To fully exploit the attribute information in graphs and dynamically fuse the features from different modalities, this letter proposes the Attributed Graph Clustering Network with Adaptive Feature Fusion (AGC-AFF) for graph clustering, where an Attribute Reconstruction Graph Autoencoder (ARGAE) with masking operation learns to reconstruct the node attributes and adjacency matrix simultaneously, and an Adaptive Feature Fusion (AFF) mechanism dynamically fuses the features from different modules based on node attention. Extensive experiments on various benchmark datasets demonstrate the effectiveness of the proposed method.

Using deep learning (DL) to achieve direction-of-arrival (DOA) estimation is an open and meaningful exploration. Existing DL-based methods achieve DOA estimation by spectrum regression or multi-label classification task. While, both of them face the problem of off-grid errors. In this paper, we proposed a cascaded deep neural network (DNN) framework named as off-grid network (OGNet) to provide accurate DOA estimation in the case of off-grid. The OGNet is composed of an autoencoder consisted by fully connected (FC) layers and a deep convolutional neural network (CNN) with 2-dimensional convolutional layers. In the proposed OGNet, the off-grid error is modeled into labels to achieve off-grid DOA estimation based on its sparsity. As compared to the state-of-the-art grid-based methods, the OGNet shows advantages in terms of precision and resolution. The effectiveness and superiority of the OGNet are demonstrated by extensive simulation experiments in different experimental conditions.

Unmanned aerial vehicle (UAV)-assisted systems have attracted a lot of attention due to its high probability of line-of-sight (LoS) connections and flexible deployment. In this paper, we aim to minimize the upload time required for the UAV to collect information from the sensor nodes in disaster scenario, while optimizing the deployment position of UAV. In order to get the deployment solution quickly, a data-driven approach is proposed in which an optimization strategy acts as the expert. Considering that images could capture the spatial configurations well, we use a convolutional neural network (CNN) to learn how to place the UAV. In the end, the simulation results demonstrate the effectiveness and generalization of the proposed method. After training, our CNN can generate UAV configuration faster than the general optimization-based algorithm.

Single-image dehazing is a challenging task in computer vision research. Aiming at the limitations of traditional convolutional neural network representation capabilities and the high computational overhead of the self-attention mechanism in recent years, we proposed image attention and designed a single image dehazing network based on the image attention: IAD-Net. The proposed image attention is a plug-and-play module with the ability of global modeling. IAD-Net is a parallel network structure that combines the global modeling ability of image attention and the local modeling ability of convolution, so that the network can learn global and local features. The proposed network model has excellent feature learning ability and feature expression ability, has low computational overhead, and also improves the detail information of hazy images. Experiments verify the effectiveness of the image attention module and the competitiveness of IAD-Net with state-of-the-art methods.

This paper reports a spatial extrapolation of the sound field with a physics-informed neural network. We investigate the spatial extrapolation of the room impulse responses with physics-informed SIREN architecture. Furthermore, we proposed a noise-robust extrapolation method by introducing a tolerance term to the loss function.

Software refactoring is an important process in software development. During software refactoring, code smell is a popular research topic that refers to design or implementation flaws in the software. Large class is one of the most concerning code smells in software refactoring. Detecting and refactoring such problem has a profound impact on software quality. In past years, software metrics and clustering techniques have commonly been used for the large class detection. However, deep-learning-based approaches have also received considerable attention in recent studies. In this study, we apply graph neural networks (GNNs), an important division of deep learning, to address the problem of large class detection. First, to support the extensive data requirements of the deep learning task, we apply a semiautomatic approach to generate a substantial number of data samples. Next, we design a new type of directed heterogeneous graph (DHG) as an input graph using the methods similarity matrix and software metrics. We construct an input graph for each class sample and make the graph classification with GNNs to identify the smelly classes. In our experiments, we apply three typical GNN model architectures for large class detection and compare the results with those of previous studies. The results show that the proposed approach can achieve more accurate and stable detection performance.

This letter introduces an innovation for the heterogeneous storage architecture of AI chips, specifically focusing on the integration of six transistors(6T) and eight transistors(8T) hybrid SRAM. Traditional approaches to reducing SRAM power consumption typically involve lowering the operating voltage, a method that often substantially diminishes the recognition rate of neural networks. However, the innovative design detailed in this letter amalgamates the strengths of both SRAM types. It operates at a voltage lower than conventional SRAM, thereby significantly reducing the power consumption in neural networks without compromising performance.

Medical imaging plays an indispensable role in precise patient diagnosis. The integration of deep learning into medical diagnostics is becoming increasingly common. However, existing deep learning models face performance and efficiency challenges, especially in resource-constrained scenarios. To overcome these challenges, we introduce a novel dendritic neural efficientnet model called DEN, inspired by the function of brain neurons, which efficiently extracts image features and enhances image classification performance. Assessments on a diabetic retinopathy fundus image dataset reveal DEN’s superior performance compared to EfficientNet and other classical neural network models.

Edge devices with strict safety and reliability requirements, such as autonomous driving cars, industrial robots, and drones, necessitate software verification on such devices before operation. The human cost and time required for this analysis constitute a barrier in the cycle of software development and updating. In particular, the final verification at the edge device should at least strictly confirm that the updated software is not degraded from the current it. Since the edge device does not have the correct data, it is necessary for a human to judge whether the difference between the updated software and the operating it is due to degradation or improvement. Therefore, this verification is very costly. This paper proposes a novel automated method for efficient verification on edge devices of an object detection AI, which has found practical use in various applications. In the proposed method, a target object existence detector (TOED) (a simple binary classifier) judges whether an object in the recognition target class exists in the region of a prediction difference between the AI’s operating and updated versions. Using the results of this TOED judgement and the predicted difference, an automated verification system for the updated AI was constructed. TOED was designed as a simple binary classifier with four convolutional layers, and the accuracy of object existence judgment was evaluated for the difference between the predictions of the YOLOv5 L and X models using the Cityscapes dataset. The results showed judgement with more than 99.5% accuracy and 8.6% over detection, thus indicating that a verification system adopting this method would be more efficient than simple analysis of the prediction differences.

Diagnosis at an early stage is clinically important for the cure of skin cancer. However, since some skin cancers have similar intuitive characteristics, and dermatologists rely on subjective experience to distinguish skin cancer types, the accuracy is often suboptimal. Recently, the introduction of computer methods in the medical field has better assisted physicians to improve the recognition rate but some challenges still exist. In the face of massive dermoscopic image data, residual network (ResNet) is more suitable for learning feature relationships inside big data because of its deeper network depth. Aiming at the deficiency of ResNet, this paper proposes a multi-region feature extraction and raising dimension matching method, which further improves the utilization rate of medical image features. This method firstly extracted rich and diverse features from multiple regions of the feature map, avoiding the deficiency of traditional residual modules repeatedly extracting features in a few fixed regions. Then, the fused features are strengthened by up-dimensioning the branch path information and stacking it with the main path, which solves the problem that the information of two paths is not ideal after fusion due to different dimensionality. The proposed method is experimented on the International Skin Imaging Collaboration (ISIC) Archive dataset, which contains more than 40,000 images. The results of this work on this dataset and other datasets are evaluated to be improved over networks containing traditional residual modules and some popular networks.

In underwater acoustic communication systems based on orthogonal frequency division multiplexing (OFDM), taking clipping to reduce the peak-to-average power ratio leads to nonlinear distortion of the signal, making the receiver unable to recover the faded signal accurately. In this letter, an Aquila optimizer-based convolutional attention block stacked network (AO-CABNet) is proposed to replace the receiver to improve the ability to recover the original signal. Simulation results show that the AO method has better optimization capability to quickly obtain the optimal parameters of the network model, and the proposed AO-CABNet structure outperforms existing schemes.

In this paper, we propose a method for video reflection removal using a video restoration framework with enhanced deformable networks (EDVR). We examine the effect of each module in EDVR on video reflection removal and modify the models using 3D convolutions. The performance of each modified model is evaluated in terms of the RMSE between the structural similarity (SSIM) and the smoothed SSIM representing temporal consistency.

Neural networks (NNs) fail to perform well or make excessive predictions when predicting out-of-distribution or unseen datasets. In contrast, Bayesian neural networks (BNNs) can quantify the uncertainty of their inference to solve this problem. Nevertheless, BNNs have not been widely adopted owing to their increased memory and computational cost. In this study, we propose a novel approach to extend binary neural networks by introducing a probabilistic interpretation of binary weights, effectively converting them into BNNs. The proposed approach can reduce the number of weights by half compared to the conventional method. A comprehensive comparative analysis with established methods like Monte Carlo dropout and Bayes by backprop was performed to assess the performance and capabilities of our proposed technique in terms of accuracy and capturing uncertainty. Through this analysis, we aim to provide insights into the advantages of this Bayesian extension.

The development of energy-efficient neural network hardware using magnetic tunnel junction (MTJ) devices has been widely investigated. One of the issues in the use of MTJ devices is large write energy. Since MTJ devices show stochastic behaviors, a large write current with enough time length is required to guarantee the certainty of the information held in MTJ devices. This paper demonstrates that quantized neural networks (QNNs) exhibit high tolerance to bit errors in weights and an output feature map. Since probabilistic switching errors in MTJ devices do not have always a serious effect on the performance of QNNs, large write energy is not required for reliable switching operations of MTJ devices. Based on the evaluation results, we achieve about 80% write-energy reduction on buffer memory compared to the conventional method. In addition, it is demonstrated that binary representation exhibits higher bit-error tolerance than the other data representations in the range of large error rates.

Currently, a significant portion of acoustic scene categorization (ASC) research is centered around utilizing Convolutional Neural Network (CNN) models. This preference is primarily due to CNN’s ability to effectively extract time-frequency information from audio recordings of scenes by employing spectrum data as input. The expression of many dimensions can be achieved by utilizing 2D spectrum characteristics. Nevertheless, the diverse interpretations of the same object’s existence in different positions on the spectrum map can be attributed to the discrepancies between spectrum properties and picture qualities. The lack of distinction between different aspects of input information in ASC-based CNN networks may result in a decline in system performance. Considering this, a feature pyramid segmentation (FPS) approach based on CNN is proposed. The proposed approach involves utilizing spectrum features as the input for the model. These features are split based on a preset scale, and each segment-level feature is then fed into the CNN network for learning. The SoftMax classifier will receive the output of all feature scales, and these high-level features will be fused and fed to it to categorize different scenarios. The experiment provides evidence to support the efficacy of the FPS strategy and its potential to enhance the performance of the ASC system.

Convolutional neural network (CNN) is widely used in acoustic scene classification (ASC) tasks. In most cases, local convolution is utilized to gather time-frequency information between spectrum nodes. It is challenging to adequately express the non-local link between frequency domains in a finite convolution region. In this paper, we propose a dual-path convolutional neural network based on band interaction block (DCNN-bi) for ASC, with mel-spectrogram as the model’s input. We build two parallel CNN paths to learn the high-frequency and low-frequency components of the input feature. Additionally, we have created three band interaction blocks (bi-blocks) to explore the pertinent nodes between various frequency bands, which are connected between two paths. Combining the time-frequency information from two paths, the bi-blocks with three distinct designs acquire non-local information and send it back to the respective paths. The experimental results indicate that the utilization of the bi-block has the potential to improve the initial performance of the CNN substantially. Specifically, when applied to the DCASE 2018 and DCASE 2020 datasets, the CNN exhibited performance improvements of 1.79% and 3.06%, respectively.

In recent years, the increasing complexity of deep network structures has hindered their application in small resource constrained hardware. Therefore, we urgently need to compress and accelerate deep network models. Channel pruning is an effective method to compress deep neural networks. However, most existing channel pruning methods are prone to falling into local optima. In this paper, we propose a channel pruning method via Improved Grey Wolf Optimizer Pruner which called IGWO-Pruner to prune redundant channels of convolutional neural networks. It identifies pruning ratio of each layer by using Improved Grey Wolf algorithm, and then fine-tuning the new pruned network model. In experimental section, we evaluate the proposed method in CIFAR datasets and ILSVRC-2012 with several classical networks, including VGGNet, GoogLeNet and ResNet-18/34/56/152, and experimental results demonstrate the proposed method is able to prune a large number of redundant channels and parameters with rare performance loss.

Handover is an important property of cellular communication that enables the user to move from one cell to another without losing the connection. It is a very crucial process for the quality of the user’s experience because it may interrupt data transmission. Therefore, good handover management is very important in the current and future cellular systems. Several techniques have been employed to improve the handover performance, usually to increase the probability of a successful handover. One of the techniques is predictive handover which predicts the target cell using some methods other than the traditional measurement-based algorithm, including using machine learning. Several studies have been conducted in the implementation of predictive handover, most of them by modifying the internal algorithm of existing network elements, such as the base station. We implemented a predictive handover algorithm using an intelligent node outside the existing network elements to minimize the modification of the network and to create modularity in the system. Using a recently standardized Open Radio Access Network (O-RAN) Near Realtime Radio Intelligent Controller (Near-RT RIC), we created a modular application that can improve the handover performance by determining the target cell using machine learning techniques. In our previous research, we modified The Near-RT RIC original software that is using vector autoregression to determine the target cell by predicting the throughput of each neighboring cell. We also modified the method using a Multi-Layer Perceptron (MLP) neural network. In this paper, we redesigned the neural network using Long Short-Term Memory (LSTM) that can better handle time series data. We proved that our proposed LSTM-based machine learning algorithms used in Near-RT RIC can improve the handover performance compared to the traditional measurement-based algorithm.

A 0.13mJ/prediction with 68.6% accuracy wired-logic deep neural network (DNN) processor is developed in a single 16-nm field-programmable gate array (FPGA) chip. Compared with conventional von-Neumann architecture DNN processors, the energy efficiency is greatly improved by eliminating DRAM/BRAM access. A technical challenge for conventional wired-logic processors is the large amount of hardware resources required for implementing large-scale neural networks. To implement a large-scale convolutional neural network (CNN) into a single FPGA chip, two technologies are introduced: (1) a sparse neural network known as a non-linear neural network (NNN), and (2) a newly developed raster-scan wired-logic architecture. Furthermore, a novel high-level synthesis (HLS) technique for wired-logic processor is proposed. The proposed HLS technique enables the automatic generation of two key components: (1) Verilog-hardware description language (HDL) code for a raster-scan-based wired-logic processor and (2) test bench code for conducting equivalence checking. The automated process significantly mitigates the time and effort required for implementation and debugging. Compared with the state-of-the-art FPGA-based processor, 238 times better energy efficiency is achieved with only a slight decrease in accuracy on the CIFAR-100 task. In addition, 7 times better energy efficiency is achieved compared with the state-of-the-art network-optimized application-specific integrated circuit (ASIC).

Federated learning is a distributed machine learning approach in which clients train models locally with their own data and upload them to a server so that their trained results are shared between them without uploading raw data to the server. There are some challenges in federated learning, such as communication size reduction and client heterogeneity. The former can mitigate the communication overheads, and the latter can allow the clients to choose proper models depending on their available compute resources. To address these challenges, in this paper, we utilize Neural ODE based models for federated learning. The proposed flexible federated learning approach can reduce the communication size while aggregating models with different iteration counts or depths. Our contribution is that we experimentally demonstrate that the proposed federated learning can aggregate models with different iteration counts or depths. It is compared with a different federated learning approach in terms of the accuracy. Furthermore, we show that our approach can reduce communication size by up to 89.4% compared with a baseline ResNet model using CIFAR-10 dataset.