An architecture for Parzen-based multivariate probability density estimation
Djordje Stanković, Andjela Draganić, Nedjeljko Lekić, Cornel Ioana, Irena Orović · 2024
This paper proposes an architecture for multivariate probability density estimation using the Parzen window approach. As a non-parametric method, the Parzen approach provides reliable density estimates when sufficient data is available. It does not require prior information about the data, making it practical for real-world applications. The proposed solution eliminates the need for sorting operations, which are challenging to implement in hardware. Simulations conducted using PSpice software (OrCad version 22.1) demonstrate that the required processing time is less than 500 ns, underscoring the method's potential for fast and efficient probability density estimation.