Academic
Publications
Efficient Greedy Learning of Gaussian Mixture Models

Efficient Greedy Learning of Gaussian Mixture Models,10.1162/089976603762553004,Neural Computation,Jakob J. Verbeek,Nikos A. Vlassis,Ben J. A. Kröse

Efficient Greedy Learning of Gaussian Mixture Models   (Citations: 117)
BibTex | RIS | RefWorks Download
This paper concerns the greedy learning of Gaussian mixtures. In the greedy ap- proach, mixture components are inserted into the mixture one after the other. We propose a heuristic for searching for the optimal component to insert. In a randomized manner a set of candidate new components is generated. For each of these candidates we find the locally optimal new component. The best local optimum is then inserted into the existing mixture. The resulting algorithm resolves the sensitivity to initializa- tion of state-of-the-art methods, like EM, and has running time linear in the number of data points and quadratic in the (final) number of mixture components. Due to its greedy nature the algorithm can be particularly useful when the optimal number of mixture components is unknown. Experimental results comparing the proposed algo- rithm to other methods on density estimation and texture segmentation are provided.
Journal: Neural Computation - NECO , vol. 15, no. 2, pp. 469-485, 2003
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...To reduce the user’s workload, we use the greedy EM algorithm [35], which builds models in an adaptive manner...
    • ...Although there is no known constructive method to find the global maximum, the greedy EM algorithm we adopted locates the global maximum using a search heuristic [35]...

    Yunhai Wanget al. Efficient Volume Exploration Using the Gaussian Mixture Model

    • ...Moreover, we also compare our approach (we refer to it as EEM) with Figueiredo-Jain algorithm (FJ-EM) [11], deterministic annealing based model selection method (DAMS) [17], greedy EM method [13], as well as variational component splitting method (VCS) [12]...

    Boyu Wanget al. Entropy penalized learning for Gaussian mixture models

    • ...The traffic phases can be identified using the fundamental diagram flow vs. density, as shown in Fig. 2 for the sensor i =7 . In this example we employ a Gaussian mixture model (GMM) [8] clustering method to classify the traffic phase, assuming a mixture of two Gaussian components for the joint distribution of flow and density:...
    • ...where M is the number of components, and gm(·|µm,Cm) is the m-th Gaussian distribution with (NP +1 )× 1 vector of mean values µm and (NP +1 )× (NP +1 )covariance matrix Cm. Parameters {αm,µ m ,Cm} M=1 are inferred from the historical data provided by the traffic monitoring network using the EM algorithm [7] [8]...
    • ...This has an impact on complexity of offline procedures - number of iterations of EM for GMM learning in the order of M 2 D where D is the dimension of dataset [8]- , and on online steps - computational cost is O(N 2...

    A. Pascaleet al. Adaptive Bayesian network for traffic flow prediction

    • ...Greedy learning of GMM, recently proposed in [32, 34], overcomes the drawbacks of the EM algorithm (e.g...
    • ...This idea was implemented in [34] and further improved in [32], and is utilized in the present work...

    Dror Lederman. An endotracheal intubation confirmation system based on carina image d...

    • ...method to learn the gaussians mixture model configuration [9]...

    Nicola Greggioet al. Unsupervised Greedy Learning of Finite Mixture Models

Sort by: