The creation of semantically relevant clusters is vital in bag-of-visual words models which are known to be very successful to achieve image classification tasks. Generally, unsupervised clustering algorithms, such as...
详细信息
The creation of semantically relevant clusters is vital in bag-of-visual words models which are known to be very successful to achieve image classification tasks. Generally, unsupervised clustering algorithms, such as K-means, are employed to create such clusters from which visual dictionaries are deduced. K-means achieves a hard assignment by associating each image descriptor to the cluster with the nearest mean. By this way, the within-cluster sum of squares of distances is minimized. A limitation of this approach in the context of image classification is that it usually does not use any supervision that limits the discriminative power of the resulting visual words (typically the centroids of the clusters). More recently, some supervised dictionary creation methods based on both supervised information and data fitting were proposed leading to more discriminative visual words. But, none of them consider the uncertainty present at both image descriptor and cluster levels. In this paper, we propose a supervised learning algorithm based on a Gaussian mixture model which not only generalizes the K-means algorithm by allowing soft assignments, but also exploits supervised information to improve the discriminative power of the clusters. Technically, our algorithm aims at optimizing, using an EM-based approach, a convex combination of two criteria: the first one is unsupervised and based on the likelihood of the training data;the second is supervised and takes into account the purity of the clusters. We show on two well-known datasets that our method is able to create more relevant clusters by comparing its behavior with the state of the art dictionary creation methods. (C) 2011 Elsevier Ltd. All rights reserved.
Students' engagements reflect their level of involvement in an ongoing learning process which can be estimated through their interactions with a computer-based learning or assessment system. A pre-requirement for ...
详细信息
Students' engagements reflect their level of involvement in an ongoing learning process which can be estimated through their interactions with a computer-based learning or assessment system. A pre-requirement for stimulating student engagement lies in the capability to have an approximate representation model for comprehending students' varied (dis)engagement behaviors. In this paper, we utilized model-based clustering for this purpose which generates K mixture Markov models to group students' traces containing their (dis)engagement behavioral patterns. To prevent the expectation-maximization (EM) algorithm from getting stuck in a local maxima, we also introduced a K-means-based initialization method named as K-EM. We performed an experimental work on two real datasets using the three variants of the EM algorithm the original EM, emEM, K-EM;and, non-mixture baseline models for both datasets. The proposed K-EM has shown very promising results and achieved significant performance difference in comparison with the other approaches particularly using the Dataset 1. Hence, we suggest to perform further experiments using large dataset(s) to validate our method. Additionally, visualization of the resultant clusters through first-order Markov chains reveals very useful insights about (dis)engagement behaviors depicted by the students. We conclude the paper with a discussion on the usefulness of our approach, limitations and potential extensions of this work.
In this research, a dynamic linear spatio-temporal model (DLSTM) was developed and evaluated for monthly streamflow forecasting. For parameter estimation, coupled expectation-maximization (EM) algorithm and Kalman fil...
详细信息
In this research, a dynamic linear spatio-temporal model (DLSTM) was developed and evaluated for monthly streamflow forecasting. For parameter estimation, coupled expectation-maximization (EM) algorithm and Kalman filter was adopted. This combination enables the model to estimate the state vector and parameters concurrently. Different forecast scenarios including various combinations of upstream stations were considered for downstream station streamflow forecasting. Several statistical criteria, nonparametric and visual tests were used for model evaluation. Results indicated that the spatio-temporal model performed acceptably in almost all scenarios. The dynamic model was able to capitalize on coupled spatial and temporal information provided that there is spatial connectivity in the studied hydrometric stations network. Moreover, threshold level method was used for model evaluation in drought andwet periods. Results indicated that, in validation phase, the model was able to forecast the drought duration and volume deficit/over threshold, although volume deficit/over threshold could not be accurately simulated.
A developmental trajectory describes the course of behavior over time. Identifying multiple trajectories within an overall developmental process permits a focus on subgroups of particular interest. We introduce a fram...
详细信息
A developmental trajectory describes the course of behavior over time. Identifying multiple trajectories within an overall developmental process permits a focus on subgroups of particular interest. We introduce a framework for identifying trajectories by using the expectation-maximization (EM) algorithm to fit semiparametric mixtures of logistic distributions to longitudinal binary data. For performance comparison, we consider full maximizationalgorithms (PROC TRAJ in SAS), standard EM, and two other EM-based algorithms for speeding up convergence. Simulation shows that EM methods produce more accurate parameter estimates. The EM methodology is illustrated with a longitudinal dataset involving adolescents smoking behaviors.
In this paper we introduce a multivariate family of distributions for multivariate count data with excess zeros, which is a multivariate extension of the univariate zero-inflated Bell distribution. We derive various g...
详细信息
In this paper we introduce a multivariate family of distributions for multivariate count data with excess zeros, which is a multivariate extension of the univariate zero-inflated Bell distribution. We derive various general properties of this multivariate distribution. In particular, the marginal distributions are univariate zero-inflated Bell distributions. The model parameters are estimated using the traditional maximum likelihood estimation method. In addition, we develop a simple EM algorithm to compute the maximum likelihood estimates of the parameters of the new multivariate distribution with closed-form expressions for the maximum likelihood estimators. Empirical applications that employ real multivariate count data are considered to illustrate the usefulness of the new class of multivariate distributions, and comparisons with the multivariate zero-inflated Poisson distribution, multivariate zero-adjusted Poisson distributions, and multivariate zero-inflated generalized Poisson distribution are made. (c) 2021 Elsevier Inc. All rights reserved.
Wireless communications over unlicensed frequency bands are expected to suffer from significant co-channel interference, which inevitably limits the error-rate performance. In this letter, an orthogonal frequency-divi...
详细信息
Wireless communications over unlicensed frequency bands are expected to suffer from significant co-channel interference, which inevitably limits the error-rate performance. In this letter, an orthogonal frequency-division multiplexing system employing bit-interleaved coded-modulation is considered and the problem of reliable decoding in the presence of unknown narrowband interference (NBI) is addressed. The proposed solution operates in an iterative fashion according to the expectation-maximization algorithm and is derived by modeling the interference power on each subcarrier as a random variable with an inverse gamma distribution. Computer simulations indicate that the resulting scheme is effective against NBI and, compared to existing alternatives, it achieves a better trade-off in terms of error rate performance, computational complexity and system overhead.
Objective: An adaptive harmonic separation (HS) technique is proposed to overcome the limitations in conventional filtering techniques for ultrasound (US) tissue harmonic imaging (THI). Methods: Based on expectation -...
详细信息
Objective: An adaptive harmonic separation (HS) technique is proposed to overcome the limitations in conventional filtering techniques for ultrasound (US) tissue harmonic imaging (THI). Methods: Based on expectation -maximization source separation, the proposed HS technique adaptively models the depth -varying fundamental and harmonic components in the frequency domain and separates the two by applying their calculated posterior probabilities. Phantom experiments with a Tx center frequency of 2 MHz are conducted to evaluate the proposed HS -based US THI schemes. Results: The phantom images show that the proposed single -pulse THI scheme utilizing the HS technique provides not only an average improvement of 19.2 % in axial resolution compared to the conventional bandpass filtering scheme but also similar image quality to that of the conventional pulse -inversion (PI) scheme which requires two Tx/Rx sequences for each scan line. Furthermore, when combined with the PI technique, the HS technique provides a uniform axial resolution over the entire 170 mm imaging depth with an average improvement of 17.1 % compared to the conventional PI scheme. Conclusion: These results show that the proposed adaptive HS technique is capable of improving both the frame rate and the image quality of US THI.
We consider the problem of multitask learning (MTL), in which we simultaneously learn classifiers for multiple data sets (tasks), with sharing of intertask data as appropriate. We introduce a set of relevance paramete...
详细信息
We consider the problem of multitask learning (MTL), in which we simultaneously learn classifiers for multiple data sets (tasks), with sharing of intertask data as appropriate. We introduce a set of relevance parameters that control the degree to which data from other tasks are used in estimating the current task's classifier parameters. The set of relevance parameters are learned by maximizing their posterior probability, yielding an expectation-maximization (EM) algorithm. We illustrate the effectiveness of our approach through experimental results on a practical data set.
The simplified joint channel estimation and symbol detection based on the EM (expectation-maximization) algorithm for space-time block code (STBC) are proposed. By assuming channel to be invariant within only one STBC...
详细信息
The simplified joint channel estimation and symbol detection based on the EM (expectation-maximization) algorithm for space-time block code (STBC) are proposed. By assuming channel to be invariant within only one STBC word and utilizing the orthogonal structure of STBC, the computational complexity and cost of this algorithm are both very low, so it is very suitable to implementation in real systems.
We propose a variable selection method for multivariate hidden Markov models with continuous responses that are partially or completely missing at a given time occasion. Through this procedure, we achieve a dimensiona...
详细信息
We propose a variable selection method for multivariate hidden Markov models with continuous responses that are partially or completely missing at a given time occasion. Through this procedure, we achieve a dimensionality reduction by selecting the subset of the most informative responses for clustering individuals and simultaneously choosing the optimal number of these clusters corresponding to latent states. The approach is based on comparing different model specifications in terms of the subset of responses assumed to be dependent on the latent states, and it relies on a greedy search algorithm based on the Bayesian information criterion seen as an approximation of the Bayes factor. A suitable expectation-maximization algorithm is employed to obtain maximum likelihood estimates of the model parameters under the missing-at-random assumption. The proposal is illustrated via Monte Carlo simulation and an application where development indicators collected over eighteen years are selected, and countries are clustered into groups to evaluate their growth over time.
暂无评论