-
Issei HAMADA, Takaharu SHIMADA, Daiki NAKATA, Kouichi HIRATA
Article type: SIG paper
Pages
01-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we classify nucleotide sequences of in uenza A viruses by using various kernels. Our kernels mainly consist of nucleotide sequence kernels by regarding nucleotide sequences as vectors, multisets and strings and phylogenetic tree kernels applied to phylogenetic trees reconstructed from a set of nucleotide sequences. Then, we evaluate that the phylogenetic tree kernels are effective to the pandemic classification and the regional analysis, while the nucleotide sequence kernels are effective to the pandemic classification and the analysis of positions in packaging signals.
View full abstract
-
Noriaki CHIKARA, Miyuki KOSHIMURA, Mitsuo NISHIDA, Yukihiro ABE, Hiros ...
Article type: SIG paper
Pages
02-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
We developed a software system which supports chemical experiment with Inductive Logic Programming (ILP). The system aims to guide experimental chemists to get good experimental results which satisfy their objectives by analyzing experimental data obtained so far. This paper shows two methods with ILP: one is for experimental data with positive examples, another is for that without positive examples.
View full abstract
-
[in Japanese], [in Japanese], [in Japanese]
Article type: SIG paper
Pages
03-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
We propose a novel formulation to find representatives based on structured sparse learning. To optimize our objective function, we propose the fast iterative shrinkage-thresholding algorithm combined with the proximal-Dykstra method and the calculation of parametric maximum ows. Experiments on three real-world image datasets validate the effectiveness of the proposed method in finding exemplars with diversity and representativeness.
View full abstract
-
Tetsuji KUBOYAMA, Kilho SHIN
Article type: SIG paper
Pages
05-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
CWC is a consistency-based filter-type feature selection algorithm, which is as accurate as and 30 to 50 times faster than the best consistency-based algorithm known in the literature. Since CWC deploys a consistency measure, it is significantly more accurate than other filter-type algorithms that are not consistency-based, and shows compatible performance in time efficiency. CWC employs the binary concistency measure that is the simplest and most rigid consistency measure. It has not been well-studied why the binary consistency measure elicits the superiror performance among the other consistency measures. To find the clues, we report an empirical comparative study of the existing consistency measures.
View full abstract
-
Koji MAEDA, Yoshimasa TAKABATAKE, Yasuo TABEI, Hiroshi SAKAMOTO
Article type: SIG paper
Pages
06-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
We propose an enumeration algorithm of short Hamming distance based on grammar compression by using split search. Pattern search and frequent pattern discovery using ESP-index[9] have been proposed, but enumeration algorithm of short Hamming distance has not been proposed yet. Using ESP-index and split search, we propose an enumeration algorithm of short Hamming distance. We experiment our algorithm for DNA text and achieve fastar calculate than FM-index[10].
View full abstract
-
[in Japanese]
Article type: SIG paper
Pages
07-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
We show a way to view the edit distance problem for trees as a pattern recognition problem.
View full abstract
-
Takuya YOSHINO, Kouichi HIRATA
Article type: SIG paper
Pages
08-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we investigate an alignment distance between the following three cyclically ordered trees. A biordered tree is an unordered tree that allows both a left-to-right and a right-to-left order among siblings. A cyclic-ordered tree is an unordered tree that allows cyclic order among siblings in a left-to-right direction. A cyclic-biordered tree is an unordered tree that allows cyclic order among siblings in both left-to-right and right-to-left directions. Then, we design the algorithms to compute the distance between biordered trees in O(n2D2) time and ones between cyclic-ordered trees and cyclic-biordered trees in O(n2D4) time, where n is the maximum number of nodes and D is the maximum degree in two given trees.
View full abstract
-
[in Japanese], [in Japanese], [in Japanese]
Article type: SIG paper
Pages
09-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
This paper considers causal discovery between discrete and continuous variables based on additive noise model. In many database, some fields are discrete while others continuous. However, the previous notion assumes that all the variables are either discrete or continuous. In this paper, we prove that for discrete (m values) and continuous variables X, Y , causality X ! Y cannot be identified for m = 2 under regular conditions, and conjecture that X ! Y can be identified for m · 3, and that Y ! X can be identified for any m. Several experiments support those properties successfully. Furthermore, using R, the program language, we implemented causal discovery between X ="month" and Y ="average temperature" in the data provided by the US National Weather Service Weather Forecast Office.
View full abstract
-
Akihiro MIYAGI, Hiroshi SAKAMOTO
Article type: SIG paper
Pages
10-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
We propose a faster online grammar compression based on frequent information in constant space. The online grammar compression based on frequent information[2] is slow due to the cost of dynamic updating of frequency table. Using the fast updatable counting frequency algorithm proposed by Ogata et al [4], we quickly update frequency in constant space. We experiment this algorithm for several texts and achieve faster grammar compression than [2].
View full abstract
-
Yoshiaki OKUBO, Makoto HARAGUCHI
Article type: SIG paper
Pages
11-
Published: July 24, 2014
Released on J-STAGE: July 01, 2021
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we are concerned with a method for retrieving objects which are similar to a given query object. Particularly, we formalize this task as a problem of finding a frequent pattern with the maximum length. The problem can be solved efficiently with an algorithm for extracting top-N colossal frequent patterns already proposed by the authors. We also discuss how to apply the proposed method to a problem of extracting similar melodies for a given query music.
View full abstract