We present a technique for the automated verification of abstract models of multithreaded programs providing fresh name generation, name mobility, and unbounded control. As high level specification language we adopt here an extension of communication finite-state machines with local variables ranging over an infinite name domain, called TDL programs. Communication machines have been proved very effective for representing communication protocols as well as for representing abstractions of multithreaded software. The verification method that we propose is based on the encoding of TDL programs into a low level language based on multiset rewriting and constraints that can be viewed as an extension of Petri Nets. By means of this encoding, the symbolic verification procedure developed for the low level language in our previous work can now be applied to TDL programs. Furthermore, the encoding allows us to isolate a decidable class of verification problems for TDL programs that still provide fresh name generation, name mobility, and unbounded control. Our syntactic restrictions are in fact defined on the internal structure of threads: In order to obtain a complete and terminating method, threads are only allowed to have at most one local variable (ranging over an infinite domain of names).

In this paper we propose some new measures of language development using network analyses, which is inspired by the recent surge of interests in network studies of many real-world systems. Children's and care-takers' speech data from a longitudinal study are represented as a series of networks, word forms being taken as nodes and collocation of words as links. Measures on the properties of the networks, such as size, connectivity, hub and authority analyses, etc., allow us to make quantitative comparison so as to reveal different paths of development. For example, the asynchrony of development in network size and average degree suggests that children cannot be simply classified as early talkers or late talkers by one or two measures. Children follow different paths in a multi-dimensional space. They may develop faster in one dimension but slower in another dimension. The network approach requires little preprocessing of words and analyses on sentence structures, and the characteristics of words and their usage emerge from the network and are independent of any grammatical presumptions. We show that the change of the two articles "the" and "a" in their roles as important nodes in the network reflects the progress of children's syntactic development: the two articles often start in children's networks as hubs and later shift to authorities, while they are authorities constantly in the adult's networks. The network analyses provide a new approach to study language development, and at the same time language development also presents a rich area for network theories to explore.

In this paper, a mathematical schema theory is developed. This theory has three roots: brain theory schemas, grid automata, and block-shemas. In Section 2 of this paper, elements of the theory of grid automata necessary for the mathematical schema theory are presented. In Section 3, elements of brain theory necessary for the mathematical schema theory are presented. In Section 4, other types of schemas are considered. In Section 5, the mathematical schema theory is developed. The achieved level of schema representation allows one to model by mathematical tools virtually any type of schemas considered before, including schemas in neurophisiology, psychology, computer science, Internet technology, databases, logic, and mathematics.

Researchers in the field of biocomputing have, for many years, successfully "harvested and exploited" the natural world for inspiration in developing systems that are robust, adaptable and capable of generating novel and even "creative" solutions to human-defined problems. However, in this position paper we argue that the time has now come for a reassessment of how we exploit biology to generate new computational systems. Previous solutions (the "first generation" of biocomputing techniques), whilst reasonably effective, are crude analogues of actual biological systems. We believe that a new, inherently inter-disciplinary approach is needed for the development of the emerging "second generation" of bio-inspired methods. This new modus operandi will require much closer interaction between the engineering and life sciences communities, as well as a bidirectional flow of concepts, applications and expertise. We support our argument by examining, in this new light, three existing areas of biocomputing (genetic programming, artificial immune systems and evolvable hardware), as well as an emerging area (natural genetic engineering) which may provide useful pointers as to the way forward.

We consider the problem of on-line prediction competitive with a benchmark class of continuous but highly irregular prediction rules. It is known that if the benchmark class is a reproducing kernel Hilbert space, there exists a prediction algorithm whose average loss over the first N examples does not exceed the average loss of any prediction rule in the class plus a "regret term" of O(N^(-1/2)). The elements of some natural benchmark classes, however, are so irregular that these classes are not Hilbert spaces. In this paper we develop Banach-space methods to construct a prediction algorithm with a regret term of O(N^(-1/p)), where p is in [2,infty) and p-2 reflects the degree to which the benchmark class fails to be a Hilbert space.

We establish a relationship between the online mistake-bound model of learning and resource-bounded dimension. This connection is combined with the Winnow algorithm to obtain new results about the density of hard sets under adaptive reductions. This improves previous work of Fu (1995) and Lutz and Zhao (2000), and solves one of Lutz and Mayordomo's "Twelve Problems in Resource-Bounded Measure" (1999).

The main theme of this workshop (Dagstuhl seminar 04351) is `Spatial Representation: Continuous vs. Discrete'. Spatial representation has two contrasting but interacting aspects (i) representation of spaces' and (ii) representation by spaces. In this paper, we will examine two aspects that are common to both interpretations of the theme, namely nerve constructions and refinement. Representations change, data changes, spaces change. We will examine the possibility of a `differential geometry' of spatial representations of both types, and in the sequel give an algebra of differential forms that has the potential to handle the dynamical aspect of such a geometry. We will discuss briefly a conjectured class of spaces, generalising the Cantor set which would seem ideal as a test-bed for the set of tools we are developing.

Artificial life models, swarm intelligent and evolutionary computation algorithms are usually built on fixed size populations. Some studies indicate however that varying the population size can increase the adaptability of these systems and their capability to react to changing environments. In this paper we present an extended model of an artificial ant colony system designed to evolve on digital image habitats. We will show that the present swarm can adapt the size of the population according to the type of image on which it is evolving and reacting faster to changing images, thus converging more rapidly to the new desired regions, regulating the number of his image foraging agents. Finally, we will show evidences that the model can be associated with the Mathematical Morphology Watershed algorithm to improve the segmentation of digital grey-scale images. KEYWORDS: Swarm Intelligence, Perception and Image Processing, Pattern Recognition, Mathematical Morphology, Social Cognitive Maps, Social Foraging, Self-Organization, Distributed Search.

We propose a new method for the estimation of parameters of hidden diffusion processes. Based on parametrization of the transition matrix, the Baum-Welch algorithm is improved. The algorithm is compared to the particle filter in application to the noisy periodic systems. It is shown that the modified Baum-Welch algorithm is capable of estimating the system parameters with better accuracy than particle filters.

From a geometric perspective most nonlinear binary classification algorithms, including state of the art versions of Support Vector Machine (SVM) and Radial Basis Function Network (RBFN) classifiers, and are based on the idea of reconstructing indicator functions. We propose instead to use reconstruction of the signed distance function (SDF) as a basis for binary classification. We discuss properties of the signed distance function that can be exploited in classification algorithms. We develop simple versions of such classifiers and test them on several linear and nonlinear problems. On linear tests accuracy of the new algorithm exceeds that of standard SVM methods, with an average of 50% fewer misclassifications. Performance of the new methods also matches or exceeds that of standard methods on several nonlinear problems including classification of benchmark diagnostic micro-array data sets.