.png){kind=icon}
.png){kind=icon}
Return
One-pass online learning from data streams with unpredictable feature evolution
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2025.112003.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A kernel-based online learning method is proposed to handle more realistic feature evolution data streams. • A support vector selection strategy based on budget and projection is introduced to ensure the model’s computational efficiency. • Theoretical analysis is provided to show the sublinear regret of the proposed method.
Keywords:
kernel-based online learning
feature evolution data streams
support vector selection
budget and projection
sublinear regret