Academic Project: Lightweight CNN for Acoustic Scene Classification
|Machine Learning Engineer / Researcher
Chennai, Tamil Nadu, India
Summary
Led the development of a lightweight Convolutional Neural Network (CNN) for acoustic scene classification, optimizing model performance and efficiency.
Highlights
Developed a low-complexity CNN utilizing MFCC features to accurately classify over 10,000 audio samples across diverse acoustic environments.
Achieved a 40% reduction in model complexity and a 30% improvement in accuracy compared to baseline models through innovative architectural design.
Implemented robust data preprocessing pipelines and augmented training data, significantly enhancing model generalization and performance.