Rosaria Rossini
Published: Thursday, January 22, 2015
il 28 gennaio, alle h. 11, in Sala seminari, si terrà l’esame finale di
Rosaria Rossini (ciclo XXVI, tutor Maria Luisa Sapino).
Leveraging salient features for exploring time series data sets
The work presented in the thesis exploits salient temporal features to develop new algorithms for querying, summarizing, visualizing, smoothing, and exploring time series data sets. A representation of the time series based on a multi-level structure is presented by using salient temporal points that are able to identify and describe salient regions in a time series. The thesis investigates the possibility to exploit these features in terms of salient events and proposes similarity criteria which take into account the importance of the relevant features and use them as a measure of similarity. Furthermore some structural features in time series create consistent salient alignments of a given pair of time series that can be used to create alternative adaptive bands based on salient temporal features to improve the Dynamic Time Warping (DTW) algorithm. The thesis also includes the use of the local properties of the time series based on their salient points to define alternative adaptive smoothing criteria for time series. A query-and feature-driven temporal data set visualization system based on salient temporal features is proposed to explore time series data set.
The work presented in the thesis exploits salient temporal features to develop new algorithms for querying, summarizing, visualizing, smoothing, and exploring time series data sets. A representation of the time series based on a multi-level structure is presented by using salient temporal points that are able to identify and describe salient regions in a time series. The thesis investigates the possibility to exploit these features in terms of salient events and proposes similarity criteria which take into account the importance of the relevant features and use them as a measure of similarity. Furthermore some structural features in time series create consistent salient alignments of a given pair of time series that can be used to create alternative adaptive bands based on salient temporal features to improve the Dynamic Time Warping (DTW) algorithm. The thesis also includes the use of the local properties of the time series based on their salient points to define alternative adaptive smoothing criteria for time series. A query-and feature-driven temporal data set visualization system based on salient temporal features is proposed to explore time series data set.
La Commissione sarà composta dai seguenti membri:
Vincenzo Moscato (Univ. Napoli), Gualtiero Volpe (Univ. Genova), Ruggero Pensa (Univ. Torino)
Supplenti: Antonio Picariello (Univ. Napoli), Antonio Camurri (Univ. Genova), Rosa Meo (Univ. Torino)