Identification of geochemical anomalies via local RX anomaly detector

摘要: Abstract Multivariate geochemical anomalies are of great significance to the mineral exploration. The general method for multivariate is application a hybrid such as combining principal component analysis (PCA) and local singularity (LSA). However, unknown probability distribution data may could not meet condition PCA detection anomalies. In this study, RX anomaly detector based on double sliding windows was used detect Based idea nonlinear manifold can be approximated linearly, converted global problem into linear in multidimensional feature space data. from southwestern Fujian district (China) were carried out validate method. map showed that majority skarn Fe deposits situated areas with high value RX(x), demonstrating detected have close spatial relationship mineralization. comparing results deep autoencoder network LSA, suggest potential identify complex geological background.