We present a novel method for estimating tree-structured covariance matrices directly from observed continuous data. Specifically, we estimate a covariance matrix from observations of p continuous random variables encoding a stochastic process over a tree with p leaves. A representation of these classes of matrices as linear combinations of rank-one matrices indicating object partitions is used...

We propose a supervised word sense disambiguation (WSD) method using tree-structured conditional random fields (TCRFs). By applying TCRFs to a sentence described as a dependency tree structure, we conduct WSD as a labeling problem on tree structures. To incorporate dependencies between word senses, we introduce a set of features on tree edges, in combination with coarse-grained tagsets, and sho...

Tree-structured data models and formats are increasingly being used in many applications to represent and store data objects. Since the structural information inherent in such data cannot be captured in conventional index structures, quick retrieval of tree-structured data necessitates special indexing techniques. A variety of algorithms, such as ViST and Data Guides, have been proposed to answ...

Several recent works have discussed tree structured sparse coding [8, 10, 7, 3], where N data points in R written as the d × N matrix X are approximately decomposed into the product of matrices WZ . HereW is a d×K dictionary matrix, and Z is a K×N matrix of coefficients. In tree structured sparse coding, the rows of Z correspond to nodes on a tree, and the columns of Z are encouraged to be nonz...

It is well known that tree-structured perfect maps can be uniquely identified by computing a maximum weight spanning tree with mutual information providing the edge weights. In this paper I generalize the edge evaluation measure by stating the conditions such a measure has to satisfy in order to be able to identify tree-structured perfect maps. In addition, I show that not only mutual informati...

Very fast tree-structured vector quantization employs scalar quantization decisions at each level, but chooses the dimension on which to quantize based on the coordinate direction of maximum variance. Because the quantization is scalar, searches are no more complex than scalar quantization − providing significant improvement in complexity over full-searched or even tree-structured vector quanti...

Much complex decision-making is performed routinely by the software of a computer system. It is appropriate to study more thoroughly the performance of this built-in decisionmaking, because it can strongly influence the efficiency of the entire system. One objective of compilers is to produce a reasonably efficient machine-language version of a user’s program. Traditionally, one of the best opp...

The success of the Internet has filled the net with lots of symbolic representations of music works. Two kinds of problems arise to the user: content-based search of music and the identification of similar works. Both belong to the pattern recognition domain. In contrast to most of the existing approaches, we pose a non-linear representation of a melody, based on trees that express the metric a...

Within the class of “determinate” object-creating database transformations, identified by Abiteboul and Kanellakis, a natural proper subclass consists of the so-called “constructive” transformations. A determinate transformation is constructive if its input-output pairs satisfy a condition discovered by Jan Paredaens. In this note we point out that when object creation is “tree-structured,” as ...

Self-similarity is a property existing in natural, social, technological, biological, and many other networks. The whole and the parts, or one part and another part of self-similar networks present some similarity. In the past dozen years, researchers have proposed some deterministic self-similar networks by utilizing various kinds of mechanisms. In this paper, we propose a new deterministic se...