Efficient Nonlinear RX Anomaly Detectors

Current anomaly detection (AD) algorithms are typically challenged by either accuracy or efficiency. More accurate nonlinear detectors slow and not scalable. In this letter, we propose two families of techniques to improve the efficiency standard kernel Reed-Xiaoli (KRX) method for AD approximating function with data-independent random Fourier features data-dependent basis Nyström approach. We compare all methods both real multi- hyperspectral images. show that proposed efficient have a lower computational cost, they perform similar (or outperform) KRX algorithm thanks their implicit regularization effect. Last but least, approach has an improved power detection.