Adversarial training has been widely considered the most effective defense against adversarial attacks.However,recent studies have demonstrated that a large discrepancy exists in the class-wise robustness of adversari...Adversarial training has been widely considered the most effective defense against adversarial attacks.However,recent studies have demonstrated that a large discrepancy exists in the class-wise robustness of adversarial training,leading to two potential issues:firstly,the overall robustness of a model is compromised due to the weakest class;and secondly,ethical concerns arising from unequal protection and biases,where certain societal demographic groups receive less robustness in defense mechanisms.Despite these issues,solutions to address the discrepancy remain largely underexplored.In this paper,we advance beyond existing methods that focus on class-level solutions.Our investigation reveals that hard examples,identified by higher cross-entropy values,can provide more fine-grained information about the discrepancy.Furthermore,we find that enhancing the diversity of hard examples can effectively reduce the robustness gap between classes.Motivated by these observations,we propose Fair Adversarial Training(FairAT)to mitigate the discrepancy of class-wise robustness.Extensive experiments on various benchmark datasets and adversarial attacks demonstrate that FairAT outperforms state-of-the-art methods in terms of both overall robustness and fairness.For a WRN-28-10 model trained on CIFAR10,FairAT improves the average and worst-class robustness by 2.13%and 4.50%,respectively.展开更多
Research on the wave field evolution law is highly significant to the fields of offshore engineering and marine resource development.Numerical simulations have been conducted for high-precision wave field evolution,th...Research on the wave field evolution law is highly significant to the fields of offshore engineering and marine resource development.Numerical simulations have been conducted for high-precision wave field evolution,thus providing short-term wave field prediction.However,its evolution occurs over a long period of time,and its accuracy is difficult to improve.In recent years,the use of machine learning methods to study the evolution of wave field has received increasing attention from researchers.This paper proposes a wave field evolution method based on deep convolutional neural networks.This method can effectively correlate the spa-tiotemporal characteristics of wave data via convolution operation and directly obtain the offshore forecast results of the Bohai Sea and the Yellow Sea.The attention mechanism,multi-scale path design,and hard example mining training strategy are introduced to suppress the interference caused by Weibull distributed wave field data and improve the accuracy of the proposed wave field evolu-tion.The 72-and 480-h evolution experiment results in the Bohai Sea and the Yellow Sea show that the proposed method in this pa-per has excellent forecast accuracy and timeliness.展开更多
A robust TV logo detection method based on the modified single shot multibox detector (SSD) is presented. Unlike most other existing methods which can only detect the TV logo from video frames, the proposed method can...A robust TV logo detection method based on the modified single shot multibox detector (SSD) is presented. Unlike most other existing methods which can only detect the TV logo from video frames, the proposed method can also detect the TV logo from photo pictures taken by smartphones or other smart terminals. Firstly, using a simple and effective way of collecting and labelling TV logo, a large-scale TV logo dataset used to train the detection model is built. Then, parameters and loss function of SSD are modified to make it more suitable for the task of TV logo detection. Moreover, a soft-NMS algorithm is introduced to remove the redundant overlapping boxes and obtain the final output box. And also an approach for hard example mining is designed to improve the detection accuracy. Finally, extensive comparison experiments are carried out which take into consideration different image resolutions, logo positions and environmental factors existing in real-world applications. Experimental results demonstrate that the proposed method achieve superior performances in robustness compared to other state-of-the-art methods.展开更多
基金supported by the National Natural Science Foundation of China(Grant Nos.U20B2049,U21B2018 and 62302344).
文摘Adversarial training has been widely considered the most effective defense against adversarial attacks.However,recent studies have demonstrated that a large discrepancy exists in the class-wise robustness of adversarial training,leading to two potential issues:firstly,the overall robustness of a model is compromised due to the weakest class;and secondly,ethical concerns arising from unequal protection and biases,where certain societal demographic groups receive less robustness in defense mechanisms.Despite these issues,solutions to address the discrepancy remain largely underexplored.In this paper,we advance beyond existing methods that focus on class-level solutions.Our investigation reveals that hard examples,identified by higher cross-entropy values,can provide more fine-grained information about the discrepancy.Furthermore,we find that enhancing the diversity of hard examples can effectively reduce the robustness gap between classes.Motivated by these observations,we propose Fair Adversarial Training(FairAT)to mitigate the discrepancy of class-wise robustness.Extensive experiments on various benchmark datasets and adversarial attacks demonstrate that FairAT outperforms state-of-the-art methods in terms of both overall robustness and fairness.For a WRN-28-10 model trained on CIFAR10,FairAT improves the average and worst-class robustness by 2.13%and 4.50%,respectively.
基金supported by the National Key Research and Development Project(No.2018YFC1407001).
文摘Research on the wave field evolution law is highly significant to the fields of offshore engineering and marine resource development.Numerical simulations have been conducted for high-precision wave field evolution,thus providing short-term wave field prediction.However,its evolution occurs over a long period of time,and its accuracy is difficult to improve.In recent years,the use of machine learning methods to study the evolution of wave field has received increasing attention from researchers.This paper proposes a wave field evolution method based on deep convolutional neural networks.This method can effectively correlate the spa-tiotemporal characteristics of wave data via convolution operation and directly obtain the offshore forecast results of the Bohai Sea and the Yellow Sea.The attention mechanism,multi-scale path design,and hard example mining training strategy are introduced to suppress the interference caused by Weibull distributed wave field data and improve the accuracy of the proposed wave field evolu-tion.The 72-and 480-h evolution experiment results in the Bohai Sea and the Yellow Sea show that the proposed method in this pa-per has excellent forecast accuracy and timeliness.
基金Supported by the National Natural Science Foundationof China(No.61702466)“Double Tops” Discipline Construction Project
文摘A robust TV logo detection method based on the modified single shot multibox detector (SSD) is presented. Unlike most other existing methods which can only detect the TV logo from video frames, the proposed method can also detect the TV logo from photo pictures taken by smartphones or other smart terminals. Firstly, using a simple and effective way of collecting and labelling TV logo, a large-scale TV logo dataset used to train the detection model is built. Then, parameters and loss function of SSD are modified to make it more suitable for the task of TV logo detection. Moreover, a soft-NMS algorithm is introduced to remove the redundant overlapping boxes and obtain the final output box. And also an approach for hard example mining is designed to improve the detection accuracy. Finally, extensive comparison experiments are carried out which take into consideration different image resolutions, logo positions and environmental factors existing in real-world applications. Experimental results demonstrate that the proposed method achieve superior performances in robustness compared to other state-of-the-art methods.
