Credit approval classification through a WASD neuronet : доклад, тезисы доклада

Описание

Тип публикации: доклад, тезисы доклада, статья из сборника материалов конференций

Конференция: Hybrid Methods of Modeling and Optimization in Complex Systems (HMMOCS-III 2024); Krasnoyarsk; Krasnoyarsk

Год издания: 2025

Идентификатор DOI: 10.1051/itmconf/20257205006

Аннотация: Because the banking business is growing, more people are seeking for bank loans, although banks can only lend to a certain number of applicants because they have a limited amount of assets to lend to. Hence, in order to save a lot of bank resources, the industry of banking is particularly concerned in developing ways to lower the rПоказать полностьюisk element involved in selecting the safe applicant. These days, selecting the safe applicant requires a lot less work thanks to machine intelligence. In light of this, a new weights and structure determination (WASD) neuronet has been developed to address the two issues of credit approval mentioned above, as well as to manage its particular features. We improve the learning process of the WASD algorithm with a novel activation function for optimal adaptation to the credit approval model, motivated by the finding that WASD neuronets perform better than traditional back-propagation neuronets in terms of slow training speed and trapping in a local minima. An experimental study with an insurance company dataset demonstrates superior performance and adaptability to issues.

Ссылки на полный текст

Издание

Журнал: ITM Web of Conferences

Номера страниц: 5006

Место издания: Krasnoyarsk

Персоны

  • Mourtas Spyridon D. (Siberian Federal University)

Вхождение в базы данных