Web-scale k-means clustering

By David Sculley. The Web Conference, 2010

1,327 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Mechanics, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/b452a856a3e3d4d37b1de837996aa6813bedfdcf

How researchers see this paper

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

In Artificial Intelligence

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

Theory and Experiments on Vector Quantized Autoencoders (2018, methodology): “These techniques have also been successfully used in minibatch K -means [26] and online EM [16, 25].”
Coarse graining molecular dynamics with graph neural networks. (2020, methodology): “A clustering algorithm such as k -means 79 is used to assign each data point in the MD simulation dataset to a discrete state.”
Learning Image Embeddings using Convolutional Neural Networks for Improved Multi-Modal Semantics (2014, methodology): “These descriptors were subsequently clustered using mini-batch k-means (Sculley, 2010) with 100 clusters.”
Reconciling meta-learning and continual learning with online mixtures of tasks (2018, methodology): “Incremental or stochastic clustering was considered in the EM setting in [37]. and in the K -means setting in [48].”
Learning to Coordinate Multiple Reinforcement Learning Agents for Diverse Query Reformulation (2018, methodology): “Specifically, we compare the random split against a mini-batch K-means clustering algorithm (Sculley, 2010).”
Scalable K-Means++ (2012, background): “Sculley [31] presented modiﬁcations to k -means for batch optimizations and to take data sparsity into account.”
Privacy-Preserving Visual Learning Using Doubly Permuted Homomorphic Encryption (2017, methodology): “zation strength (in [36]) was chosen Table 6. Clustering Results on Caltech101/256: adjusted mutual information scores given the correct number of clusters. Methods Caltech101 Caltech256 Privacy S10 [57] 0.733 0.630 7 DPHE 0.753 0.614 3 adaptively so that the sparsity of cluster centroids was more t”
High-Fidelity Image Generation With Fewer Labels (2019, neutral): “After learning the feature extractor F , we apply mini batch k -Means clustering (Sculley, 2010) on the representations of the training images.”
Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis (2023, neutral): “For performance reasons, a batched clustering strategy is used [23].”
Asynchronous parallel stochastic gradient descent: a numeric core for scalable distributed machine learning algorithms (2015, methodology): “T do 2: draw j ∈ { 1 . . . m } uniformly at random 3: update w t +1 ← w t − (cid:15)∂ w x j ( w t ) 4: return w T advantage in terms of computational cost with respect to the number of data samples is eminent: compared to batch updates of quadratic complexity, SGD updates come at linearly growing…”
Automatic Classification of Defective Photovoltaic Module Cells in Electroluminescence Images (2018, methodology): “For performance reasons, we use the fast mini-batch variant [34] of k-means.”
SONYC-UST-V2: An Urban Sound Tagging Dataset with Spatiotemporal Context (2020, background): “The model then predicted the class presence for unlabeled recordings, and recordings with class probabilities above a low fixed threshold were then clustered with minibatch k-means [7].”