[2007.11154] Rethinking CNN Models for Audio Classificationopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

In this paper, we show that ImageNet-Pretrained standard deep CNN models can be used as strong baseline networks for audio classification. Even though there is a significant difference between audio Spectrogram and standard ImageNet image samples, transfer learning assumptions still hold firmly. To understand what enables the ImageNet pretrained models to learn useful audio representations, we systematically study how much of pretrained weights is useful for learning spectrograms. We show (1) that for a given standard model using pretrained weights is better than using randomly initialized weights (2) qualitative results of what the CNNs learn from the spectrograms by visualizing the gradients. Besides, we show that even though we use the pretrained model weights for initialization, there is variance in performance in various output runs of the same model. This variance in performance is due to the random initialization of linear classification layer and random mini-batch orderings in

1 mentions: @shinmura0
Keywords: cnn
Date: 2020/09/16 14:22

Referring Tweets

@shinmura0 Rethinking CNN Models for Audio Classification t.co/D05U5Otiwz ・ImageNetで学習したモデルを、環境音認識に転移学習させる研究 ・「どのレイヤーを凍結するか」など、広域的な実験を実施 ・ESC50でSOTA近くの精度を達成 (続く)

Related Entries

Read more [2001.10072] ABCTracker: an easy-to-use, cloud-based application for tracking multiple objectscontac...
0 users, 1 mentions 2020/01/31 02:20
Read more 【オンライン】Pythonデータ分析勉強会#16 - connpass
1 users, 1 mentions 2020/02/10 15:01
Read more [1812.04960] Object-centric Auto-encoders and Dummy Anomalies for Abnormal Event Detection in Videoc...
0 users, 1 mentions 2020/02/26 05:20
Read more [1904.07223] Joint Discriminative and Generative Learning for Person Re-identificationopen searchope...
0 users, 1 mentions 2020/06/23 03:51
Read more [1812.04606] Deep Anomaly Detection with Outlier Exposureopen searchopen navigation menucontact arXi...
0 users, 1 mentions 2020/09/19 13:00