本文基于衍射神经网络平台,探究经典修正线性单元(rectified linear unit, ReLU)及其衍生激活函数对光学神经网络(ONNs)推理能力的影响,阐明非线性函数与衍射神经网络结合方式与其学习能力的关联。结果显示,衍射网络训练适应性强,但不...本文基于衍射神经网络平台,探究经典修正线性单元(rectified linear unit, ReLU)及其衍生激活函数对光学神经网络(ONNs)推理能力的影响,阐明非线性函数与衍射神经网络结合方式与其学习能力的关联。结果显示,衍射网络训练适应性强,但不当的非线性函数可能降低其性能。例如,在5层网络中,在每层后添加RTReLU(rectified translational linear unit)因光强衰减导致在MNIST测试集上的分类准确率降至91.4%,低于纯线性网络的92.6%;而每层添加PReLU(parametric rectified linear unit)则保留阈值后信息,使准确率提升至95.8%。在3层网络中,因为较少的网络深度降低了光强的损耗,在每层后添加RTReLU后推理能力优于线性网络。进一步发现,在网络中适当位置添加单个ReLU激活函数可大幅提升性能,如5层网络仅最后一层添加RTReLU可达峰值准确率96.6%。展开更多
Reliable human action recognition(HAR)in video sequences is critical for a wide range of applications,such as security surveillance,healthcare monitoring,and human-computer interaction.Several automated systems have b...Reliable human action recognition(HAR)in video sequences is critical for a wide range of applications,such as security surveillance,healthcare monitoring,and human-computer interaction.Several automated systems have been designed for this purpose;however,existing methods often struggle to effectively integrate spatial and temporal information from input samples such as 2-stream networks or 3D convolutional neural networks(CNNs),which limits their accuracy in discriminating numerous human actions.Therefore,this study introduces a novel deeplearning framework called theARNet,designed for robustHAR.ARNet consists of two mainmodules,namely,a refined InceptionResNet-V2-based CNN and a Bi-LSTM(Long Short-Term Memory)network.The refined InceptionResNet-V2 employs a parametric rectified linear unit(PReLU)activation strategy within convolutional layers to enhance spatial feature extraction fromindividual video frames.The inclusion of the PReLUmethod improves the spatial informationcapturing ability of the approach as it uses learnable parameters to adaptively control the slope of the negative part of the activation function,allowing richer gradient flow during backpropagation and resulting in robust information capturing and stable model training.These spatial features holding essential pixel characteristics are then processed by the Bi-LSTMmodule for temporal analysis,which assists the ARNet in understanding the dynamic behavior of actions over time.The ARNet integrates three additional dense layers after the Bi-LSTM module to ensure a comprehensive computation of both spatial and temporal patterns and further boost the feature representation.The experimental validation of the model is conducted on 3 benchmark datasets named HMDB51,KTH,and UCF Sports and reports accuracies of 93.82%,99%,and 99.16%,respectively.The Precision results of HMDB51,KTH,and UCF Sports datasets are 97.41%,99.54%,and 99.01%;the Recall values are 98.87%,98.60%,99.08%,and the F1-Score is 98.13%,99.07%,99.04%,respectively.These results highlight the robustness of the ARNet approach and its potential as a versatile tool for accurate HAR across various real-world applications.展开更多
为了研究出一种快速、高效的玉米病害识别方法,针对玉米叶片病害识别问题,本文以灰斑病、南方锈病、小斑病、锈病、叶斑等5种常见的玉米叶片病害为研究对象,提出一种基于改进ResNet18神经网络的玉米病害识别方法。通过在ResNet18网络的...为了研究出一种快速、高效的玉米病害识别方法,针对玉米叶片病害识别问题,本文以灰斑病、南方锈病、小斑病、锈病、叶斑等5种常见的玉米叶片病害为研究对象,提出一种基于改进ResNet18神经网络的玉米病害识别方法。通过在ResNet18网络的基础上引入金字塔卷积(Pyramidal Convolution)可以在玉米复杂的生长环境中利用多尺度的特征信息来提高模型对单叶片的识别和定位能力,以有效加快模型的收敛速度并显著提高模型的病害识别准确率;将残差结构的激活函数替换为PReLU(Parametric Rectified Linear Unit)激活函数避免模型训练过程中的神经元死亡。在收集的真实玉米叶片病害数据集上进行的实验表明,与原始ResNet18残差网络相比,本文提出的模型在玉米叶片病害识别的准确率、精确度、召回率、F1分数分别提升了1.86%、1.78%、1.78%、1.87%;模型的参数尺寸减小了1.85%。该模型可作为一种检测复杂生长环境下玉米叶片病害的有效方法。展开更多
The subcellular localization of human proteins is vital for understanding the structure of human cells.Proteins play a significant role within human cells,as many different groups of proteins are located in a specific...The subcellular localization of human proteins is vital for understanding the structure of human cells.Proteins play a significant role within human cells,as many different groups of proteins are located in a specific location to perform a particular function.Understanding these functions will help in discoveringmany diseases and developing their treatments.The importance of imaging analysis techniques,specifically in proteomics research,is becoming more prevalent.Despite recent advances in deep learning techniques for analyzing microscopy images,classification models have faced critical challenges in achieving high performance.Most protein subcellular images have a significant class imbalance.We use oversampling and under sampling techniques in this research to overcome this issue.We have used a Convolutional Neural Network(CNN)model called GapNet-PL for the multi-label classification task on the Human Protein Atlas Classification(HPA)Dataset.Authors have found that the ParametricRectified LinearUnit(PreLU)activation function is better than the Scaled Exponential LinearUnit(SeLU)activation function in the GapNet-PL model in most classification metrics.The results showed that the GapNet-PL model with the PReLU activation function achieved an area under the ROC curve(AUC)equal to 0.896,an F1 score of 0.541,and a recall of 0.473.展开更多
文摘本文基于衍射神经网络平台,探究经典修正线性单元(rectified linear unit, ReLU)及其衍生激活函数对光学神经网络(ONNs)推理能力的影响,阐明非线性函数与衍射神经网络结合方式与其学习能力的关联。结果显示,衍射网络训练适应性强,但不当的非线性函数可能降低其性能。例如,在5层网络中,在每层后添加RTReLU(rectified translational linear unit)因光强衰减导致在MNIST测试集上的分类准确率降至91.4%,低于纯线性网络的92.6%;而每层添加PReLU(parametric rectified linear unit)则保留阈值后信息,使准确率提升至95.8%。在3层网络中,因为较少的网络深度降低了光强的损耗,在每层后添加RTReLU后推理能力优于线性网络。进一步发现,在网络中适当位置添加单个ReLU激活函数可大幅提升性能,如5层网络仅最后一层添加RTReLU可达峰值准确率96.6%。
基金supported and funded by theDeanship of Scientific Research at ImamMohammad Ibn Saud Islamic University(IMSIU)(grant number IMSIU-DDRSP2504).
文摘Reliable human action recognition(HAR)in video sequences is critical for a wide range of applications,such as security surveillance,healthcare monitoring,and human-computer interaction.Several automated systems have been designed for this purpose;however,existing methods often struggle to effectively integrate spatial and temporal information from input samples such as 2-stream networks or 3D convolutional neural networks(CNNs),which limits their accuracy in discriminating numerous human actions.Therefore,this study introduces a novel deeplearning framework called theARNet,designed for robustHAR.ARNet consists of two mainmodules,namely,a refined InceptionResNet-V2-based CNN and a Bi-LSTM(Long Short-Term Memory)network.The refined InceptionResNet-V2 employs a parametric rectified linear unit(PReLU)activation strategy within convolutional layers to enhance spatial feature extraction fromindividual video frames.The inclusion of the PReLUmethod improves the spatial informationcapturing ability of the approach as it uses learnable parameters to adaptively control the slope of the negative part of the activation function,allowing richer gradient flow during backpropagation and resulting in robust information capturing and stable model training.These spatial features holding essential pixel characteristics are then processed by the Bi-LSTMmodule for temporal analysis,which assists the ARNet in understanding the dynamic behavior of actions over time.The ARNet integrates three additional dense layers after the Bi-LSTM module to ensure a comprehensive computation of both spatial and temporal patterns and further boost the feature representation.The experimental validation of the model is conducted on 3 benchmark datasets named HMDB51,KTH,and UCF Sports and reports accuracies of 93.82%,99%,and 99.16%,respectively.The Precision results of HMDB51,KTH,and UCF Sports datasets are 97.41%,99.54%,and 99.01%;the Recall values are 98.87%,98.60%,99.08%,and the F1-Score is 98.13%,99.07%,99.04%,respectively.These results highlight the robustness of the ARNet approach and its potential as a versatile tool for accurate HAR across various real-world applications.
文摘为了研究出一种快速、高效的玉米病害识别方法,针对玉米叶片病害识别问题,本文以灰斑病、南方锈病、小斑病、锈病、叶斑等5种常见的玉米叶片病害为研究对象,提出一种基于改进ResNet18神经网络的玉米病害识别方法。通过在ResNet18网络的基础上引入金字塔卷积(Pyramidal Convolution)可以在玉米复杂的生长环境中利用多尺度的特征信息来提高模型对单叶片的识别和定位能力,以有效加快模型的收敛速度并显著提高模型的病害识别准确率;将残差结构的激活函数替换为PReLU(Parametric Rectified Linear Unit)激活函数避免模型训练过程中的神经元死亡。在收集的真实玉米叶片病害数据集上进行的实验表明,与原始ResNet18残差网络相比,本文提出的模型在玉米叶片病害识别的准确率、精确度、召回率、F1分数分别提升了1.86%、1.78%、1.78%、1.87%;模型的参数尺寸减小了1.85%。该模型可作为一种检测复杂生长环境下玉米叶片病害的有效方法。
文摘The subcellular localization of human proteins is vital for understanding the structure of human cells.Proteins play a significant role within human cells,as many different groups of proteins are located in a specific location to perform a particular function.Understanding these functions will help in discoveringmany diseases and developing their treatments.The importance of imaging analysis techniques,specifically in proteomics research,is becoming more prevalent.Despite recent advances in deep learning techniques for analyzing microscopy images,classification models have faced critical challenges in achieving high performance.Most protein subcellular images have a significant class imbalance.We use oversampling and under sampling techniques in this research to overcome this issue.We have used a Convolutional Neural Network(CNN)model called GapNet-PL for the multi-label classification task on the Human Protein Atlas Classification(HPA)Dataset.Authors have found that the ParametricRectified LinearUnit(PreLU)activation function is better than the Scaled Exponential LinearUnit(SeLU)activation function in the GapNet-PL model in most classification metrics.The results showed that the GapNet-PL model with the PReLU activation function achieved an area under the ROC curve(AUC)equal to 0.896,an F1 score of 0.541,and a recall of 0.473.
