Learning Spectral Clustering, With Application To Speech Separation

By Francis R. Bach, Michael I. Jordan. Journal of machine learning research, 2006

304 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/1879378a9af4b3125e55d98337000cd862e2b75b

How researchers see this paper

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

Citations in the corpus

Fast approximate spectral clustering (2009, methodology): “Although spectral clustering algorithms that work directly with multiway partitions exist [4, 37], in the current paper we will focus on the classical recursive Ncut algorithm.”
Maximin affinity learning of image segmentation (2009, background): “The work of Bach and Jordan [2] is the closest to our work.”
Speaker-Independent Speech Separation With Deep Attractor Network (2017, methodology): “stimated from the observation and used to separate the target signal from other sources in the mixture. Methods such as computational auditory scene analysis (CASA) [11], [12] and spectral clustering [13] fall into this category [14]. Factorization models, such as non-negative matrix factorization (”
DIFFRAC: a discriminative and flexible framework for clustering (2007, background): “d(B, B) defines a distance over the set of partitions [9] which is always between 0 and(k + k − 2).”
Supervised Speech Separation Based on Deep Learning: An Overview (2017, unknown): “Speaker-independent separation can be treated as unsupervised clustering where T-F units are clustered into distinct classes dominated by individual speakers [6], [79].”
Alternative Objective Functions for Deep Clustering (2018, methodology): “…approaches have demonstrated overwhelming advantages over previous approaches including graphical modeling approaches [6], spectral clustering approaches [7], and CASA methods [8] In this study, we introduce several alternative deep clustering objectives that aim to improve clustering performance.”
Analyzing communities and their evolutions in dynamic social networks (2009, background): “Then the error rate is represented by the norm ‖Z Z T − GGT ‖F , which measures the distance between the community structure represented by Z and that represented by G (see Bach and Jordan [2006] for a detailed discussion).”
Deep clustering: Discriminative embeddings for segmentation and separation (2015, methodology): “Conversely this very complexity makes optimization of the front end processing a formidable challenge [16].”
Recursive speech separation for unknown number of speakers (2019, background): “Previously, various approaches including spectral clustering [4] computational auditory scene analysis (CASA) [5], non-negative matrix factorization (NMF) [6–9] were proposed to tackle this problem, yet showed limited success.”
Multi-Channel Deep Clustering: Discriminative Spectral and Spatial Embeddings for Speaker-Independent Speech Separation (2018, methodology): “…was the ﬁrst modern approach to perform speaker-independent separation, and demonstrated superior performance over previous attempts at speech separation, including graphical modeling approaches [3], spectral clustering approaches [4], and computational auditory scene analysis based methods [5].”
End-to-End Speech Separation with Unfolded Iterative Phase Reconstruction (2018, methodology): “The inventions of deep clustering [1, 2, 3], deep attractor networks [4, 5] and permutation free training [1, 2, 6, 7] have dramatically improved the performance of single-channel speaker-independent multispeaker speech separation, demonstrating overwhelming advantages over previous methods includin”
Bootstrapping Single-channel Source Separation via Unsupervised Spatial Clustering on Stereo Mixtures (2018, methodology): “The estimation of confidence measures of source separation estimates was explored in [14] and learning to separate sources using features derived from cues was explored in [15].”
Training Deep Networks with Structured Layers by Matrix Backpropagation (2015, methodology): “Our work relates both to the extensive literature in the area of (deep) neural networks (see [4] for a review) with particular emphasis on ConvNets[2, 5] and with (shallow) architectures that have been proven popular and successful in computer vision[6, 13, 8, 14].”
Evolutionary spectral clustering by incorporating temporal smoothness (2007, background): “[2] proved that for a given partition Z, the normalized cut can be equivalently written as”
Single-Channel Multi-Speaker Separation Using Deep Clustering (2016, background): “Such a framework was provided in subsequent work on spectral clustering [6], at the cost of prohibitive complexity.”
Handbook of Cluster Analysis (2015, methodology): “This intuition has been proved formally by [4, 25].”
Spectral Clustering for Speech Separation (2009, background): “The computation of eigenvectors and eigenvalues is a well-studied problem in numerical linear algebra (see, e.g., Golub and Loan, 1996).”