Knowledge Processing System Using Chaotic Associative Memory

The chaotic neural network is studied under the threshold activated coupling, which provides a controlled output patterns. In general the chaotic neural network constructed with chaotic neurons exhibits very rich dynamic behaviors with a nonperiodic associative memory. In the chaotic neural network, it is difficult to distinguish the stored patterns from others, because of the chaotic states of output of the network. In order to apply the nonperiodic associative memory into information search and pattern recognition, etc, it is necessary to control chaos in this chaotic neural network. Chaos in the chaotic neural network is controlled with threshold activated coupling method and the network converges on one of its stored patterns or their reverses which has the smallest Hamming distance with the initial state of the network or provides a controlled neural network with spatio-temporal patterns depending upon the nature of control.

We review our model of chaotic neural networks with chaotic dynamics and apply it to associative memory. We demonstrate that dynamical assocative memory can be realized with the chaotic neural network.

In this paper, we propose an emotion generation system based on MaC model using neural networks. In the proposed system, the chaotic neural network and the Boltzmann machine are used in the emotion generator of the MaC model. In the Boltzmann machine, the plural pattern can be recalled stochastically for the same input. The proposed system makes use of this property in order to generate different emotions for same external input. On the other hand, the chaotic neural network can recall the training patterns dynamically based on external input and the history. The proposed system makes use of this property in order to generate emotions based on its history.

In this paper, we propose a chaotic neural network with time delay term (CNNTD). The proposed model is based on the conventional chaotic neural network and has two types of connection weights; (1) normal weights for hetero associations and (2) weights with time delay for auto associations. In the proposed model, dynamic associations between patterns belong to the same group in short term and dynamic associations between groups in long term are realized. We carried out a series of computer experiments and confirmed that the effectiveness of the proposed model.

An approach for measuring the cognitive complexity of visual display formats is presented. The approach involves modeling both the knowledge that can be extracted from a format and the knowledge an operator brings to a task. A semantic network formalism is developed to capture task-relevant knowledge, from which four orthogonal predictor measures of cognitive complexity are derived. In an experiment, seven different avionic missions, performed with the aid of a horizontal situation display, were studied, and three of the predictor measures were found to correlate significantly with observed task difficulty. The results indicate that a semantic network formalism can be used to produce an objective metric of format quality in terms of cognitive complexity.

Extending the expressive power of semantic networks

Effects of correlation among stored patterns on associative dynamics of chaotic neural network

Traditional theories viewed memory as the “great storehouse of information.” Since the late 1960s, researchers have begun thinking of information as stored in the form of semantic networks. The general idea was that concepts were stored in nodes and that the links between nodes indicated an association between the concepts. Several decades later, it became clear that there is a fundamental problem with this view. The problem is that the concepts are merely labels in the network, for example the label “whale” or “tree.” The network is merely a connection of linked labels. The labels have no meaning to the network; they only have meaning to the user. As Harnad (1990) put it, the network is parasitic on us. He dubbed this the grounding problem. The grounding problem suggests that semantic networks as just described cannot be a model of human memory. A great deal of research since the publication of Harnad’s influential paper has tried to solvethe grounding problem. Mental representations need to be grounded in perception and action; they cannot be a free-floating system of symbols. A research tradition has emerged that tries to investigate how cognition is grounded in perception and action. Sometimes this research tradition is called “embodied cognition” but others, like the authors of this chapter, prefer the term “grounded cognition.” The first publications in this emerging field addressed the question of how language proces-sing is grounded in action and perception (Glenberg and Kaschak, 2002; Tucker and Ellis, 1998; Zwaan, Stanfield, and Yaxley, 2002). These studies suggested that there are interactions between language and perception and action. Over the years, convergent evidence on these mutual interactions of action and perception with cognitive processes have given rise to a promising line of research that explored ways in which different domains of cognition are grounded in various action and perception patterns (Barsalou, Simmons, Barbey, and Wilson, 2003; Borghi and Cimatti, 2010; Price, Peterson, and Harmon-Jones, 2012). Theoretical discussions on the groundedness of cognition were initially of a general nature.They dealt with the intricate relations between body and brain (Damasio, 1994), perceptual symbol systems (Barsalou, 1999; Barsalou et al., 2003), views on embodied cognition (Glenberg, 1997; Wilson, 2002) and simulation (Gallese, 2003). What they have in common is a disagreement with the traditional view of cognition that is burdened with the grounding problem(Fodor, 1975, 1983). They also share a focus on the central role of the body in shaping the mind (Wilson, 2002). This chapter will review embodied cognition perspectives and examples of empirical researchwithin the context of a memory and action domain. Relevant insights on memory systems, the role of the body in cognitive processes, and neural substrates will be discussed here. This is followed by a more in-depth discussion of empirical research that relates to the memory and action theme with a focus on different motor domains, memory systems, and tasks involved. Expertise is addressed as a special form of action-cognition interaction and relates to memorybias issues that are the result of motor fluency. The chapter continues with an exploration of new directions in memory and action research in the psychology of economics, law and collective remembering, and concludes with the evaluation of boundary issues and limitations of the embodied cognition approach that have been brought up recently (Mahon and Caramazza, 2008).

The relative ineffectiveness of semantics-based multimedia indexing systems on the Web is caused by the semantic knowledge incompleteness and semantic heterogeneity problems. Nevertheless, the need to search multimedia documents with precision on the Web is persistently growing; pressing the demand for effective and efficient indexing strategies. In this article, we present an ontology-based multimedia indexing approach that cooperatively identifies the semantic and taxonomic relations that exist between annotation words that surround multimedia documents on Webpages. In this context, multiple ontologies are jointly employed for indexing each document. We construct inverted indexes in the form of semantic networks where nodes of each network are identified and added based on a majority-voting technique, while edges represent the semantic and taxonomic relations that hold between those nodes. To alleviate the heterogeneity between the resulting networks, we employ ontology merging algorithms to integrate them into consistent networks. We also utilise concept relatedness measures to enrich the networks with semantically-relevant entities that are not recognised by the used ontologies. To validate our proposal, we have developed a prototype system based on the proposed techniques. The produced results using real-world datasets demonstrate an improvement of the effectiveness against state-of-the-art baseline metrics.

This paper presents a work on generating Bangla sentences from an interlingua representation called Universal Networking Language (UNL). UNL represents knowledge in the form of semantic network like hyper-graphs which contains disambiguated words, binary semantic relations, and speech act like attributes associated with the words, assisted by the semantically rich lexicon and a set of analysis and generation rules. We have developed a set of generation rules for converting UNL expression to Bangla sentences. Our experiment shows that these rules successfully generate correct Bangla sentences from UNL expressions.KeywordsInterlinguaUniversal Networking LanguageUniversal WordMorphologyDeConverterGeneration Rules

Ontology represents domain concepts and relations in a form of semantic network. Different from the traditional flat structured feature vectors, ontology provides more information through concept definition and relation inference. Many research works use ontologies in information matchmaking and retrieval. This trend is further accelerated by the convergence of various information sources supported by ontologies in the multimedia research field. Efforts have been made to shift traditional content-based retrieval approaches toward concept-based approaches. Both text and image features are extracted and integrated to improve the classification and retrieval performance. However, there are many open issues on the ontology understanding, definition, construction, utilization and implementation. The extra work required by ontology based approaches becomes one of the major difficulties that hedge against the development.

Ontologies represent domain concepts and relations in a form of semantic network. Many research works use ontologies in the information matchmaking and retrieval. This trend is further accelerated by the convergence of various information sources supported by ontologies. In this paper, we propose a novel multi-modality ontology model that integrates both the low-level image features and the high-level text information to represent image contents for image retrieval. By embedding this ontology into an image retrieval system, we are able to realize intelligent image retrieval with high precision. Moreover, benefiting from the soft-coded ontology model, this system has good flexibility and can be easily extended to the larger domains. Currently, our experiment is conducted on the animal domain canine. An ontology has been built based on the low-level features and the domain knowledge of canine. A prototype retrieval system is set up to assess the performance. We compare our experiment results with traditional text-based image search engine and prove the advantages of our approach.

In this paper, we proposed an improved delay feedback control method (IDFC) for a chaotic neural network. In the method, a delay feedback control signal is added into the term of the refractoriness of the chaotic neuron to resist the chaos in the chaotic neural network. The computer experiments show that the output sequence of the controlled chaotic neural network become periodic. The controlling chaos in a chaotic neural network is therefore implemented. When control parameters K and tau are taken as 1.3 and 1 respectively, the outputs of the network is periodic-16

In this paper, we propose a chaotic associative memory for successive learning (CAMSL) using internal patterns. In the CAMSL, the learning process and the recall process are not divided. When an unstored pattern is given to the network, the CAMSL can learn the pattern successively. The CAMSL distinguishes an unstored pattern from the stored patterns. When a stored pattern is given, the CAMSL recalls the pattern. When an unstored pattern is given, the CAMSL changes the internal pattern for the input pattern by chaos and presents the other pattern candidates. When the CAMSL cannot recall the desired pattern, it learns the input pattern as an unstored pattern. We carried out a series of computer simulations and confirmed the effectiveness of the CAMSL.

This paper proposes a sub-optimal multiuser detector (MUD) algorithm for CDMA system based on the neural network with a novel wavelet chaotic simulated annealing neural network (WCSANN), and gives a concrete model of the MUD after appropriate transformations and mappings. The WCSANN makes use of the wavelet and chaotic simulated annealing parameters of the recurrent neural network to control the network evolving behavior so that the network has richer and more flexible dynamics rather than conventional neural networks, so that it can be expected to have much powerful ability to search for globally optimal or sub-optimal solutions, and can refrain from the serious local optimal problem of Hopfield-type neural networks. Simulation experiments have been performed to show the effectiveness and validation of the proposed method for MUD problem.