This study targets social login registrants on an EC site and aims to clarify the difference between the purchasing tendency of social login registrants and general members by analyzing product purchasing history. We focused on the golf portal site that is the subject of this research. We analyzed the purchasing data comparing social login registrants with general members. As the results, it became clear that social login registrants have less resistance to purchasing expensive products on an EC site compared with general members and golf clubs act as a bridge for purchasing.

This paper proposes a stock price prediction model, which extracts features from time series data and social networks for prediction of stock prices and evaluates its performance. In this research, we use the features such as numerical dynamics (frequency) of news and comments, overall sentiment analysis of news and comments, as well as technical analysis of historic price and volume. We model the stock price movements as a function of these input features and solve it as a regression problem in a Multiple Kernel Learning regression framework. Experimental results show that our proposed method outperforms other baseline methods in terms of magnitude prediction measures such as RMSE, MAE and MAPE for three famous Japan companies' stocks in US stock market. The results indicate that features other than mining from stock prices themselves improved the performance.

We propose a robot architecture to integrate symbolic and non-symbolic information processings. Artificial neural networks (ANN) are quick, flexible and robust. Symbolic processing is on the other hand comprehensible, effective, controllable, and consistent. To integrate symbolic and non-symbolic methods, we consider the relation between a robot and its environment as constraints. To describe and solve such constraints we turn to constraint logic programming (CLP). To construct a robot that works in the complex environment, CLP and ANN are integrated into a unified framework such that CLP evaluates the behavior candidates proposed by ANN according to the constraints and ANN learns adequate behavior according to evaluations by CLP. We implemented the decision process in our robot that drove through a test course as we expected.

We propose a stochastic learning algorithm for multilayer perceptrons of linear-threshold function units, which theoretically converges with probability one and experimentally exhibits 100% convergence rate and remarkable speed on parity and classification problems with typical generalization accuracy. For learning the n bit parity function with n hidden units, the algorithm converged on all the trials we tested (n=2 to 12) after 5.8 x 4.1(n) presentations for 0.23 x 4.0(n-6) seconds on a 533MHz Alpha 21164A chip on average, which is five to ten times faster than Levenberg-Marquardt algorithm with restarts. For a medium size classification problem known as Thyroid in UCI repository, the algorithm is faster in speed and comparative in generalization accuracy than the standard backpropagation and Levenberg-Marquardt algorithms.

An n-bit parallel binary adder consisting of NOR gates only in single-rail input logic is proved to require at least 17n + 1 connections for any value of n. Such an adder is proved to require at least 7n + 2 gates. An adder that attains these minimal values is shown. Also, it is concluded that some of the parallel adders with the minimum number of NOR gates derived by Lai and Muroga have the minimum number of connections as well as the minimum number of gates, except for the two modules for the two least significant bit positions.