I nt rusi on Det ect i on i n Vi rt ual Networks of Cl oud Comput i ng based on Machi neLearni ng Syst em

I nt rusi on det ect i on i s based on t hehypot hesi s t hat hacker behavi or i sdi f f erent f rom t he behavi ors of anaut hori zed user t hat can be measured. wecannot expect a cl ear and preci sedi st i nct i on between an at t ack by anat t acker and t he normal use ofresources by an aut hori zed user, but weshoul d expect some of t he two t ooverl ap wi t h each ot her. Theref ore, t oaddress t hi s i ssue, t here i s a need t oext end t radi t i onal securi t y sol ut i onssuch as f i rewal l s, and i nt rusi onprevent i on or det ect i on syst ems. I naddi t i on, i dent i f yi ng t he possi bl echaract eri st i cs of network t raf f i c i san i mport ant chal l enge f or accurat ei nt rusi on det ect i on. We are usi ng t hei nf i l t rat i on UNSW-NB۱۵ dat a set . wehave achi eved hi gher accuracy andaccept abl e resul t s i n cl oud comput i ngvi rt ual networks, where t he workprocess i s such t hat f i rst t he dat a seti s sent t o t he cl assi f i er Logi st i cRegressi on i s gi ven and t hen combi nedwi t h K-Nearest Nei ghbor cl assi f i er andf i nal l y i t s resul t s are i mproved wi t hAda-boost and i t s resul t s arepresent ed.