Masanori Nakakuni

The Analysis of Pen Inputs of Handwritten Symbols using Self Organizing Maps and its Application to User Authentication

To realize ubiquitous computing, mobile computers such as PDAs and smart phones will be important components. However, the mobile computers tend to be used without authentication because they should be usable as soon as the power becomes on. In this paper, we propose an authentication method without losing the usability of PDA. With this method, the user is authenticated using the pen pressure pattern measured by tracing the symbols which are displayed on the screen with stylus. From the analyses of pen pressure patterns using self organizing maps, the symbols which are suitable for authentication are selected. 70% of the users can be authenticated by this method from the authentication experiments using the symbols.

An integration method of multi-modal biometrics using supervised pareto learning self organizing maps

This paper proposes a method for the integration of multi-modal biometrics. As the conventional authentication method, password system is mostly used. But, password mechanism has many issues. In order to solve the problems, biometric authentication methods are often used. But, the authentication method using biological characteristics, such as fingerprint, also has some problems. In this paper, we propose a authentication method using multi-modal behavior biometrics sampled from keystroke timings and handwritten patterns. And supervised Pareto learning self organizing maps which integrate the multi-modal vectors is proposed. The performance of this method is confirmed by the authentication experiments.

Application of Supervised Pareto Learning Self Organizing Maps and Its Incremental Learning

We have proposed Supervised Pareto Learning Self Organizing Maps(SP-SOM) based on the concept of Pareto optimality for the integration of multiple vectors and applied SP-SOM to the biometric authentication system which uses multiple behavior characteristics as feature vectors. In this paper, we examine performance of SP-SOM for the generic classification problem using iris data set. Furthermore, we propose the incremental learning algorithm for SP-SOM and examine effectiveness in a classification problem and adaptation ability to the change of the behavior biometric features by time.

The authentication system for multi-modal behavior biometrics using concurrent Pareto learning SOM

We have proposed the integration of behavior biometrics using Supervised Pareto learning SOM to improve the accuracy of authentication. For small systems such as mobile devices, this method may be heavy, because of the memory usage or computational power. In this paper, we propose the application of Concurrent Pareto learning SOM, which uses a small map for each user. The performance of this method is confirmed by authentication experiments using behavior biometrics of keystroke timings and key typing sounds.

Analysis of Robustness of Pareto Learning SOM to Variances of Input Vectors

We have proposed Supervised Pareto learning Self Organizing Maps (SP-SOM) which based on the concept of Pareto optimality for the integration of multiple vectors and applied SP-SOM to biometric authentication system which used multiple behavior characteristics as feature vectors. As a variant of SP-SOM, we have also proposed Supervised Full Pareto learning SOM (SFP-SOM) which processed each input attribute independently on SP-SOM. In this paper, we examine the robustness of SP-SOM and SFP-SOM to the variance of input vectors with the general classification problem using iris and abalone data set and biometric authentication problem using key typing features.

Quantitative measures to evaluate neural network weight initialization strategies

It has been reported numerous times in the neural network research literature that weight initialization in neural networks affects the learning rate, the convergence rate and the probability of correct classification. In this research paper we develop a theory for objectively testing various weight initialization strategies. Our theory provides a quantitative measure for each available weight initialization strategy. Thus for each initialization strategy and each epoch we estimate the conditional probability distribution function of correct classification given the epoch number. For each initialization strategy and for a given epoch the conditional probability is a random variable with certain probability distribution function and certain mean and variance. Based on multivariate analysis, statistics of extremes, analysis of variance and estimation theory we develop an objective framework and measurements to assess if one strategy is better than another or if the differences between strategies are not significant but they are due to random fluctuations.

Mapping of the 3D objects using computer generated hologram SOM

We propose the algorithm of CGH (Computer Generated Hologram)-SOM, in which SOM can organize the 3D information of objects on the map, using Fresnel hologram as memories of the units. The performance of CGH-SOM is also examined by experiments. Fresnel Hologram, which is conventionally implemented on photographic dry plates, can record the 3D information of an object, and can be used to recognize 3D objects. In the algorithm, we implemented Fresnel hologram as Computer Generated Hologram, which virtually simulates the photographic processing in the computer.

The adaptive authentication system for behavior biometrics using pareto learning self organizing maps

In this paper, we propose an authentication system which can adapt to the temporal changes of the behavior biometrics with accustoming to the system. We proposed the multi-modal authentication system using Supervised Pareto learning Self Organizing Maps. In this paper, the adaptive authentication system with incremental learning which is applied as the feature of neural networks is developed.

Low-bandwidth transmission algorithm for reliable wireless communication

In this research, we present a dynamic transmission algorithm that aims to provide reliable, low bandwidth communication via User Datagram Protocol (UDP) in such way that it is lighter and less rigid than Transmission Control Protocol (TCP), whereby, the degree of redundancy is a function of the noise and the probability for a bit to be corrupted. We also provide a variable number of protections depending on the importance of certain bits. In addition we provide a variable packet size depending on the noise, in order to decrease the probability of automatic repeat request. This algorithm is useful in providing lower latency communication through noisy mediums such as wireless communication.

Analysis of packet traffics and detection of abnormal traffics using Pareto learning self organizing maps

Recently, the spread of the Internet makes familiar to the incident concerning the Internet, such as a DoS attack and a DDoS attack. Some methods which detect the abnormal traffics in the network using the information from headers and payloads of IP-packets transmitted in the networks are proposed. In this research, we propose a method of Pareto Learning SOM (Self Organizing Map) for IP packet flow analysis in which the occurrence rate is used for SOM computing. The flow of the packets can be visualized on the map and it can be used for detecting the abnormal flows of packets.