Whilst cases have been regularly used in building legal case based reasoners, they have rarely been used as a means of automated discovery of legal knowledge. Significant obstacles must be overcome if knowledge discov...
详细信息
Whilst cases have been regularly used in building legal case based reasoners, they have rarely been used as a means of automated discovery of legal knowledge. Significant obstacles must be overcome if knowledge discovery techniques are to be applied in the legal domain. We argue that the use of domain expertise is vital and that an abundance of commonplace cases is necessary. Even with appropriate data, knowledge discovery techniques in law must deal with contradictory cases and use statistical techniques in order to define error and estimate performance. We illustrate these points by describing the use of the cross validation resampling technique, our own error heuristic and the method we use for dealing with contradictions for the training of neural networks in the domain of property proceedings in Australian Family Law.
We propose a general approach to the electron-impact ionization problem that involves two distinct steps. First, using exterior complex scaling, we calculate the outgoing portion of the scattering wave function withou...
We propose a general approach to the electron-impact ionization problem that involves two distinct steps. First, using exterior complex scaling, we calculate the outgoing portion of the scattering wave function without explicit use of boundary conditions for ionization or any other channel. Once that wave function is known for a finite region of space, we make use of a general formula for extracting the integral cross section for the breakup process from any wave function. That formula is a projected form of the optical theorem and has no relation to complex coordinates or other methods of computing the scattering wave function. A preliminary calculation on the s-wave radial limit form of electron–hydrogen-atom ionization is presented in excellent agreement with calculations by other approaches.
We describe a variational method for calculating photoionization cross sections from a matrix element of the resolvent operator. Scattering boundary conditions are enforced by expanding the resolvent in a basis of squ...
We describe a variational method for calculating photoionization cross sections from a matrix element of the resolvent operator. Scattering boundary conditions are enforced by expanding the resolvent in a basis of square-integrable (L2 ) and outgoing wave continuum basis functions. By employing several continuum basis functions, with overlaps defined by a suitable analytic continuation, we can, in a single calculation, express the cross section over a continuous range of energies without explicitly resolving the variational equations at each desired energy. The method is illustrated by calculation of the photoionization cross section of atomic Be in the autoionizing region between the 1s2 2s and 1s2 2p states of Be+ .
Exact inference in densely connected Bayesian networks is computationally intractable, and so there is considerable interest in developing effective approximation schemes. One approach which has been adopted is to bou...
Exact inference in densely connected Bayesian networks is computationally intractable, and so there is considerable interest in developing effective approximation schemes. One approach which has been adopted is to bound the log likelihood using a mean-field approximating distribution. While this leads to a tractable algorithm, the mean field distribution is assumed to be factorial and hence unimodal. In this paper we demonstrate the feasibility of using a richer class of approximating distributions based on mixtures of mean field distributions. We derive an efficient algorithm for updating the mixture parameters and apply it to the problem of learning in sigmoid belief networks. Our results demonstrate a systematic improvement over simple mean field theory as the number of mixture components is increased.
The complex Kohn variational method is used to study low-energy electron collision processes involving CH3Cl. The elastic differential and momentum transfer cross sections are studied from low electron energy (0.5 eV)...
The complex Kohn variational method is used to study low-energy electron collision processes involving CH3Cl. The elastic differential and momentum transfer cross sections are studied from low electron energy (0.5 eV) through the resonance region (3.5 eV) to 10 eV. The effects of target correlation and polarization are studied. A comparison is made between the results of this study and experimental results available for this system.
This paper presents and investigates the application of multi-level planning techniques, together with meta and micro level knowledge architecture model, in building the domain knowledge, planning and navigating the c...
详细信息
ISBN:
(纸本)0780342534
This paper presents and investigates the application of multi-level planning techniques, together with meta and micro level knowledge architecture model, in building the domain knowledge, planning and navigating the curriculum of an intelligent tutoring system (ITS). Curriculum issues have been important to ITS. In order to truly individualize instruction, ITS must be able to reason about the curriculum, understand its implications, and be able to dynamically redesign the curriculum. This leads to the dramatic increase in the information a planner handles. The multi-level planning, made possible by the meta and micro level knowledge model, efficiently manages large domain knowledge base, in that it largely eases the planning task both in complexity and in storage.
作者：
Houston, S. KennethKen Houston was educated at Queen's University
Belfast where he gained BSc (Hons) and PhD degrees. He is Professor of Mathematical Studies in the School of Computing and Mathematics at the University of Ulster. He is interested in the teaching learning and assessment of Mathematical Modelling and all else pertaining to introducing students to ‘the way of life of an applied mathematician’. This includes the development of ‘enterprise’ competencies the use of Information Technology and innovative teaching strategies like peer tutoring. He is editor of “Innovations in Mathematics Teaching’ SEDA paper 87 1994 and co-author of ‘Mathematical Modelling’ with John Berry
It is suggested that "mathematical modelling is a way of life" for everyone. The M's of mathematics—methods, models and modelling, and the R's of mathematics—reasoning, reading, 'riting and ...
One of the most important problems in linear multivariable control theory is that of controlling a given plant in order to have its output track (or reject) a reference (or disturbance) signal produced by some externa...
详细信息
One of the most important problems in linear multivariable control theory is that of controlling a given plant in order to have its output track (or reject) a reference (or disturbance) signal produced by some external generator, called the exosystem. The disturbance or reference contains only the frequency components of the exosystem. This is called the output regulation problem. We generalize the problem framework in several directions. One is to model disturbances and references by an exosystem that is not, unlike the traditional approach, autonomous but is driven. Another direction is to incorporate the derivative information in the problem formulation. Consequently, we formulate two classes of generalized exact and almost output regulation problems. We provide solvability conditions and the solutions to these problems. Finally, and most importantly, we show that if solvability conditions for regulation are not satisfied, one can design a regulator to achieve "best" tracking (or rejection) of reference (or disturbance) signal is any desired induced norm sense.
We present a method for approximating the invariant measure of a randomly perturbed mapping S of R(d). Cases where the unperturbed mapping is singular are considered, as are cases where the sizes of random jumps are u...
We present a method for approximating the invariant measure of a randomly perturbed mapping S of R(d). Cases where the unperturbed mapping is singular are considered, as are cases where the sizes of random jumps are unbounded. Existence of an invariant measure and convergence of the scheme are proved when the mapping has strong contraction properties. The implementation is designed to be useful in experimental situations where the map is known only approximately and the distribution of the noise is unknown. Under the same contraction conditions, we also prove that the invariant measure, the stationary measure of the associated Markov chain, is approached exponentially under iteration of the perturbed map, independent of the starting point.
Solutions of the unsteady two-dimensional Navier-Stokes equations and boundary-layer equations are considered for the flow induced by a thick-core vortex convecting along a surface in an incompressible flow. Vortex-in...
详细信息
暂无评论