A Tai mapping between two rooted labeled trees (trees, for short) is a one-to-one node correspondence preserving ancestors and siblings (if trees are ordered). The variations of the Tai mapping are known to provide a hierarchy, called a Tai mapping hierarchy. In this paper, we characterize the Tai mapping hierarchy relevant to a common subforest by focusing on the connections of nodes and the arrangements of subtrees in a common subforest. Then, we ll a gap in the Tai mapping hierarchy by introducing several new variations. Furthermore, we summarize and investigate the time complexity of computing the variations of the edit distance as the minimum cost of the variations of the Tai mapping.

It was very recent that a formal definition of multiple alignments is given to general data structures that include not only strings but also trees and graphs. Also, it has been shown that the center star method, which is to compute approximately optimal multiple alignments for strings, is effective for the generalized multiple alignments. In this paper, we report the results of our experiments to prove effectiveness of the generalized multiple alignments and the extended center star method by taking trees as an example.

We propose an index and search algorithm based on the edit distance with moves. Edit distance with moves (EDM) is a distance measure between two strings, including substring moves in addition to edit distance to turn one string to the other. It is known to computing exact EDM is hard.Thus, an index structure and a search algorithm with EDM doesn't exist. We show the first algorithm named string index for edit distance with moves (siEDM) for indexing and searching strings with an approximate EDM. Using edit sensitive parsing (ESP), our method builds an index structure and prunes the search space for searching query by proving a bound of the approximate EDM. This bound for fast query searching is equal to the bound of ESP.

We study the loss version of adversarial multi-armed bandit problems with one lossless arm. We show an adversary's strategy that forces any player to suffer K - 1 - O(1/T) loss where K is the number of arms and T is the number of rounds.

This paper evaluates three incremental matrix approximation algorithms, Brute Force, Incremental Singular Value Decomposition (iSVD) and Frequent Directions (FD), in terms of running time and accuracy. The authors assume that each feature vector comes from data stream one-by-one, and the algorithms try to incrementally update already-known SVD. As a result of experiments for 256-by-7291 dense matrix from a handwritten digits dataset, iSVD ran faster than brute-force algorithm. Importantly, since iSVD is a straightforward incremental technique for SVD updating, the result always gave nearly optimal low-rank approximation. By contrast, FD shrinks from the original 256-by-7291 matrix to a tiny 256-by-l matrix, and provides approximated singular vectors. The singular vectors were nearly-optimal to estimate the best low-rank approximation of the original matrix, and running time was much faster than iSVD. As conclusions, this paper discusses possible applications and future directions of these incremental algorithms.

相互情報量を推定するR パッケージmi.correct を公開した。その応用として、今回は、株式の銘柄間の関連性を見出すことを試みた。

We implement the simple algorithm for counting frequent items in stream data by a constant space, and compare the performance with a naive algorithm.

In this paper, first we formulate a kissing hairpin between two RNA sequences as RNA tertiary structures kissing two hairpin loops for every RNA sequence. Then, we design the method of finding all of the maximal kissing hairpins between two RNA sequences, where a kissing hairpin is maximal if the length of its kissing part is maximal. Finally, by connecting maximal kissing hairpins with the same kissing part between two RNA segments, we give experimental results of finding maximal kissing hairpins through 8 RNA segments in influenza A (H3N2) viruses.

Highlly repetitive texts have been increasing still now. Examples are individual genomes, version controlled documents and repositries of source codes. Although grammar compression is an efficient for highly repetitive texts, the working space is still large for a data structure called reverse phrase dictionary. In this paper, we propose two methods for ruducing the space of reverse phrase dictionary. One is a space efficient dynamic 1-dimentional range search for reverse phrase dictionary, the other is a space reduction method of reverse phrase dictionary for a succinct represetation of grammar compression.