Ricardo Ribeiro Gudwin

Nonlinear context adaptation in the calibration of fuzzy sets

Abstract In this note we elaborate on the concept and use of context adaptation. The underlying idea hinges upon a nonlinear transformation of an actual reference unit universe of discourse into a subset of reals, say [a, b], that is implied by actually available data (current context). Assuming a collection of fuzzy sets A = {A1, A2, …, An} defined over [0, 1], the adaptation gives rise to a new frame of cognition A ′= {A1′, A2′, …, An′} expressed over [a,b]. Owing inherent nonlinearity of the developed mapping, different elements (fuzzy sets) of A can be “stretched” or “expanded” according to the given experimental data. Proposed is a neural network as a relevant optimization tool.

Artificial Cognition Systems

Emphasizing how cognitive processes can be meaningful to artificial systems, this text presents recent research efforts in artificial intelligence about building artificial systems capable of performing cognitive tasks.

A fuzzy elevator group controller with linear context adaptation

We introduce a fuzzy elevator group controller using a linear context adaptation technique. We first describe the elevator group control problem and the schemes usually employed to solve it. We detail the fuzzy controller used in our development and an example system used in simulation experiments. The focus is on the comparison between the standard fuzzy controller and the fuzzy controller with linear context adaptation. Simulation results are included to show the usefulness of the fuzzy control strategy suggested.

Context adaptation in fuzzy processing and genetic algorithms

In this paper we introduce the use of contextual transformation functions to adjust membership functions in fuzzy systems. We address both linear and nonlinear functions to perform linear or nonlinear context adaptation, respectively. The key issue is to encode knowledge in a standard frame of reference, and have its meaning tuned to the situation by means of an adequate transformation reflecting the influence of context in the interpretation of a concept. Linear context adaptation is simple and fast. Nonlinear context adaptation is more computationally expensive, but due to its nonlinear characteristic, different parts of base membership functions can be stretched or expanded to best fit the desired format. Here we use a genetic algorithm to find a nonlinear transformation function, given the base membership functions and a set of data extracted from the environment classified by means of fuzzy concepts.

Emergence of self-organized symbol-based communication in artificial creatures

In this paper, we describe a digital scenario where we simulated the emergence of self-organized symbol-based communication among artificial creatures inhabiting a virtual world of unpredictable predatory events. In our experiment, creatures are autonomous agents that learn symbolic relations in an unsupervised manner, with no explicit feedback, and are able to engage in dynamical and autonomous communicative interactions with other creatures, even simultaneously. In order to synthesize a behavioral ecology and infer the minimum organizational constraints for the design of our creatures, we examined the well-studied case of communication in vervet monkeys. Our results show that the creatures, assuming the role of sign users and learners, behave collectively as a complex adaptive system, where self-organized communicative interactions play a major role in the emergence of symbol-based communication. We also strive in this paper for a careful use of the theoretical concepts involved, including the concepts of symbol and emergence, and we make use of a multi-level model for explaining the emergence of symbols in semiotic systems as a basis for the interpretation of inter-level relationships in the semiotic processes we are studying.

Vox Populi: Evolutionary Computation for Music Evolution

Publisher Summary This chapter introduces a new system, Vox Populi , based on evolutionary computation, for composing music in real time. A population of chords is properly codified according to the MIDI protocol and evolves by the application of genetic algorithms. A fitness criterion is defined to indicate the best chord in each generation, and this chord is selected as the next element in the sequence to be played. Each new generated chord is a new sound palette that a musician can use to continue the music evolution. Graphic controls provide user-friendly manipulation of the fitness and of the sound attributes. Evolutionary computation is used to stimulate the user with novel sounds, and it allows the user to respond. Associating the dynamic behavior of genetic algorithms with these tools for real-time interaction, Vox Populi becomes a musical instrument. But unlike a traditional instrument, Vox Populi is able to create its own sound raw material and to provide choice criteria simultaneously. All these features enhance the users music capabilities and mark this system as state of the art in computer music. Next, a general description of the main components of the computational environment and melodic, harmonic, and voice range criteria for musical fitness are defined. The chapter also explains the genetic encoding of notes and the evolutionary cycle for chord production.

Text document classification using swarm intelligence

This paper presents an algorithm for the automatic grouping of PDF documents, and with potential application for Web document classification. The algorithm developed is based on an ant-clustering algorithm, which was inspired by the behavior of some ant species in the organization their nests. To apply the ant-clustering algorithm for text document classification, two modifications had to be introduced in the standard algorithm: 1) the use of a metric to evaluate the similarity degree of text data, instead of numeric data; and 2) the proposal of a cooling schedule for a user-defined parameter so as to improve the convergence properties of the algorithm. To illustrate the behavior of the modified algorithm, it was applied to sets of real-world documents taken from the IEEE WCCI -1998 CD.

Feedforward neural network initialization: an evolutionary approach

The initial set of weights to be used in supervised learning for multilayer neural networks has a strong influence in the learning speed and in the quality of the solution obtained after convergence. An inadequate initial choice of the weight values may cause the training process to get stuck in a poor local minimum or to face abnormal numerical problems. There are several proposed techniques that try to avoid both local minima and numerical instability, only by means of a proper definition of the initial set of weights. This paper focuses on the application of genetic algorithms (GA) as a tool to analyze the space of weights, in order to achieve good initial conditions for supervised learning. GAs almost-global sampling compliments connectionist local search techniques well, and allows one to find some very important characteristics in the initial set of weights for multilayer networks. The results presented are compared, for a set of benchmarks, with that produced by other approaches found in the literature.

Genetic algorithms and discrete event systems: an application

We propose an approach for discrete event systems control optimization, based on the theory developed by P.J. Ramadge and W.M. Wonham (1987; 1989) and on the limited lookahead policy strategy proposed by Sheng-Luen Chung and S. Lafortune (1992). By considering a performance index, i.e. a measure of how well a sequence of events meets its objectives, a genetic algorithm is derived to find optimal decisions for this class of systems. After introducing the theoretic background, an application example concerning the supervisory control of elevator systems is also included.<<ETX>>

Symbolic Communication in Artificial Creatures: An Experiment in Artificial Life

This is a project on Artificial Life where we simulate an ecosystem that allows cooperative interaction between agents, including intra-specific predator-warning communication in a virtual environment of predatory events. We propose, based on Peircean semiotics and informed by neuroethological constraints, an experiment to simulate the emergence of symbolic communication among artificial creatures. Here we describe the simulation environment and the creatures’ control architectures, and briefly present obtained results.