This　paper　proposes　an　autoencoder　parameter　extraction　technique　for　parametric　modeling　of　electromagnetic　(EM)　behavior　of　the　third-order　waveguide　filter.　In　this　technique,　in　order　to　extract　the　obtained　feature　parameters,　the　neural　networks　are　trained　to　autonomously　optimize　the　weights　of　the　encoder　and　decoder.　The　proposed　technique　can　extract　the　feature　parameters　more　smoothly　and　continuously　compared　with　the　existing　vector　fitting　technique.　　©　2024　IEEE.