The Internet of Things (IoT) revolution has led to unprecedented data generation, necessitating a shift from traditional centralized computing to more decentralized approaches. To address the challenges of data processing closer to the source, the paradigm of Mobile Edge Computing (MEC) has emerged. It facilitates task offloading to nearby edge servers, thereby reducing delay and enhancing privacy. However, limited computation resources at the edge necessitate intelligent resource allocation through effective scheduling to maintain Quality of Service (QoS). Typically, a task comprises multiple subtasks with inherent dependencies, some of which are locally dependent and unsuitable for offloading. In subtask scheduling, one must account for both inter-subtask and local dependencies, deploying different subtask types to near-optimal computing devices, whether edge servers or User Equipment (UEs). This requirement presents significant challenges to scheduling strategies. Furthermore, since task offloading requests are inherently online, without prior task information before their arrival, improper scheduling can result in resource wastage and increased delays. To tackle these challenges, we formulate the Online Multi-DAG task Scheduling with Dependency awareness (OMSD) problem within a DAG-MEC framework. This problem is modeled as an Integer Linear Programming (ILP) problem and proven to be NP-hard. We propose a Dynamic Priority List Scheduling (DPLS) algorithm to address this problem effectively. Our algorithm strategically determines subtask execution order by evaluating upward and downward ranks, task volume, and contention levels. Simulation results demonstrate that DPLS significantly outperforms existing benchmark algorithms regarding mean task completion time, server load balance, and maximum task completion time, offering a robust solution to the OMSD challenge in MEC environments.

Edge computing brings computational resources to network edge, enabling mobile devices (MDs) to offload computing-intensive tasks to nearby edge servers. This significantly reduces the energy consumption of MDs and supports latency-sensitive applications. Meanwhile, the advancement of full-duplex (FD) cell-free massive multiple-input multiple-output (MIMO) technology provides a promising opportunity to enhance end-edge communication efficiency, particularly in scenarios with coexisting uplink (UL) and downlink (DL) users. In this paper, we investigate the joint task offloading and resource scheduling problem in FD cell-free massive MIMO-enabled edge computing networks. The problem is formulated as a two-stage optimization framework. In the first stage, we develop a hybrid simulated annealing–particle swarm optimization (SA-PSO) algorithm, which incorporates the Metropolis criterion to enhance global search capability, aiming to maximize spectral efficiency. In the second stage, we propose a diffusion-augmented prioritized deep deterministic policy gradient (DAP-DDPG) algorithm. This algorithm integrates prioritized experience replay with diffusion models to minimize the total energy consumption of MDs while satisfying stringent latency constraints. Simulation results demonstrate that, compared with benchmark schemes, the proposed SA-PSO algorithm achieves a 13.4% to 146% improvement in spectral efficiency, while the DAP-DDPG algorithm reduces the energy consumption of MDs by 12.5% to 33.1%.

Diffusion models has threatened image authenticity by enabling highly realistic fakes. Proactive defense offers protection by adding a "protective layer" that resists such manipulation. However, current proactive defenses mainly focus on text-guided editing but are less effective for the more challenging exemplar-guided tasks. To bridge this gap, we propose Proactive Spatial-Frequency Defense (PSFD), a novel proactive defense for exemplar-guided image editing. PSFD leverages adversarial attack to add subtle perturbations to make images immune to editing. We apply protections in both the frequency and spatial domains. Spatial perturbation disrupts feature extraction by forcing visual encoders to map the image to "bad" representations. Frequency perturbation tweaks the high-frequency components to distort the image’s texture information. We design two optimization strategies: PSFD-U that aims to generate maximal variation and PSFD-T that seeks to achieve specific editing styles. Extensive experiments on MS-COCO and ImageNet demonstrate PSFD’s strong defensive capabilities and transferability.

Adversarial examples can assess the robustness of machine learning models, which has attracted the attention of many researchers to adversarial example generation methods. Transferability and imperceptibility stand out as two crucial metrics for evaluating the quality of adversarial examples. However, achieving a balance between these two indicators poses a formidable challenge. In this paper, we propose a low-frequency guided adversarial attack method (LGA) to generate adversarial examples with strong transferability and good imperceptibility. Specifically, we enhance the transferability of adversarial examples by increasing the diversity of attack algorithms, and introduce the guiding principle and the triplet loss constraint to ensure that the generated adversarial examples are optimized away from the class regions of the clean examples. We find that the low-frequency component in the frequency domain of the image contains the vast majority of the semantic information of the image. Therefore, we constrain the attack perturbations to low-frequency component space to enhance the covert nature while maintaining visual coherence, rendering the adversarial examples more difficult to perceive. We conduct extensive experiments on various models with different network structures and multiple defense strategies, and the experimental results demonstrate that our method outperforms existing methods in the tradeoff between transferability and imperceptibility, achieving the SOTA performance.

Fine-tuning Pre-trained Language Models (PLMs) is a popular Natural Language Processing (NLP) paradigm for addressing Named Entity Recognition (NER) tasks. However, neural network models often demonstrate poor generalization capabilities due to significant disparities between the knowledge learned by PLMs and the distribution of the target dataset, as well as data scarcity issues. In addition, token omission in predictions due to insufficient learning remains a challenge in NER. In this paper, we propose a kNN retrieval enhancement algorithm (WkNER) that incorporates word segmentation information to enhance the model’s generalization ability and alleviate the problem of missing entity tokens in prediction. The introduction of word segmentation information is used to preliminarily determine the boundaries of entities and alleviate the common prediction errors of missing tokens within entities made by the fine-tuned model. Secondly, we find that non-entities in the retrieval table contain a large amount of redundant information, and explore the effects of introducing non-entity information of different scales on the model. Experimental results show that our proposed method significantly improves the performance of baseline models, and achieves better or compared recognition accuracy than previous state-of-the-art models in multiple public Chinese and English datasets. Especially in low-resource scenarios, our method achieves higher accuracy on 20% of the dataset than the original method on the full dataset.

Recently, the lightweight single image super-resolution (SISR) model based on information distillation has attracted the attention of many researchers due to its ability to recover high-resolution images quickly. We reassess and delve into the advantages and disadvantages of information distillation structures, and propose an information scaling distillation network (ISDN) for lightweight single image super-resolution, which can accurately and efficiently restore high-resolution images. By optimizing the distillation branch and feature branch of the information distillation, we meticulously designed the stacked block deep scaling distillation block (DSDB) to enlarge the receptive field and increase the network depth. We mainly optimize and design from two aspects. Firstly, we extract redundant information in the distillation branch and integrate it into multiple layers to form deep information transmission. Secondly, we design a blueprint deep scaling residual (BDSR) in the feature branch, which can extract advanced semantic image information, compress and expand feature channels. The qualitative and quantitative results on various benchmark datasets demonstrate the advantages of our model in terms of model parameters, multiply-accumulate operations, test efficiency, and image reconstruction quality. Code is available at https://github.com/ycLi-CV/ISDN-main.

As one of the classic problems of pattern recognition, the online Handwritten Chinese Character Recognition (OLHCCR) has attracted the attention of many researchers. Yet, it remains challenging due to complex glyphs, numerous strokes, and huge categories. Existing methods utilize temporal features or spatial features to recognize handwritten characters, which results in recognition errors due to the character with non-standard stroke order. This paper proposes a new OLHCCR model based on 1-D Convolution and Two-Streams Transformers. The model has a 1-D Transformer and a Vision Transformer, and the 1-D Transformer contains a 1-D Convolution layer and Transformers, that is, the model has overall structure of Two-Streams Transformers with 1-D Convolution. So, the model is named as C-TST. It can fuse temporal and spatial features of Chinese character to achieve high recognition accuracy and fast recognition speed. Specifically, each online handwritten Chinese character is represented by a trajectory sequence. The original trajectory sequence is preprocessed to enhance the information density of each trajectory point and features difference among trajectory points. Then, the result after preprocessing is input into the 1-D convolution layer to obtain shallow temporal features, which are used also as the input of the Transformers to capture the temporal features. Simultaneously, character image is generated by processing the original trajectory sequence, and then fed into the Vision Transformer to capture the spatial features. By fusing the captured temporal and spatial features of online handwritten Chinese character, the proposed C-TST achieves a recognition accuracy of 97.90% on ICDAR-2013 and a state-of-the-art recognition accuracy of 97.38% on IAHCC-UCAS2016. The code will be available athttps://github.com/cwnuiot/Two-Streams-Transformer

In recent years, a number of lightweight single-image super-resolution (SISR) network methods heave been proposed. However, most existing approaches do not make full use of the information before and after the convolution and the high-frequency information of the image. In this paper, we propose a lightweight deep feature aggregation network (DFAnet), which fuses the outputs of all the deep feature aggregation blocks (DFAB) through the designed nonlinear global feature fusion (NGFF) module. The DFAB includes deep feature aggregation structure (DFAS) and non-local sparse attention mechanism (NLSA), where DFAS consists of several aggregation convolutions and information rearrangement operations. Then the output of DFAS is assessed by non-local sparse attention module to form our basic block DFAB. Furthermore, we design a nonlinear global feature fusion (NGFF) module to learn the nonlinear relationship between the output of each DFAB, which encourages every DFAB to pay attention to different patterns of the image. The qualitative and quantitative experimental results on several benchmark datasets show the proposed method achieves the state-of-the-art results in term of reconstruction accuracy, computational complexity and memory consumption.

The rapid development of Internet of Vehicle (IoV) technology has brought the improvement of user experience satisfaction. Subsequently, a variety of vehicle applications put forward higher requirements for information transmission and storage space. A new intelligent edge content caching mechanism is proposed to adapt to the dynamic change of vehicles and differences of storage space of existing facilities. Firstly, based on Long Short Term Memory(LSTM), a resource request prediction model is proposed to effectively estimate the number of content request over a period of time. Then, considering the popularity of the requested content and the preferences of different vehicle users, this paper proposes Dynamic content cache algorithm(DCCA). Among them, the model is built through Markov Decision Process(MDP) to update and optimize the request content using Double Deep Q-Network(DDON). This experiment shows that DCCA is able to improve the hit rate by 50% and reduce the average delay by 20ms in the face of complex and varied request contents with limited cache capacity

Deep neural network (DNN) models have been widely used in many tasks due to their superior performance. However, DNN models are usually vulnerable to adversarial example attacks, which limits their applications in many safety-critic scenarios. How to effectively detect adversarial examples to enhance the robustness of DNN models has attracted much attention in recent years. Most adversarial example detection methods require modifying or retraining the model, which is impractical and reduces the classification accuracy of normal examples. In this paper, we propose an adversarial example detection approach that does not require modification of the DNN models and meanwhile retains the classification accuracy of normal examples. The key observation is that when we transform the input example with some operations (e.g., masking a pixel with a reference value), feed the transformed example to the target model, and use the output of the intermediate layers to predict the label of the example, the generated label sequences of adversarial examples will be extremely discrepant but the label sequences of normal examples keep nearly unchanged. Motivated by this observation, we design an approach to detect adversarial examples based on the label sequence discrepancy (LSD) of the given examples. The experimental results against five mainstream adversarial attacks on three benchmark datasets demonstrate that LSD outperforms the state-of-the-art solutions in the detection rate of adversarial examples. Moreover, LSD performs well at various confidence levels and exhibits good generalizability between different attacks.

Recently, the siamese convolutional neural network plays an important role in the field of visual tracking, which can obtain high tracking accuracy and good real-time performance. However, the requirement of offline training a specific neural network results in the hardware source and time consumption. In order to improve the tracking efficiency and save computation resources, we adopt pre-trained densely connected neural network to extract robust target features. Since the pre-trained model is mainly used for classification task, it is not appropriate to directly adopt these deep features for visual tracking. We design a regression network to measure the importance of each channel to the target, and then propose a weighting fusion strategy to select the suitable features for visual tracking. Besides, we provide deep analysis about the proposed channel weighting method to demonstrate the superiority of this method through visualization of feature heatmaps. Extensive experiments on four classical benckmarks show that compared with state-of-the-art methods, our algorithm achieves the best results on several standard indicators and comparable results on other indicators.

Single image super-resolution attempts to reconstruct a high-resolution (HR) image from its corresponding low-resolution (LR) image, which has been a research hotspot in computer vision and image processing for decades. To improve the accuracy of super-resolution images, many works adopt very deep networks to model the translation from LR to HR, resulting in memory and computation consumption. In this article, we design a lightweight dense connection distillation network by combining the feature fusion units and dense connection distillation blocks (DCDB) that include selective cascading and dense distillation components. The dense connections are used between and within the distillation block, which can provide rich information for image reconstruction by fusing shallow and deep features. In each DCDB, the dense distillation module concatenates the remaining feature maps of all previous layers to extract useful information, the selected features are then assessed by the proposed layer contrast-aware channel attention mechanism, and finally the cascade module aggregates the features. The distillation mechanism helps to reduce training parameters and improve training efficiency, and the layer contrast-aware channel attention further improves the performance of model. The quality and quantity experimental results on several benchmark datasets show the proposed method performs better tradeoff in term of accuracy and efficiency.

This paper considers the multi-user MIMO collaborative spatial multiplexing uplink scheduling scheme in multi-cell time-division duplexing system. An transmitter centric (TC) beamforming and scheduling scheme has been proposed to reduce the inter-cell interference opportunistically. Most previous works in the area of mimo beamforming and scheduling are aiming at improve the signal to interference plus noise power ratio at receiver directly. These receiver centric (RC) schemes requires interference covariance at receiver. However, in multi-user system, the uplink interference at base station (BS) is fluctuant and can't be acquired for scheduling due to the dynamics of the scheduled user set in other cells. It makes the uplink scheduling problem challenging. Our proposed TC scheme considers the pollution which is defined as the sum power of user signal arriving at other BSs. User selects the transmit beamforming vector with highest signal to pollution plus noise ratio (SPR) and informs its home BS. To enhance the signal and suppress the pollution, a scheduling criterion is derived based on SPR metric with awareness of intra-cell interference. The proposed scheduling scheme is compared with conventional RC and single-user TC spatial multiplexing in the scenarios with various antenna configurations, BS and user numbers. The simulation result shows that our proposed TC scheme can better reduce the pollution and significantly improve the system capacity.

We propose an efficient low bit error rate (BER) and low complexity multiple-input multiple-output (MIMO) multiuser detection (MUD) method for use with multiuser MIMO orthogonal frequency division multiplexing (OFDM) systems. It is a hybrid method combining a multiuser-interference-cancellation-based decision feedback equalizer using error feedback filter (MIMO MIC DFE-EFF) and a differential algorithm. The proposed method, termed ‘MIMO MIC DFE-EFF with a differential algorithm’ for short, has a multiuser feedback structure. We describe the schemes of MIMO MIC DFE-EFF and MIMO MIC DFE-EFF with a differential algorithm, and compare their minimum mean square error (MMSE) performance and computational complexity. Simulation results show that a significant performance gain can be achieved by employing the MIMO MIC DFE-EFF detection algorithm in the context of a multiuser MIMO-OFDM system over frequency selective Rayleigh channel. MIMO MIC DFE-EFF with the differential algorithm improves both computational efficiency and BER performance in a multistage structure relative to conventional DFE-EFF, though there is a small reduction in system performance compared with MIMO MIC DFE-EFF without the differential algorithm.

In this paper, we consider a multiuser relay network under amplify and forward (AF) scheme, where each mobile station (MS) is supported by a unique relay station (RS). While the RSs forwards the scaled signal to the MSs, the interference caused by the multiuser nature of the system propagates via the RSs, which significantly degrades the performance gain of the relay technique. We propose two beamforming methods, zero forcing (ZF) method and maximizing signal to leakage ratio (MSLR) method, to suppress the multiuser interference in the relay networks. ZF method can effectively cancel multiuser interference for all MSs and RSs, while it requires a relatively large number of transmit antennas equipped at the base station (BS). The other method, MSLR, aims to maximize the received signal at each MS and RS, while minimize the energy leaking to other MSs and RSs. The advantage of MSLR method is that it does not impose a condition on the relation between the number of transmit antennas equipped at the BS and the receive antennas at the RSs and MSs. Simulations show that MSLR method has significant performance gains over ZF method.

In 802.16e OFDMA systems, noise and co-channel interference on pilots degrade the accuracy of channel state information (CSI) obtained from channel estimation (CE) and system performance is deteriorated. To evaluate the system performance loss due to imperfect CSI, CE model should be used to incorporate noise and interference's impact on CSI into system level simulator (SLS). In this paper, a CE model, which considers 802.16e OFDMA system's pilot pattern and subcarrier randomization, are proposed. The signal model of 802.16e OFDMA system's CE under noise and interference is presented first. Then the error of CSI estimation by LMMSE channel estimator is derived. The proposed CE model covers the main parts of CSI error. Simulation results show it is suitable for SLS by providing a good approximation to estimated CSI instead of performing true LMMSE CE.