Jixiang Li(Xiaomi), Chuming Liang(Xiaomi), Bo Zhang(Xiaomi), Zhao Wang(Xiaomi), Fei Xiang(Xiaomi) and Xiangxiang Chu(Xiaomi)
Convolutional neural networks are widely adopted in Acoustic Scene Classification (ASC) tasks, but they generally carry a heavy computational burden. In this work, we propose a high-performance yet lightweight baseline network inspired by MobileNetV2, which replaces square convolutional kernels with unidirectional ones to extract features alternately in temporal and frequency dimensions. Furthermore, we explore a dynamic architecture space built on the basis of the proposed baseline with the recent Neural Architecture Search (NAS) paradigm, which first train a supernet that incorporates all candidate architectures and then apply a well-known evolutionary algorithm NSGA-II to discover more efficient networks with higher accuracy and lower computational cost from the supernet. Experimental results demonstrate that our searched network is competent in ASC tasks, which achieves 90.3% F1-score on the DCASE2018 task 5 evaluation set, marking a new state-of-the-art performance while saving 25% of FLOPs compared to our baseline network.