Few-shot intent detection is a practical challenge task,because new intents are frequently emerging and collecting large-scale data for them could be costly.Meta-learning,a promising technique for leveraging data from...Few-shot intent detection is a practical challenge task,because new intents are frequently emerging and collecting large-scale data for them could be costly.Meta-learning,a promising technique for leveraging data from previous tasks to enable efficient learning of new tasks,has been a popular way to tackle this problem.However,the existing meta-learning models have been evidenced to be overfitting when the meta-training tasks are insufficient.To overcome this challenge,we present a novel self-supervised task augmentation with meta-learning framework,namely STAM.Firstly,we introduce the task augmentation,which explores two different strategies and combines them to extend meta-training tasks.Secondly,we devise two auxiliary losses for integrating self-supervised learning into meta-learning to learn more generalizable and transferable features.Experimental results show that STAM can achieve consistent and considerable performance improvement to existing state-of-the-art methods on four datasets.展开更多
For the Android app market,malware uses code encryption techniques for block detection.For Android applications,scholars have proposed a method to determine whether Android applications are malicious software by analy...For the Android app market,malware uses code encryption techniques for block detection.For Android applications,scholars have proposed a method to determine whether Android applications are malicious software by analyzing the behavioral characteristics of software operation.The most traditional method is static detection,which is characterized by fast detection speed and less resource occupation.However,Android software cannot be detected by static methods after using encryption technology.The APK package of the application isfirst decom-piled to detect and extract key features,behavioral patterns,and invocation infor-mation using Frida and Camille.Subsequently,the long short-term memory net-work(LSTM)is employed to analyze software intent for determining the presence of malware.The experimental results demonstrate that the static method achieves an accuracy of approximately 80%,whereas the dynamic method achieves an accuracy of 91%.Through the utilization of software intention analysis and per-mission usage checks in combination,the accuracy rate can be further enhanced to 94%.Upon comparison of the different algorithms utilized in each detection method,it is concluded that both the KNN and random forest algorithms exhibit higher accuracy in the application of such detection methods.展开更多
基金the National Natural Science Foundation of China under Grant Nos.61936012 and 61976114。
文摘Few-shot intent detection is a practical challenge task,because new intents are frequently emerging and collecting large-scale data for them could be costly.Meta-learning,a promising technique for leveraging data from previous tasks to enable efficient learning of new tasks,has been a popular way to tackle this problem.However,the existing meta-learning models have been evidenced to be overfitting when the meta-training tasks are insufficient.To overcome this challenge,we present a novel self-supervised task augmentation with meta-learning framework,namely STAM.Firstly,we introduce the task augmentation,which explores two different strategies and combines them to extend meta-training tasks.Secondly,we devise two auxiliary losses for integrating self-supervised learning into meta-learning to learn more generalizable and transferable features.Experimental results show that STAM can achieve consistent and considerable performance improvement to existing state-of-the-art methods on four datasets.
文摘For the Android app market,malware uses code encryption techniques for block detection.For Android applications,scholars have proposed a method to determine whether Android applications are malicious software by analyzing the behavioral characteristics of software operation.The most traditional method is static detection,which is characterized by fast detection speed and less resource occupation.However,Android software cannot be detected by static methods after using encryption technology.The APK package of the application isfirst decom-piled to detect and extract key features,behavioral patterns,and invocation infor-mation using Frida and Camille.Subsequently,the long short-term memory net-work(LSTM)is employed to analyze software intent for determining the presence of malware.The experimental results demonstrate that the static method achieves an accuracy of approximately 80%,whereas the dynamic method achieves an accuracy of 91%.Through the utilization of software intention analysis and per-mission usage checks in combination,the accuracy rate can be further enhanced to 94%.Upon comparison of the different algorithms utilized in each detection method,it is concluded that both the KNN and random forest algorithms exhibit higher accuracy in the application of such detection methods.
