Representative Clustering of Uncertain Data

... their drawbacks. Clustering Using Expected Distances. The main drawback of approaches based on expected distances [41, 21, 23] is the information loss incurred by describing a complex probability distance function by a single scalar. Considering additional moments of a probability distance function, ...

... their drawbacks. Clustering Using Expected Distances. The main drawback of approaches based on expected distances [41, 21, 23] is the information loss incurred by describing a complex probability distance function by a single scalar. Considering additional moments of a probability distance function, ...

Using extended feature objects for partial similarity

... above complexity to increase by a factor of m2 (all edge sequences of arbitrary length). The ‘continuous’ problem of finding similar portions of the two polygons starting and ending at an arbitrary point (not necessarily a vertex) on any edge of the contour of the polygon is even more difficult and, ...

... above complexity to increase by a factor of m2 (all edge sequences of arbitrary length). The ‘continuous’ problem of finding similar portions of the two polygons starting and ending at an arbitrary point (not necessarily a vertex) on any edge of the contour of the polygon is even more difficult and, ...

finding or not finding rules in time series

... Our claim is surprising since it calls into question the contributions of dozens of papers. In fact, the existence of so much work based on STS clustering offers an obvious counter argument to our claim. It could be argued: “Since many papers have been published which use time series subsequence clu ...

... Our claim is surprising since it calls into question the contributions of dozens of papers. In fact, the existence of so much work based on STS clustering offers an obvious counter argument to our claim. It could be argued: “Since many papers have been published which use time series subsequence clu ...

Improving student model for individualized learning

... been used to enhance human learning. These environments aim at increasing student achievement by providing individualized instructions. It has been recognized that individualized learning is more effective than the conventional learning. Student models which are used to capture student knowledge und ...

... been used to enhance human learning. These environments aim at increasing student achievement by providing individualized instructions. It has been recognized that individualized learning is more effective than the conventional learning. Student models which are used to capture student knowledge und ...

Semantically-grounded construction of centroids for datasets with

... 2009). More flexible approaches are based on constructing a new synthetic centroid using an averaging function, which is applied to all the attribute values of the objects in the dataset. In both cases, distance functions are needed. The case of centroid construction for numerical databases has been ...

... 2009). More flexible approaches are based on constructing a new synthetic centroid using an averaging function, which is applied to all the attribute values of the objects in the dataset. In both cases, distance functions are needed. The case of centroid construction for numerical databases has been ...

Unconditional Quantile Regressions

... influence of an individual observation on a distributional statistic of interest. Influence functions of commonly used statistics are either well known or easy to derive. For example, the influence function of the mean µ = E [Y ] is the demeaned value of the outcome variable, Y − µ. Adding back the ...

... influence of an individual observation on a distributional statistic of interest. Influence functions of commonly used statistics are either well known or easy to derive. For example, the influence function of the mean µ = E [Y ] is the demeaned value of the outcome variable, Y − µ. Adding back the ...

Tree-Based State Generalization with Temporally Abstract Actions

... refers to techniques that group sequences of actions together and treat them as one abstract action, e.g. [5–9]. Using a function approximator for the value function, e.g. [10], can, in theory, subsume both state and temporal abstraction, but the authors are unaware of any of these techniques that, ...

... refers to techniques that group sequences of actions together and treat them as one abstract action, e.g. [5–9]. Using a function approximator for the value function, e.g. [10], can, in theory, subsume both state and temporal abstraction, but the authors are unaware of any of these techniques that, ...

Synthetic Datasets for Clustering Algorithms

... calculated using a distance measure.1 The family of Lk -norm distances, Mahalanobis distance functions are a few distance measures to mention. The major goal of clustering is to identify underlying patterns based on the similarities between the objects. Clustering is also knows as un-supervised lear ...

... calculated using a distance measure.1 The family of Lk -norm distances, Mahalanobis distance functions are a few distance measures to mention. The major goal of clustering is to identify underlying patterns based on the similarities between the objects. Clustering is also knows as un-supervised lear ...

Scale-free Clustering - UEF Electronic Publications

... might also be usable in other classification problems. In addition, a similarity measure based on mutual information to be used with hierarchical clustering [Koj04] as well as a metric based on Fisher information matrix [KSP01] have also been presented. The latter is a metric which is learned from t ...

... might also be usable in other classification problems. In addition, a similarity measure based on mutual information to be used with hierarchical clustering [Koj04] as well as a metric based on Fisher information matrix [KSP01] have also been presented. The latter is a metric which is learned from t ...

# Expectation–maximization algorithm

In statistics, an expectation–maximization (EM) algorithm is an iterative method for finding maximum likelihood or maximum a posteriori (MAP) estimates of parameters in statistical models, where the model depends on unobserved latent variables. The EM iteration alternates between performing an expectation (E) step, which creates a function for the expectation of the log-likelihood evaluated using the current estimate for the parameters, and a maximization (M) step, which computes parameters maximizing the expected log-likelihood found on the E step. These parameter-estimates are then used to determine the distribution of the latent variables in the next E step.