Thumbnail Image
Publication

Optimizing Neural Efficiency: Implementing Binarized Neural Networks for Binary Classification

Journal Title
Keywords
Readers/Advisors
Wang, Xiaoliang
Journal Title
Term and Year
Publication Date
2025
Book Title
Publication Volume
Publication Issue
Publication Begin
Publication End
Number of pages
Collections
Quest
Show all metadata
Files
Thumbnail Image
Quest2025_037.pdf
Adobe PDF, 442.9 KB
Research Projects
Organizational Units
Journal Issue
URI
http://hdl.handle.net/20.500.12648/16594
Abstract
To improve training stability and performance, this study introduces an improved Binary Neural Network (BNN) that uses a tan-based gradient approximation for binarization. The model's effectiveness in classifying tasks is demonstrated by comparing it to an implementation based on NumPy. Results from experiments show increased accuracy and convergence, which makes it more useful in settings with limited resources.
Citation
DOI
Description
Accessibility Statement
Embedded videos