The Million Song Dataset

By Thierry Bertin-Mahieux, D. Ellis, Brian Whitman, Paul Lamere. 2011

1,500 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Evaluation & Benchmarks, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Evaluation & Benchmarks

In Signal Processing

Citations in the corpus

Variational Autoencoders for Collaborative Filtering (2018, methodology): “Million Song Dataset (MSD): This data contains the user-song play counts released as part of the Million Song Dataset [3].”
On the Effectiveness of Linear Models for One-Class Collaborative Filtering (2016, methodology): “We evaluated all methods on four datasets: (a) M L 1M, the MovieLens 1M dataset, (b) K OBO , an e-book purchase dataset from Rakuten Kobo Inc. 4 , (c) L AST FM, a song play count dataset from LastFM (Celma 2008), and (d) M SD , the Million Song dataset (Bertin-Mahieux et al. 2011).”
NAS-Bench-101: Towards Reproducible Neural Architecture Search (2019, methodology): “We use 3 popular UCI (Lichman, 2013) datasets for regression: protein structure (Rana, 2013), slice localization (Graf et al., 2011) and year prediction (Bertin-Mahieux et al., 2011).”
Fast Training of Convolutional Networks through FFTs (2013, background): “While early benchmark datasets in machine learning contained thousands or tens of thousands of samples [7, 3, 10], current datasets are of the order of millions [6, 2].”
Neural Audio Synthesis of Musical Notes with WaveNet Autoencoders (2017, background): “Audio signals found in the wild contain multi-scale dependencies that prove particularly diﬃcult to model (Raﬀel, 2016; Bertin-Mahieux et al., 2011; King et al., 2008; Thickstun et al., 2016), leading many previous eﬀorts at data-driven audio synthesis to focus on more constrained domains such as…”
Divide and conquer kernel ridge regression: a distributed algorithm with minimax optimal rates (2013, methodology): “We use the Million Song Dataset [4], which consists of 463,715 training examples and a second set of 51,630 testing examples.”
MUSAN: A Music, Speech, and Noise Corpus (2015, background): “Other corpra, such as the Million Song database [3] circumvent intellectual property issues by providing only features.”
COALA: Co-Aligned Autoencoders for Learning Semantically Enriched Audio Representations (2020, methodology): “Similarly in MGC, the sam- pleCNN classifier, pre-trained on the Million Song Dataset (MSD) (Lee et al., 2018) produces much better results than our approach.”
Learning Binary Trees via Sparse Relaxation (2020, methodology): “Year (Bertin-Mahieux et al., 2011): Temporal regression task constructed from the Million Song Dataset; Microsoft (Qin & Liu, 2013): Regression approach to the MSLR-Web10k Query–URL relevance prediction for learning to rank; Yahoo (Chapelle & Chang, 2011): Regression approach to the C14 learning-to-”
Efficient Distributed SGD with Variance Reduction (2015, methodology): “We see that for MILLIONSONG, increasing the number of local workers initially decreases convergence time, but speed levels out for large numbers of workers, likely due to the smaller size of the local dataset fragments.”
Reinforcement Learning based Recommender Systems: A Survey (2021, methodology): “Two popular datasets used in RL-based RSs are MovieLens [110] and Million Song dataset [111].”
Automatic Tagging Using Deep Convolutional Neural Networks (2016, methodology): “In crowd-sourced music tag datasets [2,13], most of the tags are false (0) for most of the clips, which makes accuracy or mean square error inappropriate as a measure.”
A modern maximum-likelihood theory for high-dimensional logistic regression (2018, methodology): “Trust Case Control Consortium, WTCCC, [26]) • A subset of the million song dataset [14], called the YearPredictionMSD data (Source: UCI machine learning repository [41]) • The Physical Unclonable Functions (PUFs) dataset (Source: UCI machine learning repository [41]) The first dataset constitutes ge”
Learning-Based Methods for Comparing Sequences, with Applications to Audio-to-MIDI Alignment and Matching (2016, methodology): “1, the remainder of this thesis focuses on matching and aligning this collection to the Million Song Dataset (MSD) (Bertin-Mahieux et al., 2011).”
Discovering Discrete Latent Topics with Neural Variational Inference (2017, neutral): “edu/millionsong/musixmatch (Bertin-Mahieux et al., 2011) (3)http://qwone.”
Coresets for Nonparametric Estimation - the Case of DP-Means (2015, background): “For MSYP, the impact of construction time can be clearly seen.”