Frontiers of Information Technology & Electronic Engineering最新文献

Federated learning (FL), a cutting-edge distributed machine learning training paradigm, aims to generate a global model by collaborating on the training of client models without revealing local private data. The cooccurrence of non-independent and identically distributed (non-IID) and long-tailed distribution in FL is one challenge that substantially degrades aggregate performance. In this paper, we present a corresponding solution called federated dual-decoupling via model and logit calibration (FedDDC) for non-IID and long-tailed distributions. The model is characterized by three aspects. First, we decouple the global model into the feature extractor and the classifier to fine-tune the components affected by the joint problem. For the biased feature extractor, we propose a client confidence re-weighting scheme to assist calibration, which assigns optimal weights to each client. For the biased classifier, we apply the classifier re-balancing method for fine-tuning. Then, we calibrate and integrate the client confidence re-weighted logits with the re-balanced logits to obtain the unbiased logits. Finally, we use decoupled knowledge distillation for the first time in the joint problem to enhance the accuracy of the global model by extracting the knowledge of the unbiased model. Numerous experiments demonstrate that on non-IID and long-tailed data in FL, our approach outperforms state-of-the-art methods.

Orthogonal time–frequency space (OTFS) is a new modulation technique proposed in recent years for high Doppler wireless scenes. To solve the parameter estimation problem of the OTFS-integrated radar and communications system, we propose a parameter estimation method based on sparse reconstruction preprocessing to reduce the computational effort of the traditional weighted subspace fitting (WSF) algorithm. First, an OTFS-integrated echo signal model is constructed. Then, the echo signal is transformed to the time domain to separate the target angle from the range, and the range and angle of the detected target are coarsely estimated by using the sparse reconstruction algorithm. Finally, the WSF algorithm is used to refine the search with the coarse estimate at the center to obtain an accurate estimate. The simulations demonstrate the effectiveness and superiority of the proposed parameter estimation algorithm.

Fueled by the explosive growth of ultra-low-latency and real-time applications with specific computing and network performance requirements, the computing force network (CFN) has become a hot research subject. The primary CFN challenge is to leverage network resources and computing resources. Although recent advances in deep reinforcement learning (DRL) have brought significant improvement in network optimization, these methods still suffer from topology changes and fail to generalize for those topologies not seen in training. This paper proposes a graph neural network (GNN) based DRL framework to accommodate network traffic and computing resources jointly and efficiently. By taking advantage of the generalization capability in GNN, the proposed method can operate over variable topologies and obtain higher performance than the other DRL methods.

With the booming development of fifth-generation network technology and Internet of Things, the number of end-user devices (EDs) and diverse applications is surging, resulting in massive data generated at the edge of networks. To process these data efficiently, the innovative mobile edge computing (MEC) framework has emerged to guarantee low latency and enable efficient computing close to the user traffic. Recently, federated learning (FL) has demonstrated its empirical success in edge computing due to its privacy-preserving advantages. Thus, it becomes a promising solution for analyzing and processing distributed data on EDs in various machine learning tasks, which are the major workloads in MEC. Unfortunately, EDs are typically powered by batteries with limited capacity, which brings challenges when performing energy-intensive FL tasks. To address these challenges, many strategies have been proposed to save energy in FL. Considering the absence of a survey that thoroughly summarizes and classifies these strategies, in this paper, we provide a comprehensive survey of recent advances in energy-efficient strategies for FL in MEC. Specifically, we first introduce the system model and energy consumption models in FL, in terms of computation and communication. Then we analyze the challenges regarding improving energy efficiency and summarize the energy-efficient strategies from three perspectives: learning-based, resource allocation, and client selection. We conduct a detailed analysis of these strategies, comparing their advantages and disadvantages. Additionally, we visually illustrate the impact of these strategies on the performance of FL by showcasing experimental results. Finally, several potential future research directions for energy-efficient FL are discussed.

Multi-exit architecture allows early-stop inference to reduce computational cost, which can be used in resource-constrained circumstances. Recent works combine the multi-exit architecture with self-distillation to simultaneously achieve high efficiency and decent performance at different network depths. However, existing methods mainly transfer knowledge from deep exits or a single ensemble to guide all exits, without considering that inappropriate learning gaps between students and teachers may degrade the model performance, especially in shallow exits. To address this issue, we propose Multi-exit self-distillation with Appropriate TEachers (MATE) to provide diverse and appropriate teacher knowledge for each exit. In MATE, multiple ensemble teachers are obtained from all exits with different trainable weights. Each exit subsequently receives knowledge from all teachers, while focusing mainly on its primary teacher to keep an appropriate gap for efficient knowledge transfer. In this way, MATE achieves diversity in knowledge distillation while ensuring learning efficiency. Experimental results on CIFAR-100, TinyImageNet, and three fine-grained datasets demonstrate that MATE consistently outperforms state-of-the-art multi-exit self-distillation methods with various network architectures.

This paper proposes a kind of programmable logic element (PLE) based on Sense-Switch pFLASH technology. By programming Sense-Switch pFLASH, all three-bit look-up table (LUT3) functions, partial four-bit look-up table (LUT4) functions, latch functions, and d flip flop (DFF) with enable and reset functions can be realized. Because PLE uses a choice of operational logic (COOL) approach for the operation of logic functions, it allows any logic circuit to be implemented at any ratio of combinatorial logic to register. This intrinsic property makes it close to the basic application specific integrated circuit (ASIC) cell in terms of fine granularity, thus allowing ASIC-like cell-based mappers to apply all their optimization potential. By measuring Sense-Switch pFLASH and PLE circuits, the results show that the “on” state driving current of the Sense-Switch pFLASH is about 245.52 µA, and that the “off” state leakage current is about 0.1 pA. The programmable function of PLE works normally. The delay of the typical combinatorial logic operation AND3 is 0.69 ns, and the delay of the sequential logic operation DFF is 0.65 ns, both of which meet the requirements of the design technical index.

Adversarial training with online-generated adversarial examples has achieved promising performance in defending adversarial attacks and improving robustness of convolutional neural network models. However, most existing adversarial training methods are dedicated to finding strong adversarial examples for forcing the model to learn the adversarial data distribution, which inevitably imposes a large computational overhead and results in a decrease in the generalization performance on clean data. In this paper, we show that progressively enhancing the adversarial strength of adversarial examples across training epochs can effectively improve the model robustness, and appropriate model shifting can preserve the generalization performance of models in conjunction with negligible computational cost. To this end, we propose a successive perturbation generation scheme for adversarial training (SPGAT), which progressively strengthens the adversarial examples by adding the perturbations on adversarial examples transferred from the previous epoch and shifts models across the epochs to improve the efficiency of adversarial training. The proposed SPGAT is both efficient and effective; e.g., the computation time of our method is 900 min as against the 4100 min duration observed in the case of standard adversarial training, and the performance boost is more than 7% and 3% in terms of adversarial accuracy and clean accuracy, respectively. We extensively evaluate the SPGAT on various datasets, including small-scale MNIST, middle-scale CIFAR-10, and large-scale CIFAR-100. The experimental results show that our method is more efficient while performing favorably against state-of-the-art methods.

As cross-chain technologies enable interactions among different blockchains (hereinafter “chains”), multi-chain consensus is becoming increasingly important in blockchain networks. However, more attention has been paid to single-chain consensus schemes. Multi-chain consensus schemes with trusted miner participation have not been considered, thus offering opportunities for malicious users to launch diverse miner behavior (DMB) attacks on different chains. DMB attackers can be friendly in the consensus process on some chains, called mask chains, to enhance their trust value, while on others, called kill chains, they engage in destructive behaviors on the network. In this paper, we propose a multi-chain consensus scheme named Proof-of-DiscTrust (PoDT) to defend against DMB attacks. The idea of distinctive trust (DiscTrust) is introduced to evaluate the trust value of each user across different chains. The trustworthiness of a user is split into local and global trust values. A dynamic behavior prediction scheme is designed to enforce DiscTrust to prevent an intensive DMB attacker from maintaining strong trust by alternately creating true or false blocks on the kill chain. Three trusted miner selection algorithms for multi-chain environments can be implemented to select network miners, chain miners, and chain miner leaders, separately. Simulation results show that PoDT is secure against DMB attacks and more effective than traditional consensus schemes in multi-chain environments.

This paper deals with the search-and-rescue tasks of a mobile robot with multiple interesting targets in an unknown dynamic environment. The problem is challenging because the mobile robot needs to search for multiple targets while avoiding obstacles simultaneously. To ensure that the mobile robot avoids obstacles properly, we propose a mixed-strategy Nash equilibrium based Dyna-Q (MNDQ) algorithm. First, a multi-objective layered structure is introduced to simplify the representation of multiple objectives and reduce computational complexity. This structure divides the overall task into subtasks, including searching for targets and avoiding obstacles. Second, a risk-monitoring mechanism is proposed based on the relative positions of dynamic risks. This mechanism helps the robot avoid potential collisions and unnecessary detours. Then, to improve sampling efficiency, MNDQ is presented, which combines Dyna-Q and mixed-strategy Nash equilibrium. By using mixed-strategy Nash equilibrium, the agent makes decisions in the form of probabilities, maximizing the expected rewards and improving the overall performance of the Dyna-Q algorithm. Furthermore, a series of simulations are conducted to verify the effectiveness of the proposed method. The results show that MNDQ performs well and exhibits robustness, providing a competitive solution for future autonomous robot navigation tasks.

Due to factors such as motion blur, video out-of-focus, and occlusion, multi-frame human pose estimation is a challenging task. Exploiting temporal consistency between consecutive frames is an efficient approach for addressing this issue. Currently, most methods explore temporal consistency through refinements of the final heatmaps. The heatmaps contain the semantics information of key points, and can improve the detection quality to a certain extent. However, they are generated by features, and feature-level refinements are rarely considered. In this paper, we propose a human pose estimation framework with refinements at the feature and semantics levels. We align auxiliary features with the features of the current frame to reduce the loss caused by different feature distributions. An attention mechanism is then used to fuse auxiliary features with current features. In terms of semantics, we use the difference information between adjacent heatmaps as auxiliary features to refine the current heatmaps. The method is validated on the large-scale benchmark datasets PoseTrack2017 and PoseTrack2018, and the results demonstrate the effectiveness of our method.