The University of Osaka · 컴퓨터과학
Joe Suzuki 교수의 연구실은 정보 이론과 베이지안 학습을 기반으로 한 기계학습 및 통계적 추론 기법을 중심으로 연구를 전개합니다. 특히 최소 기술 길이(MDL) 원리에 기반한 베이지안 belief 네트워크 학습, 상호정보량의 베이지안 추정, 마르코프 체인 기반 유전자 알고리즘의 수렴 분석 등에서 핵심 기여를 하고 있습니다. 연구는 이론적 엄밀성과 실용적 적용성을 동시에 고려하여, 데이터 기반 모델 선택, 구조 학습, 독립성 검증 등에 응용됩니다. 특히 비모수적이고 일반적인 확률 변수에 적용 가능한 상호정보량 추정 기법 개발에도 기여하고 있습니다.
표시된 성과는 수집된 데이터 기준으로 산출되며, 일부 차이가 있을 수 있습니다.
This paper addresses a Markov chain analysis of genetic algorithms (GAs), in particular for a variety called a modified elitist strategy. The modified elitist strategy generates the current population of M individuals by reserving the individual with the highest fitness value from the previous generation and generating M-1 individuals through a generation change. The author's analysis is based on a Markov chain: by assuming a simple GA in which the genetic operation in the generation changes is
In this paper, the problem of learning a Bayesian belief network (BBN) from given examples based on the minimum description length (MDL) principle is addressed. Given examples, the learning algorithm based on the MDL principle computes for each network the total of description length of the network and that of the examples given the network, and finds a network with the minimum value. We provide a search algorithm that reduces the computation time and at the same time is sure to find the network
This paper addresses the problem of learning Bayesian belief networks (BBN) based on the minimum description length (MDL) principle. First, we give a formula of description length based on which the MDL-based procedure learns a BBN. Secondly, we point out that the difference between the MDL-based and Cooper and Herskovits procedures is essentially in the priors rather than in the approaches (MDL and Bayesian), and recommend a class of priors from which the formula is obtained. Finally, we show a
This paper considers model selection in classification. In many applications such as pattern recognition, probabilistic inference using a Bayesian network, prediction of the next in a sequence based on a Markov chain, the conditional probability P(Y=y|X=x) of class yisinY given attribute value xisinX is utilized. By model we mean the equivalence relation in X: for x,x'isinXx~x'hArrP(Y=y|X=x)=P(Y=y|X=x'), forall yisinY. By classification we mean the number of such equivalence classes is finite. W
Given data, not knowing the distribution, we wish to construct a forest (Markov graph) relative to which the description length is minimized, connecting edges with larger estimated mutual information of each pair of random variables step by step (the Chow-Liu algorithm) to balance simplicity of the forest and fitness of the data, where the random variables are not to be either discrete or continuous. To this end, we construct a Bayesian measure over the data sequences, and propose the Bayesian e
