sdyna is a general framework designed to address large stochastic reinforcement learning (rl) problems. Unlike previous model-based methods in Factored mdps (fmdps), it incrementally learns the structure of a rl problem using supervised learning techniques. spiti is an instantiation of sdyna that uses decision trees as factored representations. First, we show that, in structured rl problems, sp...

Institute of Child Welfare, University of Minnesota The Wallin series of performance tests designed for use with very young children consists of four boards, each measuring 3 by 14 inches, % of an inch thick. Each board contains recesses for 6 pegs, similar in size but varying as to form. Board A includes only round pegs, Board B only square pegs, in Board C three round and three square pegs ar...

Noncentral chi-square distributions play an important role in statistics. They occur for instance as exact or limiting distributions of test statistics under alternatives. To examine the power functions of such tests we come across the problem of comparing distribution functions of noncentral chi-square variables. In this section inequalities are presented, which make such a comparison possible...

In this paper we consider the problem of change detection and non stationary signals tracking. Using parametric estimation of signals based on least square lattice adaptive filters we consider for change detection statistical parametric methods using likelihood ratio and hypothesis tests. In order to track signals dynamics, we introduce a compensation procedure in the adaptive estimation. This ...

A version of the fundamental mean-square convergence theorem is proved for stochastic differential equations (SDEs) in which coefficients are allowed to grow polynomially at infinity and which satisfy a one-sided Lipschitz condition. The theorem is illustrated on a number of particular numerical methods, including a special balanced scheme and fully implicit methods. The proposed special balanc...

If A is unitary then A = eiH = cos H + i sin H where H is Hermitian. We examine the problem of approximating H . The standard approach to approximating logarithms is to take successive square roots until A1/2 n is close to the identity and then apply a Padé approximation to log A = 2n log A1/2 n . For the unitary problem we have A1/2 n = cos(H/2n) + i sin(H/2n) which suggests the use of halfang...

The exchange of light pseudoscalars between fermions leads to a spinindependent potential in order g4, where g is the Yukawa pseudoscalar-fermion coupling constant. This potential gives rise to detectable violations of both the weak equivalence principle (WEP) and the gravitational inverse-square law (ISL), even if g is quite small. We show that when previously derived WEP constraints are combi...