i.e., the crowdsourcer) can block malicious behaviors via leveraging the market power through task allocation and pricing, we propose two novel frameworks based on the mechanism design game theory (i.e., the reverse game theory). To the best of our knowledge, we are the first to exploit the market power and to apply the mechanism design game theory in thwarting malicious behaviors in crowdsourcing. The first proposed framework is built on a requestor-dominant mechanism design game (Rd-MDG), where the game rule is determined solely by the requestor. The second proposed framework is based on the worker-assisted mechanism design game (WaMDG), where the worker (i.e., the contributor) can assist the requestor to determine the game rules by offering advices. These two frameworks have the following salient features: i) neither of them requires the workers to reveal their private information; ii) the game rules of each framework are designed to be able to force the workers to calculate their best strategies based on their actual private information; iii) our theoretical analysis shows that equilibriums exist for both frameworks; and iv) our extensive simulation results demonstrate that these two frameworks can thwart malicious behaviors by driving the workers with a higher attack intent into obtaining lower utilities.

Facial expression recognition, which many researchers have put much effort in, is an important portion of affective computing and artificial intelligence. However, human facial expressions change so subtly that recognition accuracy of most traditional approaches largely depend on feature extraction. Meanwhile, deep learning is a hot research topic in the field of machine learning recently, which intends to simulate the organizational structure of human brain's nerve and combine low-level features to form a more abstract level. In this paper, we employ a deep convolutional neural network (CNN) to devise a facial expression recognition system, which is capable to discover deeper feature representation of facial expression to achieve automatic recognition. The proposed system is composed of the Input Module, the Pre-processing Module, the Recognition Module and the Output Module. We introduce both the Japanese Female Facial Expression Database(JAFFE) and the Extended Cohn-Kanade Dataset(CK+) to simulate and evaluate the recognition performance under the influence of different factors (network structure, learning rate and pre-processing). We also introduce a K-nearest neighbor (KNN) algorithm compared with CNN to make the results more convincing. The accuracy performance of the proposed system reaches 76.7442% and 80.303% in the JAFFE and CK+, respectively, which demonstrates feasibility and effectiveness of our system.

Semantic extraction based on key words is essential method to find relationships of entities. Enterprise relations analysis is becoming more and more popular in financial field, because it gives explanation on business processes and decision-making reference on investments. Most of previous studies focused on the analysis of specific situations and built relations on targeted enterprises. Hence, there is no method based on key words through decomposing sentence to figure out semantics. In this paper, we adopt EOLN model to build enterprise map (EM), taking explicit knowledge and implied relations extraction into consideration. Explicit knowledge means the knowledge can be directly found without reasoning, and implied relations need to be generated by reasoning and computation. We perform comprehensive experiments on data set collected from 2515 different companies, and build the complete EM using the key words based relationship extraction. Experimental results validate the robustness of our proposed approach.

The use of mobile phones while driving has become a major source of distraction to drivers, leading to a large number of car accidents. In this paper, we study the problem of automatically detecting driver phone calls by monitoring smartphone activities and utilizing the vehicle on-board unit. The challenges to overcome include: i) passenger phone calls should be allowed while the calls of the driver should be blocked; ii) the detection mechanism should be phone position-independent and phone owner-independent as the driver may put the smartphone at any position in the front row and make calls via an earphone, or the driver may borrow a passenger's phone to make a call; iii) the in-vehicle environment is noisy resulted from the operating engine, the music the driver and passenger may listen to, and the conversation between passengers and/or the driver; and iv) the computational cost at the smartphone should be light as realtime phone call detection is expected to effectively block an ongoing call to and from the driver. To overcome these challenges and achieve our objective of detecting driver phone calls, we take advantage of the uniqueness of individual's voice features. Through a short period of learning stage, our proposed system can recognize the driver's voice from the collected audio data. Combined with the smartphone's call state, our scheme can determine whether the driver is participating in the current phone call or not. Our strategy takes into account the complicated in-vehicle environment, and the proposed algorithm does not rely on the location of the phone within the vehicle nor the ownership of the smartphone, as the most existing driver phone call detection mechanisms do. We develop a client-server based system with the smartphones being the clients and the vehicle on-board unit being the server. To validate our mechanism, we perform extensive real-world experiments under different scenarios. The results demonstrate a high probability of detecting driver phone calls with a small false alarm rate.

Air pollution has raised people's intensive concerns especially in developing countries such as China and India. Different from using expensive or unreliable methods like sensor-based or social network based one, photo based air pollution estimation is a promising direction, while little work has been done up to now. Focusing on this immediate problem, this paper devises an effective convolutional neural network to estimate air's quality based on photos. Our method is comprised of two ingredients: first a negative log-log ordinal classifier is devised in the last layer of the network, which can improve the ordinal discriminative ability of the model. Second, as a variant of the Rectified Linear Units (ReLU), a modified activation function is developed for photo based air pollution estimation. This function has been shown it can alleviate the vanishing gradient issue effectively. We collect a set of outdoor photos and associate the pollution levels from official agency as the ground truth. Empirical experiments are conducted on this real-world dataset which shows the capability of our method.

As a promising technology, crowdsourcing aims to accomplish a complex task via eliciting services from a large group of workers. However, recent observations indicate that the success of crowdsourcing is being hindered by the malicious behaviors of the workers. In this paper, we analyze the attack problem using an iterated prisoner's dilemma (IPD) game and propose an zero-determinant (ZD) strategy based algorithm. Simulation results demonstrate that the requestor can incentivize the worker to keep on cooperating.

Due to the popularity of auction mechanisms in real-world applications and the increasing awareness of securing private information, auctions are in dire need of bid-privacy protection. In this paper, we design three secure, multi-unit, sealed-bid, first-price auction schemes. The first is a secure auction using homomorphic encryption and is denoted by SAHE; the second is a secure action using masking values and is denoted by SAMV; and the third has an improved masked noise algorithm, denoted by ISAMV. In the first, SAHE, the auction is processed on encrypted bids by a server, and the final output is only known by the auctioneer. Neither the auctioneer nor the server can obtain the full information of the bidders. The second and third auctions, SAMV and ISAMV, decrease computational complexity. Instead of homomorphic encryption, they use random noise to mask the bid values. By using a masking method, the server only knows the noise, and the auctioneer only knows the auction results; neither will see the private information of the bidders. All three schemes enable the auctioneer to verify that the winners have paid the correct amounts. A thorough theoretical analysis is performed to evaluate the security properties, computational complexity, and communication complexity of the auctions. Copyright © 2016 John Wiley & Sons, Ltd.

Cognitive radio provides an efficient technology to solve the problem of spectrum resource scarcity while cooperative communications can increase channel capacity. It is a common sense that combining the two benefits the system performance of cognitive radio networks (CRNs). A challenging problem of intelligent cooperation in CRNs is how to make control decisions when a secondary user cooperates with a primary user to get an optimal cooperation outcome. In this paper, we consider a scenario where a secondary user provides effort to the primary user to win the competition from many secondary users and simultaneously achieves optimal throughput. We establish a novel cooperation scheme termed Low Price to Win (LPW), and abstract the cooperation problem in CRNs as an optimization problem with multiple constraints. Unlike some traditional methods that provide direct solutions, we design a novel greedy algorithm using the Lyapunov optimization technique, by which the sophisticated optimization problem can be divided into a few subproblems and then cross-layer optimization can be applied. Our simulation results demonstrate the efficiency and effectiveness of the proposed algorithm.

As a distributed system, a wireless network, usually faces a complex environment (transient faults and topology changes occur frequently). The connected dominating set (CDS) problem has been widely studied due to its important applications in wireless communication and networks, especially the important role as a virtual backbone for efficient routing. In this paper, under SINR (Signal-to-Interference-plus-Noise-Ratio) model, we propose a distributed self-stabilizing maximal independent set (MIS) algorithm (DSSMIS). Based on DSSMIS, we design a distributed self-stabilizing algorithm (DSSCDS) for CDS construction with constant approximation within O(log n) rounds. To best of our knowledge, this is the first self-stabilizing CDS algorithm under SINR model.

Mobile opportunistic networks make use of a new networking paradigm that takes advantage of node mobility to distribute information. Studying their inherent properties of information dissemination can provide a straightforward explanation on the potentials of mobile opportunistic networks to support emerging applications such as mobile commerce, emergency services, and so on. In this paper, we investigate the inherent properties of information dissemination using the Lévy mobility model to characterize the movement pattern of the nodes. Because Lévy mobility can closely mimic human walk, the analysis model we adopt is practical. Our analyses are taken from the perspectives of small- and large-scales. From the perspective of small-scale, the distribution of the minimum time needed by the information to spread to a given region is investigated, from the perspective of large-scale, the bounds of the probability of the earliest time at which the information arrives in a region that is sufficiently farther away are obtained. We also provide the rate that such probability approaches zero as the distance to the region increases to infinity. Finally, our main results are validated by the numerical simulations.

In this paper, we propose an extensible and flexible truthful auction framework that is individual-rational and self-collusion resistant. By properly setting one simple parameter, this framework can yield efficient auctions (like VCG) and (sub)optimal auctions (like Myerson's Optimal Mechanism (MOM)) with a more computationally-efficient procedure compared to VCG and MOM; by carefully choosing virtual valuation functions for the bidders, it can produce attribute-aware auctions that take the channel diversity into consideration. The framework adopts a novel procedure that can prevent bidder self-collusion resulted from the bid diversity. Theoretical analysis and case studies demonstrate the strength of our auction framework in handling various considerations in a practical heterogeneous spectrum market.

In this paper we investigate the dynamic traffic relationship characterized by a similarity value from one road point to another in vehicle networks. Due to the regularity of human mobility, traffic exhibits strong correlations in both temporal domain and spatial domain. By exploiting the similarity values, we derive application-specific message update rules for affinity propagation, based on which we propose an instant traffic clustering algorithm to partition the road points into time variant clusters, where the traffics within the same cluster are strongly spatially correlated. Online traffic clustering is also considered by clustering combination via evidence accumulation for further influence study. We also present a neural network based traffic prediction algorithm to predict the traffic conditions cluster by cluster for a future time based on the current and historical traffic data. Simulation study on real traffic data demonstrates that our proposed algorithms are able to identify the true influences among road points and provide accurate traffic predictions.

Target localization and tracking in the Internet of Things (IoT) environment have been paid more and more attention recently. The knowledge and information generated from wireless sensor nodes of the IoT make huge contributions to localization and tracking of targets with high mobility. This paper presents a square-root unscented Kalman filtering (SR-UKF) based localization and tracking algorithm for mobile target in an IoT environment. First, a localization initialization model is proposed for an IoT scenario. Then, according to information of neighboring sensor nodes, we employ the SR-UKF idea for the further localization and tracking of the target. Simulation results demonstrate that the proposed algorithm achieves lower localization and tracking error under the same computational complexity, compared with some conventional extended Kalman filtering (EKF) or UKF based methods. The proposed algorithm is of great significance in the field of IoT information processing.

The vision of Internet of Things is to connect everyday objects through embedding wireless devices, so that they can interact with each other and provide new services. One of major challenges for the IoT is the heterogeneous problem. Information generated by different IoT objects will not be compatible, which hinders data communications between the IoT objects. In this paper, we study semantic technology for integrating heterogeneous information in the IoT. We propose to use ontologies to model the schema of data generated by the IoT objects, and propose a hybrid ontology matching approach to solve the heterogeneous problem. The proposed approach uses multiple matchers to compute similarities between ontology elements, and computes weight for each matcher by making use of the hierarchical structure of ontology. Experimental results show that our method can effectively filter out wrong mappings and obtain alignments with high quality between ontologies.

Mining data streams for knowledge discovery is important to many applications, including Web click stream mining, network intrusion detection, and on-line transaction analysis. In this paper, by analyzing data characteristics, we propose an efficient algorithm SWSS (Sequential pattern mining with the weighted sliding window model in SPAM) to mine frequent sequential patterns based on the weighted sliding windows model. This algorithm provides more space for users to specify which sequences they are more interested in. Extensive experiments show that the proposed algorithm is feasible and efficient for mining all sequential patterns as users specified.

DNA Microarray technique has produced large amount of gene expression data. To analyze these data, many excellent machine learning techniques have been proposed in recent related work. In this paper, we try to perform the clustering of microarray data by combining the recently proposed Locality Preserving Projection (LPP) method with PCA, i.e. PCA-LPP. The comparison between PCA and PCA-LPP is performed based on two clustering algorithms, K-means and agglomerative hierarchical clustering. As we already known, clustering with the components extracted by PCA instead of the original variables does improve cluster quality. Moreover, our empirical study shows that by using LPP to perform further process the dimensions of components extracted by PCA can be further reduced and the quality of the clusters can be improved greatly meanwhile. Particularly, the first few components obtained by PCA-LPP capture more information of the cluster structure than those of PCA.

In the recent few years, web mining has become a hotspot of data mining with the development of Internet. Web pages classification is one of the essential techniques for web mining since classifying web pages of an interesting class is often the first step of mining the web. The high dimensional text vocabulary space is one of the main challenges of web pages. In this paper, we study the capabilities of Bayesian classifiers for web pages categorization. Several feature selection techniques, such as Chi Squared, Information Gain and Gain Ratio are used for selecting relevant words in web pages. Results on benchmark dataset show that the performances of Aggregating One-Dependence Estimators (AODE) and Hidden Naive Bayes (HNB) are both more competitive than other traditional methods.

As the e-commerce is done faster, there is a continuous flourishing of e-marketplaces. Matchmaking is an important aspect of e-commerce interactions. Recently, an approach has been taken to service matchmaking based on semantic Web technologies; the designed matching rule can be used to find the sellers' compatible advertisements for buyers. In this paper, we define three categories of attributes for the matchmaking service. Then we present three factors: semantic matching degree, semantic support and relational confidence to capture the semantic characteristics and relationships of the attributes. And we design a semantic ranking MSC combining the three factors to rank the results of advertisements matchmaking. MSC can capture the semantic aspect of matchmaking results; evaluation shows that MSC makes the process of matchmaking more accurately and the advertisement with the highest MSC is better

In this paper, a novel image registration method is proposed. In the proposed method, kernel independent component analysis (KICA) is applied to extract features from the image sets, and these features are input vectors of feedforward neural networks (FNN). Neural network outputs are those translation, rotation and scaling parameters with respect to reference and observed image sets. Comparative experiments are performed between KICA based method and other six feature extraction based method: principal component analysis (PCA), independent component analysis (ICA), kernel principal component analysis (KPCA), the discrete cosine transform (DCT), Zernike moment and the complete isometric mapping (Isomap). The results show that the proposed method is much improved not only at accuracy but also remarkably at robust to noise.