An Enhancement Method in Few-Shot Scenarios for Intrusion Detection in Smart Home Environments

Different devices in the smart home environment are subject to different levels of attack. Devices with lower attack frequencies confront difficulties collecting data, which restricts ability train intrusion detection models. Therefore, this paper presents a novel method called EM-FEDE (enhancement based on feature enhancement and data enhancement) generate adequate training for expanding few-shot datasets. Training models an expanded dataset can enhance performance. Firstly, adaptively extends features by analyzing historical records homes, achieving format alignment device data. Secondly, performs cleaning operations reduce noise redundancy uses random sampling mechanism ensure diversity obtained sampling. Finally, processed is used as input CWGAN, loss between generated real calculated using Wasserstein distance. Based loss, CWGAN adjusted. generator outputs effectively According experimental findings, accuracy J48, Random Forest, Bagging, PART, KStar, KNN, MLP, CNN has been enhanced 21.9%, 6.2%, 19.4%, 9.2%, 6.3%, 7%, 3.4%, 5.9%, respectively, when compared original dataset, along optimal generation sample ratio each algorithm. The findings demonstrate effectiveness approach completing sparse