With the increase of semiconductor integration density,in order to cope with the increase of wafer defect complexity and types,especially the low recognition accuracy of overlapping mixed defects and unknown wafer def...With the increase of semiconductor integration density,in order to cope with the increase of wafer defect complexity and types,especially the low recognition accuracy of overlapping mixed defects and unknown wafer defects,this study proposes a lightweight model for wafer defect detection called LightWMNet.First,using a hierarchical attention Encoder-Decoder architecture,the features of wafer defect pattern(WDP)are channel recalibrated to generate high-resolution fine-grained features and low-resolution coarse-grained features.Secondly,the backbone network incorporates two novel attention modules—feedforward spatial attention(FFSa)and feedforward channel attention(FFCa)—to amplify responses in critical defect regions and suppress noise from stochastic discrete pixels.These mechanisms synergistically enhance feature discriminability without introducing significant parametric overhead.Finally,the Dice loss function and the cross entropy loss function are combined to jointly evaluate the segmentation and classification accuracy of the model.Experimental results on the public mixed wafer defect dataset MixedWM38 show that the pixel accuracy(PA),intersection over union(IoU)and Dice coefficient of the proposed network reach 98.26%,94.83%and 97.22%,respectively.Without significantly increasing the computational complexity and size of the model,compared with the existing state-of-the-art(SOTA)model,the classification accuracy of lightWMNet in single defect,three mixed defects and four mixed defects is improved by 0.5%,0.25%and 0.89%respectively.Furthermore,we used transfer learning for the first time to evaluate the model's generalisation ability for unseen defect categories.The results showed that LightWMNet still has a certain recognition ability even in untrained wafer defects.展开更多
A novel siphon-based divide-and-conquer(SbDaC)policy is presented in this paper for the synthesis of Petri net(PN)based liveness-enforcing supervisors(LES)for flexible manufacturing systems(FMS)prone to deadlocks or l...A novel siphon-based divide-and-conquer(SbDaC)policy is presented in this paper for the synthesis of Petri net(PN)based liveness-enforcing supervisors(LES)for flexible manufacturing systems(FMS)prone to deadlocks or livelocks.The proposed method takes an uncontrolled and bounded PN model(UPNM)of the FMS.Firstly,the reduced PNM(RPNM)is obtained from the UPNM by using PN reduction rules to reduce the computation burden.Then,the set of strict minimal siphons(SMSs)of the RPNM is computed.Next,the complementary set of SMSs is computed from the set of SMSs.By the union of these two sets,the superset of SMSs is computed.Finally,the set of subnets of the RPNM is obtained by applying the PN reduction rules to the superset of SMSs.All these subnets suffer from deadlocks.These subnets are then ordered from the smallest one to the largest one based on a criterion.To enforce liveness on these subnets,a set of control places(CPs)is computed starting from the smallest subnet to the largest one.Once all subnets are live,this process provides the LES,consisting of a set of CPs to be used for the UPNM.The live controlled PN model(CPNM)is constructed by merging the LES with the UPNM.The SbDaC policy is applicable to all classes of PNs related to FMS prone to deadlocks or livelocks.Several FMS examples are considered from the literature to highlight the applicability of the SbDaC policy.In particular,three examples are utilized to emphasize the importance,applicability and effectiveness of the SbDaC policy to realistic FMS with very large state spaces.展开更多
Monge–Ampere equations(MAEs)are fully nonlinear second-order partial differential equations(PDEs),which are closely related to various fields including optimal transport(OT)theory,geometrical optics and affine geomet...Monge–Ampere equations(MAEs)are fully nonlinear second-order partial differential equations(PDEs),which are closely related to various fields including optimal transport(OT)theory,geometrical optics and affine geometry.Despite their significance,MAEs are extremely challenging to solve.Although some classical numerical approaches can solve MAEs,their computational efficiency deteriorates significantly on fine grids,with convergence often heavily dependent on the quality of initial estimate.Research on deep learning methods for solving MAEs is still in its early stages,which predominantly addresses simple formulations with basic Dirichlet boundary conditions.Here,we propose a deep learning method based on physicsdriven deep neural networks,enabling the solution of both simple and generalised MAEs with transport boundary conditions.In this method,we deal with two first-order sub-equations separated from MAE instead of solving the single MAE directly,which facilitates the imposition of transport boundary conditions and simplifies the training of neural networks.Moreover,we constrain the convexity of solution using the Lagrange multiplier method and maintain the optimisation process differentiable with bilinear interpolation.We provide three progressively complex examples ranging from a simple MAE with an analytical solution to a highly nonlinear variant arising in phase retrieval to validate the effectiveness of our method.For comparison,we benchmark against state-of-the-art deep learning approaches that have been systematically adapted to accommodate the specific requirements of each example.展开更多
Neutrophil extracellular traps(NETs)have emerged as key mediators of cardiovascular diseases(CVDs),linking innate immune activation to vascular injury,thrombosis,and maladaptive remodeling.This review synthesizes rece...Neutrophil extracellular traps(NETs)have emerged as key mediators of cardiovascular diseases(CVDs),linking innate immune activation to vascular injury,thrombosis,and maladaptive remodeling.This review synthesizes recent insights into the molecular and cellular pathways driving NET formation,including post-translational modifications,metabolic reprogramming,inflammasome signaling,and autophagy.It highlights the role of NETs in atherosclerosis,thrombosis,myocardial ischemia-reperfusion injury,and hypertension,emphasizing common control points such as peptidylarginine deiminase 4(PAD4)-dependent histone citrullination and nicotinamide adenine dinucleotide phosphate oxidases 2(NOX2)-mediated oxidative stress.Mechanistic interpretation of circulating biomarkers,includingmyeloperoxidase(MPO)-DNA complexes,citrullinated histoneH3,and cell-free DNA,provides a translational bridge between NET biology and patient stratification.Therapeutic strategies targeting NETs are examined through three main approaches:inhibition of NET initiation,enhancement of chromatin clearance,and neutralization of toxic extracellular components,with attention to both established and emerging interventions.In contrast to previous reviews,this study highlights the novelty of a mechano-therapeutic framework by providing a mechanistic roadmap linking NET formation pathways to therapeutic targeting in cardiovascular disease.Moving forward,integrating mechanistic information with biomarker discovery,precision profiling,and targeted therapies offers innovative strategies to reduce vascular inflammation and improve outcomes in cardiovascular disease.展开更多
Train Mass Rapid Transit(MRT)was put into service in 1987,and has since been augmented by and linked to the Light Rapid Transit.Combined,you can often get you within walking distance of most destinations.The maps on t...Train Mass Rapid Transit(MRT)was put into service in 1987,and has since been augmented by and linked to the Light Rapid Transit.Combined,you can often get you within walking distance of most destinations.The maps on the metro system are easy to read,complete with English version.You can easily purchase an EZ-Link card or a NETS Flashpay Card(stored value cards)at all MRT stations and bus interchange.展开更多
Deep Learning-based systems for Finger vein recognition have gained rising attention in recent years due to improved efficiency and enhanced security.The performance of existing CNN-based methods is limited by the pun...Deep Learning-based systems for Finger vein recognition have gained rising attention in recent years due to improved efficiency and enhanced security.The performance of existing CNN-based methods is limited by the puny generalization of learned features and deficiency of the finger vein image training data.Considering the concerns of existing methods,in this work,a simplified deep transfer learning-based framework for finger-vein recognition is developed using an EfficientNet model of deep learning with a self-attention mechanism.Data augmentation using various geometrical methods is employed to address the problem of training data shortage required for a deep learning model.The proposed model is tested using K-fold cross-validation on three publicly available datasets:HKPU,FVUSM,and SDUMLA.Also,the developed network is compared with other modern deep nets to check its effectiveness.In addition,a comparison of the proposed method with other existing Finger vein recognition(FVR)methods is also done.The experimental results exhibited superior recognition accuracy of the proposed method compared to other existing methods.In addition,the developed method proves to be more effective and less sophisticated at extracting robust features.The proposed EffAttenNet achieves an accuracy of 98.14%on HKPU,99.03%on FVUSM,and 99.50%on SDUMLA databases.展开更多
基金supported by the National Natural Science Foundation of China under Grant 61573183.
文摘With the increase of semiconductor integration density,in order to cope with the increase of wafer defect complexity and types,especially the low recognition accuracy of overlapping mixed defects and unknown wafer defects,this study proposes a lightweight model for wafer defect detection called LightWMNet.First,using a hierarchical attention Encoder-Decoder architecture,the features of wafer defect pattern(WDP)are channel recalibrated to generate high-resolution fine-grained features and low-resolution coarse-grained features.Secondly,the backbone network incorporates two novel attention modules—feedforward spatial attention(FFSa)and feedforward channel attention(FFCa)—to amplify responses in critical defect regions and suppress noise from stochastic discrete pixels.These mechanisms synergistically enhance feature discriminability without introducing significant parametric overhead.Finally,the Dice loss function and the cross entropy loss function are combined to jointly evaluate the segmentation and classification accuracy of the model.Experimental results on the public mixed wafer defect dataset MixedWM38 show that the pixel accuracy(PA),intersection over union(IoU)and Dice coefficient of the proposed network reach 98.26%,94.83%and 97.22%,respectively.Without significantly increasing the computational complexity and size of the model,compared with the existing state-of-the-art(SOTA)model,the classification accuracy of lightWMNet in single defect,three mixed defects and four mixed defects is improved by 0.5%,0.25%and 0.89%respectively.Furthermore,we used transfer learning for the first time to evaluate the model's generalisation ability for unseen defect categories.The results showed that LightWMNet still has a certain recognition ability even in untrained wafer defects.
基金The authors extend their appreciation to King Saud University,Saudi Arabia for funding this work through the Ongoing Research Funding Program(ORF-2025-704),King Saud University,Riyadh,Saudi Arabia.
文摘A novel siphon-based divide-and-conquer(SbDaC)policy is presented in this paper for the synthesis of Petri net(PN)based liveness-enforcing supervisors(LES)for flexible manufacturing systems(FMS)prone to deadlocks or livelocks.The proposed method takes an uncontrolled and bounded PN model(UPNM)of the FMS.Firstly,the reduced PNM(RPNM)is obtained from the UPNM by using PN reduction rules to reduce the computation burden.Then,the set of strict minimal siphons(SMSs)of the RPNM is computed.Next,the complementary set of SMSs is computed from the set of SMSs.By the union of these two sets,the superset of SMSs is computed.Finally,the set of subnets of the RPNM is obtained by applying the PN reduction rules to the superset of SMSs.All these subnets suffer from deadlocks.These subnets are then ordered from the smallest one to the largest one based on a criterion.To enforce liveness on these subnets,a set of control places(CPs)is computed starting from the smallest subnet to the largest one.Once all subnets are live,this process provides the LES,consisting of a set of CPs to be used for the UPNM.The live controlled PN model(CPNM)is constructed by merging the LES with the UPNM.The SbDaC policy is applicable to all classes of PNs related to FMS prone to deadlocks or livelocks.Several FMS examples are considered from the literature to highlight the applicability of the SbDaC policy.In particular,three examples are utilized to emphasize the importance,applicability and effectiveness of the SbDaC policy to realistic FMS with very large state spaces.
基金supported by CAAI-Huawei MindSpore Open Fund(CAAIXSJLJJ-2022-010A).
文摘Monge–Ampere equations(MAEs)are fully nonlinear second-order partial differential equations(PDEs),which are closely related to various fields including optimal transport(OT)theory,geometrical optics and affine geometry.Despite their significance,MAEs are extremely challenging to solve.Although some classical numerical approaches can solve MAEs,their computational efficiency deteriorates significantly on fine grids,with convergence often heavily dependent on the quality of initial estimate.Research on deep learning methods for solving MAEs is still in its early stages,which predominantly addresses simple formulations with basic Dirichlet boundary conditions.Here,we propose a deep learning method based on physicsdriven deep neural networks,enabling the solution of both simple and generalised MAEs with transport boundary conditions.In this method,we deal with two first-order sub-equations separated from MAE instead of solving the single MAE directly,which facilitates the imposition of transport boundary conditions and simplifies the training of neural networks.Moreover,we constrain the convexity of solution using the Lagrange multiplier method and maintain the optimisation process differentiable with bilinear interpolation.We provide three progressively complex examples ranging from a simple MAE with an analytical solution to a highly nonlinear variant arising in phase retrieval to validate the effectiveness of our method.For comparison,we benchmark against state-of-the-art deep learning approaches that have been systematically adapted to accommodate the specific requirements of each example.
文摘Neutrophil extracellular traps(NETs)have emerged as key mediators of cardiovascular diseases(CVDs),linking innate immune activation to vascular injury,thrombosis,and maladaptive remodeling.This review synthesizes recent insights into the molecular and cellular pathways driving NET formation,including post-translational modifications,metabolic reprogramming,inflammasome signaling,and autophagy.It highlights the role of NETs in atherosclerosis,thrombosis,myocardial ischemia-reperfusion injury,and hypertension,emphasizing common control points such as peptidylarginine deiminase 4(PAD4)-dependent histone citrullination and nicotinamide adenine dinucleotide phosphate oxidases 2(NOX2)-mediated oxidative stress.Mechanistic interpretation of circulating biomarkers,includingmyeloperoxidase(MPO)-DNA complexes,citrullinated histoneH3,and cell-free DNA,provides a translational bridge between NET biology and patient stratification.Therapeutic strategies targeting NETs are examined through three main approaches:inhibition of NET initiation,enhancement of chromatin clearance,and neutralization of toxic extracellular components,with attention to both established and emerging interventions.In contrast to previous reviews,this study highlights the novelty of a mechano-therapeutic framework by providing a mechanistic roadmap linking NET formation pathways to therapeutic targeting in cardiovascular disease.Moving forward,integrating mechanistic information with biomarker discovery,precision profiling,and targeted therapies offers innovative strategies to reduce vascular inflammation and improve outcomes in cardiovascular disease.
文摘Train Mass Rapid Transit(MRT)was put into service in 1987,and has since been augmented by and linked to the Light Rapid Transit.Combined,you can often get you within walking distance of most destinations.The maps on the metro system are easy to read,complete with English version.You can easily purchase an EZ-Link card or a NETS Flashpay Card(stored value cards)at all MRT stations and bus interchange.
文摘Deep Learning-based systems for Finger vein recognition have gained rising attention in recent years due to improved efficiency and enhanced security.The performance of existing CNN-based methods is limited by the puny generalization of learned features and deficiency of the finger vein image training data.Considering the concerns of existing methods,in this work,a simplified deep transfer learning-based framework for finger-vein recognition is developed using an EfficientNet model of deep learning with a self-attention mechanism.Data augmentation using various geometrical methods is employed to address the problem of training data shortage required for a deep learning model.The proposed model is tested using K-fold cross-validation on three publicly available datasets:HKPU,FVUSM,and SDUMLA.Also,the developed network is compared with other modern deep nets to check its effectiveness.In addition,a comparison of the proposed method with other existing Finger vein recognition(FVR)methods is also done.The experimental results exhibited superior recognition accuracy of the proposed method compared to other existing methods.In addition,the developed method proves to be more effective and less sophisticated at extracting robust features.The proposed EffAttenNet achieves an accuracy of 98.14%on HKPU,99.03%on FVUSM,and 99.50%on SDUMLA databases.
