MBPD: Motif-Based Period Detection
... of this work being time series makes it suitable for other kinds of data such as multimedia because they can be converted to time series e.g. the extraction of MFCC from audio as it is used for one of the datasets in our experiments. Several methods have been proposed to detect periods in data. Most ...
... of this work being time series makes it suitable for other kinds of data such as multimedia because they can be converted to time series e.g. the extraction of MFCC from audio as it is used for one of the datasets in our experiments. Several methods have been proposed to detect periods in data. Most ...
Outlier Detection using Semi-supervised and Unsupervised Learning on High Dimensional Data
... parameter. But in “local mode” when k ≪ n can cause problems with scores discrimination in high-dimensional settings. III. ...
... parameter. But in “local mode” when k ≪ n can cause problems with scores discrimination in high-dimensional settings. III. ...
MapReduce-Based Pattern Finding Algorithm
... space [15]. However all these algorithms mentioned above ignored to consider the limitation of the main memory of one computer. So for further improving the performance of pattern finding and breaking through single computer resource constraints, we design a parallel pattern finding algorithm based ...
... space [15]. However all these algorithms mentioned above ignored to consider the limitation of the main memory of one computer. So for further improving the performance of pattern finding and breaking through single computer resource constraints, we design a parallel pattern finding algorithm based ...
Learning Transformation Models for Ranking and Survival Analysis
... question how to solve the estimation equations numerically is often approached in an ad hoc way (if at all, see Kalbfleisch and Prentice, 2002 and references). We find that the class of transformation models is a powerful tool to model data arising from survival studies for different reasons. The fi ...
... question how to solve the estimation equations numerically is often approached in an ad hoc way (if at all, see Kalbfleisch and Prentice, 2002 and references). We find that the class of transformation models is a powerful tool to model data arising from survival studies for different reasons. The fi ...
Clustering Data with Measurement Errors
... perspectives. See [10, 11] for a review. In recent years probability models have been proposed as a basis for cluster analysis [1, 4, 7, 9, 15]. Methods of this type have shown promise in a number of practical applications [1, 4, 7]. In this approach, the data are viewed as coming from a mixture of ...
... perspectives. See [10, 11] for a review. In recent years probability models have been proposed as a basis for cluster analysis [1, 4, 7, 9, 15]. Methods of this type have shown promise in a number of practical applications [1, 4, 7]. In this approach, the data are viewed as coming from a mixture of ...
ADR-Miner - An Ant-Based Data Reduction Algorithm for Classification
... shown a successful track record with combinatorial optimization problems in multiple fields of application, and has since been extended to handle other types of problems, including data mining problems [38]. In this dissertation I introduce ADR-Miner: an ant-inspired algorithm designed to do data re ...
... shown a successful track record with combinatorial optimization problems in multiple fields of application, and has since been extended to handle other types of problems, including data mining problems [38]. In this dissertation I introduce ADR-Miner: an ant-inspired algorithm designed to do data re ...
MOSAIC: A Proximity Graph Approach to Agglomerative Clustering
... In summary, MOSAIC merges pairs of neighboring clusters maximizing an externally given fitness function q, and this process is continued until only one cluster is left. Finally, the best clustering is determined and returned. Using cluster representatives obtained from a representative-based cluster ...
... In summary, MOSAIC merges pairs of neighboring clusters maximizing an externally given fitness function q, and this process is continued until only one cluster is left. Finally, the best clustering is determined and returned. Using cluster representatives obtained from a representative-based cluster ...
Multi-Dimensional Characterization of Temporal Data Mining on
... Memory hierarchy: There are several types of memory accessible by an execution core on an NVIDIA GPU. Located on-chip, each multiprocessor contains a set of 32-bit registers along with a shared memory region which is quickly accessible by any core on the multiprocessor, but hidden from other multipr ...
... Memory hierarchy: There are several types of memory accessible by an execution core on an NVIDIA GPU. Located on-chip, each multiprocessor contains a set of 32-bit registers along with a shared memory region which is quickly accessible by any core on the multiprocessor, but hidden from other multipr ...
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.