Asosiy kontentga oʻtish
Maqola

Improving machine learning algorithms using methodological stochastic differential equations

Isha SalujaSymbiosis Law School Pune (SLS-P), Symbiosis Centre for Advanced Legal Studies and Research (SCALSAR), Symbiosis International (Deemed University), Pune, Maharashtra, 411014, IndiaSrijita BhattacharjeeDepartment of Engineering and Technology (CSE), Bharati Vidyapeeth (Deemed to be University), Kharghar, Navi Mumbai, Maharashtra, 410210, IndiaVimal BibhuDepartment of Computer Science & Engineering, Noida International University, Greater Noida, Uttar Pradesh, 203201, IndiaS. Mohan Mahalakshmi NaiduDepartment of Electronics & Telecommunication, International Institute of Information Technology (I²IT), Pune, Maharashtra, 411057, IndiaVanita S. BuradkarSamariddin MakhmudovDepartment of Economics, Mamun University, Khiva, 220900, Uzbekistan
2026en
ABI

Annotatsiya

The study examines the application of scientific stochastic differential equations (SDEs) to machine learning techniques to make them more robust and predictive. This is an approach that enhances generalization in evolutionary and complex environments because it characterizes the skepticism and clatter of data that are constructed in with SDEs. We come up with new SDE-based systems that can modify the fast rate at which they learn as well as the frequency with which they revise parameters. This ensures that converging is more stable. Big changes in measures of speed are seen in experimental results on test datasets when compared to traditional optimization methods. The proposed approach provides a solid mathematical means of incorporating stochastic dynamics that provide more data of how algorithms behave in the case of doubt.

Hali tarjima qilinmagan

Identifikatorlar

Iqtiboslar va manbalar

0 ta iqtibos0 ta foydalanilgan manba