Transfer Learning for Music Classification and Regression Tasks (2017-03-27T00:00:00.000000Z)

TL;DR

This paper proposes to use a pre-trained convnet feature, a concatenated feature vector using the activations of feature maps of multiple layers in a trained convolutional network, and shows how it can serve as general-purpose music representation.

Abstract

In this paper, we present a transfer learning approach for music classification and regression tasks. We propose to use a pre-trained convnet feature, a concatenated feature vector using the activations of feature maps of multiple layers in a trained convolutional network. We show how this convnet feature can serve as general-purpose music representation. In the experiments, a convnet is trained for music tagging and then transferred to other music-related classification and regression tasks. The convnet feature outperforms the baseline MFCC feature in all the considered tasks and several previous approaches that are aggregating MFCCs as well as low- and high-level music features.

Authors

Kyunghyun Cho

23 papers

Keunwoo Choi

8 papers

György Fazekas

6 papers

TL;DR

Abstract

Authors

References63 items

Kapre: On-GPU Audio Preprocessing Layers for a Quick Implementation of Deep Neural Network Models with Keras

The Effects of Noisy Labels on Deep Convolutional Neural Networks for Music Tagging

The Effects of Noisy Labels on Deep Convolutional Neural Networks for Music Classification

On the Robustness of Deep Convolutional Neural Networks for Music Classification

Look, Listen and Learn

Multi-Level and Multi-Scale Feature Aggregation Using Pretrained Convolutional Neural Networks for Music Auto-Tagging

Designing efficient architectures for modeling temporal features with convolutional neural networks

Towards Multi-Purpose Spectral Rhythm Features: An Application to Dance Style, Meter and Tempo Estimation

SoundNet: Learning Sound Representations from Unlabeled Video

How to Use t-SNE Effectively

Convolutional recurrent neural networks for music classification

Scale and shift invariant time/frequency representation using auditory statistics: Application to rhythm description

Deep Convolutional Neural Networks and Data Augmentation for Environmental Sound Classification

The extended ballroom dataset

Explaining Deep Convolutional Neural Networks on Music Classification

Revisiting Priorities: Improving MIR Evaluation Practices

Towards Playlist Generation Algorithms Using RNNs Trained on Within-Track Transitions

Automatic Tagging Using Deep Convolutional Neural Networks

Theano: A Python framework for fast computation of mathematical expressions

Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification

Deep Image Features in Music Information Retrieval

Adam: A Method for Stochastic Optimization

Audio feature reduction and analysis for automatic music genre classification

A Dataset and Taxonomy for Urban Sound Research

Learning sparse dictionaries for music and speech classification

Very Deep Convolutional Networks for Large-Scale Image Recognition

Music genre classification based on local feature selection using a self-adaptive harmony search algorithm

Learning and Transferring Mid-level Image Representations Using Convolutional Neural Networks

Distributed Representations of Sentences and Documents

On-line continuous-time music mood regression with deep recurrent neural networks

Network In Network

Visualizing and Understanding Convolutional Networks

1000 songs for emotional analysis of music

Efficient Estimation of Word Representations in Vector Space

ImageNet classification with deep convolutional neural networks

Invariant Scattering Convolution Networks

Scikit-learn: Machine Learning in Python

A Survey on Transfer Learning

Self-adaptive harmony search algorithm for optimization

Enhanced polyphonic music genre classification using high level features

Vocal detection in music with support vector machines

A tutorial on support vector regression

Extreme learning machine: a new learning scheme of feedforward neural networks

Evaluating Rhythmic descriptors for Musical Genre Classification

Factors in automatic musical genre classification of audio signals

Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy

Least Squares Support Vector Machine Classifiers

An Introduction to the Psychology of Hearing

kapre: Keras audio preprocessors

Beat histogram features for rhythm-based musical genre classification using multiple novelty functions

AURALISATION OF DEEP CONVOLUTIONAL NEURAL NETWORKS: LISTENING TO LEARNED FEATURES

Content-Aware Collaborative Music Recommendation Using Pre-trained Neural Networks

Traffic Sign Recognition Using Extreme Learning Classifier with Deep Convolutional Features

Keras: Deep learning library for theano and tensorflow

Transfer Learning by Supervised Pre-training for Audio-based Music Classification

Transfer Learning In Mir: Sharing Learned Latent Representations For Music Audio Classification And Similarity

Learning Feature Representations with K-Means

The Million Song Dataset

Visualizing Data using t-SNE

Gtzan musicspeech

Proceedings of the Twenty-Second International Joint Conference on Artificial Intelligence WSABIE: Scaling Up To Large Vocabulary Image Annotation

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names