CONvergence of iterated CORrelations (CONCOR) 
Given an adjacency matrix, or a set of adjacency matrices for different relations, a correlation matrix can be formed by the following procedure. Form a profile vector for a vertex i by concatenating the ith row in every adjacency matrix; the i,jth element of the correlation matrix is the Pearson correlation coefficient of the profile vectors of i and j. This (square, symmetric) matrix is called the first correlation matrix. The procedure can be performed iteratively on the correlation matrix until convergence. Each entry is now 1 or -1. This matrix is used to split the data into two blocks such that members of the same block are positively correlated, members of different blocks are negatively correlated. CONCOR uses the above technique to split the initial data into two blocks. Successive splits are then applied to the separate blocks. At each iteration all blocks are submitted for analysis, however blocks containing two vertices are not split. Consequently n-partitions of the binary tree can produce up to 2n blocks. Note that any similarity matrix can be used as input.
http://…/concor-in-r …
Qualitative Dueling Bandit (QDB)
We formulate and study a novel multi-armed bandit problem called the qualitative dueling bandit (QDB) problem, where an agent observes not numeric but qualitative feedback by pulling each arm. We employ the same regret as the dueling bandit (DB) problem where the duel is carried out by comparing the qualitative feedback. Although we can naively use classic DB algorithms for solving the QDB problem, this reduction significantly worsens the performance—actually, in the QDB problem, the probability that one arm wins the duel over another arm can be directly estimated without carrying out actual duels. In this paper, we propose such direct algorithms for the QDB problem. Our theoretical analysis shows that the proposed algorithms significantly outperform DB algorithms by incorporating the qualitative feedback, and experimental results also demonstrate vast improvement over the existing DB algorithms. …
TextContourNet
We study the problem of extracting text instance contour information from images and use it to assist scene text detection. We propose a novel and effective framework for this and experimentally demonstrate that: (1) A CNN that can be effectively used to extract instance-level text contour from natural images. (2) The extracted contour information can be used for better scene text detection. We propose two ways for learning the contour task together with the scene text detection: (1) as an auxiliary task and (2) as multi-task cascade. Extensive experiments with different benchmark datasets demonstrate that both designs improve the performance of a state-of-the-art scene text detector and that a multi-task cascade design achieves the best performance. …
Like this:
Like Loading…
Related