Dimensionality Reduction for Fast Similarity Search in Large Time Series Databases

The problem of similarity search in large time series databases has attracted much attention recently. It is a non-trivial problem because of the inherent high dimensionality of the data. The most promising solutions involve first performing dimensionality reduction on the data, and then indexing the reduced data with a spatial access method. Three major dimensionality reduction techniques have been proposed: Singular Value Decomposition (SVD), the Discrete Fourier transform (DFT), and more recently the Discrete Wavelet Transform (DWT). In this work we introduce a new dimensionality reduction technique which we call Piecewise Aggregate Approximation (PAA). We theoretically and empirically compare it to the other techniques and demonstrate its superiority. In addition to being competitive with or faster than the other methods, our approach has numerous other advantages. It is simple to understand and to implement, it allows more flexible distance measures, including weighted Euclidean queries, and the index can be built in linear time. Received 16 May 2000 / Revised 18 December 2000 / Accepted in revised form 2 January 2001     

Converting heterogeneous statistical tables on the web to searchable databases

Much of the world’s quantitative data reside in scattered web tables. For a meaningful role in Big Data analytics, the facts reported in these tables must be brought into a uniform framework. Based on a formalization of header-indexed tables, we proffer an algorithmic solution to end-to-end table processing for a large class of human-readable tables. The proposed algorithms transform header-indexed tables to a category table format that maps easily to a variety of industry-standard data stores for query processing. The algorithms segment table regions based on the unique indexing of the data region by header paths, classify table cells, and factor header category structures of two-dimensional as well as the less common multidimensional tables. Experimental evaluations substantiate the algorithmic approach to processing heterogeneous tables. As demonstrable results, the algorithms generate queryable relational database tables and semantic-web triple stores. Application of our algorithms to 400 web tables randomly selected from diverse sources shows that the algorithmic solution automates end-to-end table processing.     

Relevance feedback techniques in the MARS image retrieval system

Content-based image retrieval (CBIR) has become one of the most active research areas in the past few years. Many visual feature representations have been explored and many systems built. In this paper, we focus on an important component of these systems - relevance feedback - and how we incorporated it into the MARS retrieval system. Relevance feedback techniques are based on an interactive retrieval approach to effectively take into account user preferences to provide an improved search experience. We present a series of coherent strategies, from single-point to multipoint and multifeature approaches that we have seamlessly integrated into our system and present experimental results to show their retrieval performance characteristics.Keywords: Image retrieval - Query refinement - Relevance feedbackMichael Ortega-Binderberger: michaelo@us.ibm.comThis work was performed while the author was a Ph.D. student at the University of Illinois at Urbana-Champaign. Correspondence to: This material is based on work supported in part by the National Science Foundation under Award Numbers CAREER IIS-9734300, 9996140, 0083489, 0331707, and 0331690 and in part by the Army Research Laboratory under Cooperative Agreement No. DAAL01-96-2-0003. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation or the Army Research Laboratory. Michael Ortega-Binderberger was supported in part by CONACYT award # 89061.     

An algorithm for engineering design optimization

In this paper a new concept for development of algorithms for optimal design of engineering systems is presented. The basic idea is to use upper and lower bounds on optimum cost to develop iterative search strategies. The main feature of the concept is that it does not rely on one-dimensional search to compute a step size at any design iteration. Implication of the feature is that the algorithms based on this concept require evaluation of constraint functions only once at any design iteration. This is highly desirable for optimal design of engineering systems because evaluation of functions for such systems is very expensive due to their implicit dependence on design variables. An algorithm based on the new concept is derived in the paper. Several new step sizes are introduced and their relation to proper reduced optimal design problems are presented. A new step size based on the constant cost requirement at some design iterations is introduced. Numerical aspects for the algorithm are also presented. Based on the new algorithm, a general-purpose computer code GRP2 is developed. The code is used to solve several problems to gain experience and insight for the algorithm. Numerical experience with examples is discussed. It is concluded that algorithms based on bounding optimum cost have substantial potential for applications in optimal design of engineering systems.     

Relative humidity based model for estimation of reference evapotranspiration for western plateau and hills region of India

As an important constituent of hydrological cycle, estimation of Reference Evapotranspiration (ET_o) is essential for planning of projects related to water resources. This article compares the results obtained from the available pan evaporation based equations to estimate ET_o from pan evaporation (E_p) and proposes a new simplified methodology for estimation of ET_o, which needs only the data of relative humidity (H). The resulting model is based upon the collected meteorological data of a selected study area i.e. Nagpur District in Maharashtra State, which lies in western plateau and hills region of India. In the proposed methodology, E_p rates are adjusted to the values expected for 50 % relative humidity. Then, the relationship between ET_o and the adjusted E_p is established. The validation of the proposed new model has been carried out by comparing its results with the results obtained by the pan evaporation method, for the study area as well as for the data set of another area. This new model is found to be reasonably accurate for estimation of reference evapotranspiration for the study area and as such, is expected to be applicable to the river basins located in the regions having climatic conditions similar to that of the study area. Although, if the conditions are different then another site specific model can be developed by adopting the methodology proposed in this article.