PyOD OCSVM with polynomial kernel and degree 3 does not generate similar output to OCSVM in sklearn

[JuanMarcosRamirez]

Dear developers,

I built an OCSVM model with a polynomial kernel and degree 3 using Sklearn. This model OCSVM generated an outlier pattern from an input dataset with a given contamination (OCSVM in Sklearn does not require the contamination parameter). We build the same OCSVM model using PyOD with the contamination from the outlier pattern obtained with the Sklearn model. The outlier pattern obtained with PyOD is entirely different from that obtained with Sklearn. Please revise the OCSVM code for polynomial kernels.

Regards,

Juan Marcos.