Dovey, K 1993, 'Organisational Form and People Development: The Team as a Vehicle for Developing the Individual-in-Community', British Journal of Guidance & Counselling, vol. 21, no. 2, pp. 124-132.
View/Download from: Publisher's site
View description>>
It is argued that all social action serves specific power interests and that the organisational form of social agencies is strongly influenced by the theoretical assumptions and practical aims of those who establish them. The paper presents the case of radical humanism, as an appropriate theory of social action within social democracies in the late-twentieth century, and argues that the team is a highly effective form of social organisation which leads to the establishment of an organisational culture compatible with radical humanist principles. © 1993, Taylor & Francis Group, LLC. All rights reserved.
Voinov, AA & Zharova, NA 1993, 'HYDRO: simulation of hydrodynamics and water pollution', Environmental Software, vol. 8, no. 4, pp. 209-218.
View/Download from: Publisher's site
View description>>
A user-friendly package for simulations of wind-induced currents and dispersion of non-conservative pollutants in aquatic media is discussed. The hydrodynamics are modelled by a stationary shallow-water approximation of the Eckman type. The generated patterns of currents are fed into the 2-D advection-diffusion model to calculate the concentration fields of a pollutant coming with inflows or injected directly into the water body. The package runs on IBM compatible PCs with a mathcoprocessor being very desirable. The package is simple to learn. It may be useful for preliminary qualitative analysis of water pollution, as well as for education and demonstration purposes. © 1993 Elsevier Science Publishers Ltd.
Lin, CT & Lee, CSG 1970, 'Reinforcement structure/parameter learning for neural-network-based fuzzy logic control systems', 1993 IEEE International Conference on Fuzzy Systems, pp. 88-93.
View description>>
This paper proposes a Reinforcement Neural-Network-Based Fuzzy Logic Control System (RNN-FLCS) for solving various reinforcement learning problems. The proposed RNN-FLCS is best applied to learning environments where obtaining exact training data is expensive. It is constructed by integrating two Neural-Network-Based Fuzzy Logic Controllers (NN-FLCs), each of which is a connectionist model with a feedforward multi-layered network developed for the realization of a fuzzy logic controller. One NN-FLC functions as a fuzzy predictor and the other as a fuzzy controller. Using the temporal difference prediction method, the fuzzy predictor can predict the external reinforcement signal and provide a more informative internal reinforcement signal to the fuzzy controller. The fuzzy controller performs a stochastic exploratory algorithm to adapt itself according to the internal reinforcement signal. During the learning process, the proposed RNN-FLCS can construct a fuzzy logic control system automatically and dynamically through a reward-penalty signal or through very simple fuzzy information feedback; both structure learning and parameter learning are performed simultaneously in the two NN-FLCs using the fuzzy similarity measure. Simulation results are presented to illustrate the performance and applicability of the proposed RNN-FLCS.
Lister, R 1970, 'Annealing networks and fractal landscapes', IEEE International Conference on Neural Networks, IEEE International Conference on Neural Networks, IEEE, pp. 257-262.
View/Download from: Publisher's site
View description>>
©1993 IEEE. The conventional explanation for the poor scaling of Hopfield and Tank networks is that they have difficulty in balancing the trade-off between the path length and the legality components of the energy function. We first describe an experiment which suggests that the conventional explanation is either wrong, or at best incomplete. We propose an alternative explanation: these networks might scale better if their dynamics effectively implemented a "divide-and-conquer" strategy. That is, if they recursively decomposed the problem into smaller independent sub-problems. An annealing network can do so if the energy landscape has a self-similar "quasi-fractal" structure. We believe this proposition applies to both discrete and analog networks. We support our proposition by describing our work on finding low cost solutions for traveling salesman problems. We then consider the implications for two other optimization problems: graph bisection and coloring.
Lister, R 1970, 'Visualizing weight dynamics in the N-2-N encoder', IEEE International Conference on Neural Networks, IEEE International Conference on Neural Networks, IEEE, pp. 684-689.
View/Download from: Publisher's site
View description>>
© 1993 IEEE. Kruglyak proved that sets of weights exist so that Multi-Layer Perceptrons can solve arbitrarily large N-2-N encoder problems. We extend Kruglyak's static geometric construction to give a way of visualizing weights dynamics during learning. This visualization provides insight as to why Back Propagation has difficulty in finding suitable N-2-N encoder weights for N>8. We believe this new insight has general consequences, relating to the danger of utilizing intermediate activity values in hidden units, and difficulties with finding solutions for tightly constrained (but not necessarily large) problems.
