← Research

MACHINE LEARNING · 2023

In-Vehicle Noise Classification

Few-shot and continual-learning methods for classifying in-vehicle acoustic events.

Yonsei University 2023
In-Vehicle Noise Classification preview

Problem

In-vehicle acoustic classifiers must recognize diverse noise sources while adapting to limited data and new classes.

Work

I worked on few-shot and continual-learning approaches for in-vehicle noise classification, including prototype-based representation learning and replay strategies.

Outcomes

The work led to an IEEE Access paper on a disentangled prototypical convolutional network and a KDMS paper on dynamic prototype-guided memory replay.