https://people.dsv.su.se/~fk/id2010/articles/Jennings_1999.pdf

On agent-based software engineering

Lectures on Cauchy’s Problem in Linear Partial Differential Equations. By J. Hadamard. Pp. viii+316. 15s.net. 1923. (Per Oxford University Press.)

The impact of supply chain complexity on manufacturing plant performance

Sustainability science has developed from a new research field into a vibrant discipline in its own right, with scientific conferences, journals and scientific societies dedicated to its pursuit. Characterized more by its research purpose than by a common set of methods or objects, sustainability science can be subdivided into the more traditional disciplinary-based science for sustainability and the transdisciplinary science of sustainability. Whereas the former consists of more descriptive, analytical and basic science, the latter is characterized by reflexivity and applicability; on a meta level, the emergence of the latter can be understood as a new step in the evolution of science. This review provides an overview of the state of sustainability science, identifying action orientation, integrated assessments and interdisciplinarity as overall characteristics. The review also focuses on methodological issues, highlighting differences in project organization and management, and the ways in which stakeholder participation can be organized in interdisciplinary and transdisciplinary research projects. Sustainability science is recognized as essential for progress towards sustainability, and as an opportunity to bring science closer to the people, requiring significant changes in the way science is organized and conducted.

/pdf/sustainability-science-a-review-an-analysis-and-some-157ra5swjv.pdf

Sustainability science: a review, an analysis and some empirical lessons

The modernization of the US electric power infrastructure, especially in lieu of its aging, overstressed networks; shifts in social, energy and environmental policies, and also new vulnerabilities, is a national concern. Our system are required to be more adaptive and secure more than every before. Consumers are also demanding increased power quality and reliability of supply and delivery. As such, power industries, government and national laboratories and consortia have developed increased interest in what is now called the Smart Grid of the future. The paper outlines Smart Grid intelligent functions that advance interactions of agents such as telecommunication, control, and optimization to achieve adaptability, self-healing, efficiency and reliability of power systems. The author also presents a special case for the development of Dynamic Stochastic Optimal Power Flow (DSOPF) technology as a tool needed in Smart Grid design. The integration of DSOPF to achieve the design goals with advanced DMS capabilities are discussed herein. This reference paper also outlines research focus for developing next generation of advance tools for efficient and flexible power systems operation and control.

Smart grid design for efficient and flexible power networks operation and control

Dealing with Complexity

In the development of learning systems and neural networks, the issue of complexity occurs at many levels of analysis.

A Brain-Like Design to Learn Optimal Decision Strategies in Complex Environments

An important task in practical applications of neural networks is to design a network architecture. Network parameters are usually determined for a fixed architecture which requires us to solve a non-linear optimization problem in a multidimensional parameter space. An alternative approach is to use a dynamically allocated architecture and determine the final set of network parameters in a series of steps, each taking place in a lower dimensional space. There have been considered various types of such architecture dynamics in which network units or connections are either added or deleted. The simplest type is incremental architecture where in each step an architecture is extended by adding one new unit.

Incremental Approximation by Neural Networks

1 Recurrent Neural Networks: Some Systems-Theoretic Aspects.- 1 Introduction.- 2 System-Theory Results: Statements.- 3 System-Theory Results: Discussion.- 4 Computational Power.- 5 Some Remarks.- 2 The Use of State Space Control Theory for Analysing Feedforward Neural Networks.- 1 Introduction.- 2 State Space Theory.- 3 State Space Representation of Feedforward Neural Networks.- 4 Observability of Feedforward Neural Networks.- 5 Controllability.- 6 Stability.- 7 Discussion.- 8 Appendix: Linear Systems of Equations [7].- 3 Statistical Decision Making and Neural Networks.- 1 Introduction.- 2 Statistical Decision Making.- 3 Bayesian Learning.- 4 On Ingredients of Bayesian Learning.- 5 Interlude on Gaussian Linear Regression Model.- 6 Approximate On-Line Estimation.- 7 Conclusions.- 4 A Tutorial on the EM Algorithm and its Applications to Neural Network Learning.- 1 Introduction.- 2 The EM Algorithm.- 3 Practical Applications.- 4 Convergence Properties.- 5 Concluding Remarks.- 5 On the Effectiveness of Memory-Based Methods inMachine Learning.- 1 Introduction.- 2 Background.- 3 The Curse of Dimensionality.- 4 The Barron-Jones Theory.- 5 Experimental Results.- 6 Analysis of Memory-Based Methods.- 7 Discussion.- 6 A Study of Non Mean Square Error Criteria for the Training of Neural Networks.- 1 Introduction.- 2 Statement of the Problem.- 3 Cost Function Minimisation for ? = E(y/x).- 4 Cost Function Minimisation for the Median of p(y/x).- 5 Simulation Results.- 6 Conclusion.- 7 A Priori Information in Network Design.- 1 Introduction.- 2 Preliminaries.- 3 Recurrent Networks and Relative Order.- 4 Simulations.- 5 Conclusions.- 8 Neurofuzzy Systems Modelling: A Transparent Approach.- 1 Empirical Data Modelling.- 2 Neurofuzzy Construction Algorithms.- 3 Modelling Case Studies.- 4 Conclusions.- 9 Feature Selection and Classification by a Modified Model with Latent Structure.- 1 Introduction.- 2 Modified Model with Latent Structure.- 3 Optimizing Model Parameters.- 4 Approach to Feature Selection.- 5 Pseudo-Bayes Decision Rule.- 6 Experiments.- 7 Summary and Conclusion.- 10 Geometric Algebra Based Neural Networks.- 1 Introduction.- 2 Complex-Valued Neural Networks.- 3 Comments on the Applicability of CVNNs to n-Dimensional Signals.- 4 Generalisations of CVNNs Within a GA Framework.- 5 Summary.- 11 Discrete Event Complex Systems: Scheduling with Neural Networks.- 1 Introduction.- 2 The DNN Architecture.- 3 Continuous Time Control Law.- 4 Real-Time Scheduling.- 5 Simulation Results.- 6 Summary.- 12 Incremental Approximation by Neural Networks.- 1 Introduction.- 2 Approximation of Functions by One-Hidden-Layer Networks.- 3 Rates of Approximation of Incremental Approximants.- 4 Variation with Respect to a Set of Functions.- 5 Incremental Approximation by Perceptron and RBF Networks.- 6 Discussion.- 13 Approximation of Smooth Functions by Neural Networks.- 1 Introduction.- 2 Preliminaries.- 3 Complexity Theorems.- 4 Local Approximation.- 5 Some Open Problems.- 14 Rates of Approximation in a Feedforward Network Depend on the Type of Computational Unit.- 1 Introduction.- 2 Feedforward Networks with Various Computational Units.- 3 Discussion.- 15 Recent Results and Mathematical Methods for Functional Approximation by Neural Networks.- 1 Introduction.- 2 Individual vs Variable Context.- 3 Nonlinear Approximation.- 4 Feedforward Architectures.- 5 Lower Bounds on Rate of Approximation.- 6 Uniqueness of Approximation by Neural Networks.- 7 Other Approaches.- 16 Differential Neurocontrol of Multidimensional Systems.- 1 Introduction.- 2 Neurophysiological Basis.- 3 Scheme of the Differential Neurocontroller.- 4 Multiplicative Units.- 5 Feedback Block.- 6 Feedforward Block.- 7 Convergence of Learning.- 8 Computer Simulations.- 9 Conclusions.- 17 The Psychological Limits of Neural Computation.- 1 Neural Networks and Turing Machines.- 2 Function Approximation.- 3 Representation of Logical Functions Using Neural Networks.- 4 The Complexity of Learning in Neural Networks.- 5 Learning Logical Functions.- 6 The Optimization of Circuits.- 7 Final Remarks.- 18 A Brain-Like Design to Learn Optimal Decision Strategies in Complex Environments.- 1 Introduction.- 2 Time-Chunked Approximate Dynamic Programming.- 3 Temporal Chunking with Neural Networks.- 4 Spatial Chunking and Critical Subsystems.- 5 Adding the Third Brain.- Research Acknowledgements.

Dealing with complexity : a neural networks approach

Recurrent nets have been introduced in control, computation, signal processing, optimization, and associate memory applications. Given matrices A ∈ ℝ n ×n , B ∈ ℝ n ×m , C ∈ ℝ p ×n , as well as a fixed Lipschitz scalar function σ : ℝ → ℝ, the continuous time recurrent network Σ with activation function σ and weight matrices (A, B,C) is given by: 

$$\frac{{dx}}{{dt}}(t) = {\overrightarrow \sigma ^{(n)}}\left( {Ax(t) + Bu(t)} \right) , y\left( t \right) = Cx\left( t \right)$$

(1)

Vera Kůrková

Papers

Dealing with Complexity

A Brain-Like Design to Learn Optimal Decision Strategies in Complex Environments

Incremental Approximation by Neural Networks

Dealing with complexity : a neural networks approach

Recurrent Neural Networks: Some Systems-Theoretic Aspects