With ever-improving information technologies and high performance computational power, recent techniques in granular computing, soft computing and cognitive science have allowed an increasing understanding of normal and abnormal brain functions, especially in the research of human's pattern recognition by means of computational intelligence. It is well understood that normal brains have high intelligence to recognize different geometrical patterns, but a systematic framework of biological neural network has not yet be established. In this paper, we propose the genetic granular cognitive fuzzy neural networks (GGCFNN) in order to efficiently testify artificial neural networks' learning capability on human's pattern recognition in term of symmetric and similar geometry patterns. In contrast to other information systems, the GGCFNN is a highly hybrid intelligent system integrating the techniques of genetic algorithms, granular computing, and fuzzy neural networks with cognitive science for pattern recognition. Our ability to simulate biological neural networks makes it possible a more comprehensive quantitative analysis on the pattern recognition of human brains, and our preliminary experiment results would shed lights on the future research of cognitive science and brain informatics.
Pooling under a softmax operation and Gaussian-like tuning in the form of a normalized dot-product were proposed as the key operations in a recent model of object recognition in the ventral stream of visual cortex. We investigate how these two operations might be implemented by plausible circuits of a few hundred neurons in cortex. We consider two different sets of circuits whose different properties may correspond to the conditions in visual and barrel cortices, respectively. They constitute a plausibility proof that stringent timing and accuracy constraints imposed by the neuroscience of object recognition can be satisfied with standard spiking and synaptic mechanisms. We provide simulations illustrating the performance of the circuits, and discuss the relevance of our work to neurophysiology as well as what bearing it may have on the search for maximum and tuning circuits in cortex.
This chapter reviews and examines two important research topics related to intelligent email processing, namely, email filtering and email-mediated applications. We present a framework to show a full process of email filtering. Within the framework, we suggest a new method of combining multiple filters and propose a novel filtering model based on ensemble learning. For email-mediated applications, we introduce the concept of operable email (OE). It is argued that operable email will play a fundamental role in future email systems, in order to meet the need of the World Wide Wisdom Web (W4). We demonstrate the use of OE in implementing an email assistant and other intelligent applications on the World Social Email Network (WSEN).
Intelligence concerns many aspects of human mental activity and is considered difficult to be defined clearly. Apart from its relation to mental activity, however, it is possible to discuss intelligence formally based on information it deals with. This paper defines intelligence as the capability to upgrading information and notes that there have been four important phases in the progress of intelligence. These are: (1) language acquisition, (2) knowledge discovery, (3) conceptualization and (4) granulation. These phases are discussed in this paper.
Conversational Informatics is a field of research that focuses on investigating human conversational behaviors and designing conversational artifacts that can interact with people in a conversational fashion. It is aimed at unveiling meaning creation and interpretation through sophisticated mechanisms of the verbal / nonverbal interactions during conversations. I overview major ideas and outcomes of Conversational Informatics and discuss the role of Conversational Informatics as a glue connecting Web Intelligence and Brain Informatics.
The researchers in DERI Innsbruck have been building an execution infrastructure for the Semantic Web Services (SWS) based on the Services Oriented Architecture (SOA) paradigm of loosely coupled components. While SOA is widely acknowledged for its potential to revolutionize the world of computing, that success depends on resolving several fundamental challenges, and especially in the case of open SOA environment the existing specifications do not address several issues. We aim in DERI Innsbruck to define a skeleton of the SWS system and implement the overall infrastructure with the aim of automating service discovery, negotiation, adaptation, composition, invocation, and monitoring as well as service interaction requiring data, protocol, and process mediation. We call this infrastructure a Semantically Enabled Service oriented Architecture (SESA). While there are already several specifications in the space for Web Services there are still elements missing, for example there is no specification describing how the particular components/services of the SWS infrastructure would work together. That work is carried out by DERI researchers in standardization bodies such as OASIS and W3C. In the near future a service-oriented world will consist of an uncountable number of services. Computation will involve services searching for services based on functional and non-functional requirements and an interoperating with those that they select. Services will not be able to interact automatically and SOAs will not scale without signification mechanization of a fixed set of components/services. Hence, machine processable semantics are critical for the next generation of computing, services and SOAs, to reach their full potential. The contribution of DERI Innsbruck is to define and implement the fixed set of services of an infrastructure that must be provided to enable a dynamic discovery, selection, mediation, invocation and inter-operation of the Semantic Web Services to facilitate the SOA revolution towards open environments. We recognize in DERI Innsburck that SOA outside of tightly controlled environment cannot succeed until/unless the semantics issues are addressed. Only with semantics can critical subtasks can be automated leaving humans to focus on higher level problems.
In this paper, a new method of supervised classification of documents is proposed. It utilizes discrete trasforms to extract features from classified objects and adopts adaptive potential active hypercontours (APAH) for document classification. The idea of APAH generalizes classic contour methods of image segmentation. It has two main advantages: it can use almost any knowledge during the search for an optimal classification function and it can operate in a feature space where only metric is defined. Here, both of them are utilized - the first one by using expert knowledge about significance of documents from training set and the second one by inducing new metrics in feature spaces. The method has been evaluated on the subset of open directory project (ODP) database and compared with k-NN, the well known classification technique.
Recent advances in computing, communications, digital storage technologies, and high-throughput data-acquisition technologies, make it possible to gather and store incredible volumes of data. It creates unprecedented opportunities for large-scale knowledge discovery from database. Data mining (DM) technology has emerged as a means of performing this discovery. It is a useful tool in many fields such as marketing, decision making, etc. There are countless researchers working on designing efficient data mining techniques, methods, and algorithms. Unfortunately, most data mining researchers pay much attention to technique problems for developing data mining models and methods, while little to basic issues of data mining. What is data mining? What is the product of a data mining process? What are we doing in a data mining process? What is the rule we should obey in a data mining process? In this paper, we will address these questions and propose our answers based on a conceptual data mining model. Our answer would be "data mining is a process of knowledge transformation". It is consistent with the process of human knowledge understanding. Based on analysis of the user-driven and "data-driven" data mining approaches proposed by many other researchers, a conceptual knowledge transformation model and a conceptual domain-oriented data-driven data mining (3DM) model are proposed. It integrates user-driven data mining and data-driven data mining into one system. Some future works for developing such a 3DM data mining system are proposed.
Neuroscience becomes a strategic growing field for both academic institutions and industrial companies because of their profound impact on human health, clinical therapy, and basic research such as brain informatics and cognitive science. To track activities of the rapidly developing field and observe new trends, neuroscientists must keep up-to-date with all the relevant information on the internet. In this paper, we present an ontology-based mining system, which is able to probe Competitive Intelligence in Neuroscience by using a well-defined ontology. It is able to support decision making by searching neuroscientific discoveries semantically and tracking new trends statistically. The experiments performed on 15,433,668 MEDLINE articles yield evidence for the feasibility and validity of our system.
Inductive reasoning is one of the most important higher level cognitive functions of the human brain, and we still know very little about its neural mechanism. In the present study, event-related potential (ERP) and event-related fMRI are used to explore the dynamic spatiotemporal characteristics of inductive reasoning process. We hypothesize that the process of numerical inductive reasoning is partially dissociable over time and space. A typical task of inductive reasoning, function-finding, was adopted. Induction tasks and calculation tasks were performed in the experiments, respectively. ERP results suggest that the time course of inductive reasoning process is partially dissociable as the following three sub-processes: number recognition (the posterior P100 and N200), strategy formation (P300) and hypothesis generation and verification (the positive slow waves). fMRI results show many activations, including prefrontal gyrus (BA 6), inferior parietal lobule (BA 7, 40), and occipital cortex (BA 18). After the respective discussions, the two kinds of data are combined qualitatively, then the dynamic spatiotemporal characteristic of inductive reasoning process are depicted using a conceptual figure. This study is a preliminary effort towards deeply understanding the dynamic information processing mechanism of human inductive reasoning process.
A novel web surfer model, where the transition probabilities are fuzzy quantities, is proposed in this article. Based on the theory of Fuzzy Markov Chains, we introduce FuzzRank, which is the counterpart of PageRank. Apart from discussing the theoretical aspects of fuzzy surfer models and FuzzRank, we have also compared its ranking, convergence and robustness properties with PageRank. Extensive experimental results and a detailed example depict the advantages of FuzzRank over PageRank.
Brain Informatics is a new interdisciplinary field of science, which studies the mechanisms of human information processing and, in some cases, also mechanisms causing the development of cognitive and other mental disturbances. Like other e-Science disciplines, it is accompanied with collection of large distributed datasets, which have to be efficiently managed, processed and analyzed. The Grid is an appropriate platform allowing solution of such ambitious tasks. This paper presents a Grid-based infrastructure called GridMiner providing the kernel functionality, like workflow management, visualization, data preprocessing, filtering, transformation and integration, data mining and data warehousing services, etc., supporting analytical tasks required by the Brain Informatics research and applications.
Acupuncture is a traditional Chinese healing technique, which is gaining popularity as an alternative and complementary therapeutic intervention in many worldwide countries. The acupoints are arranged on so-called meridians, which represent a network of channels each connected to a functional organic system. Our experiment is to investigate the mechanism of acupuncture at Liv3 (Taichong) and possible post-effect of acupuncture. Functional magnetic resonance imaging (fMRI) of the whole brain was performed in 18 healthy right-handed young volunteers during two stimulation paradigms: ten subjects received real acupuncture (RA) at acupoints rights Liv3 (on the hand) and other 8 subjects received sham acupuncture (SA) near . Liv3.fMRI data were analyzed using SPM99. Acupuncture at Liv3 resulted in activation of bilateral cerebella, prefrontal lobe (PF), superior parietal lobule (SPL) and inferior parietal lobule (IPL), occipital lobe, parahippocampal gyrus, insula, thalamus, lentiform nucleus; contralateral temporal pole and anterior cingulated gyrus (ACG), posterior cingulated gyrus (PCG). The PE of RA activated bilateral cerebella, PF, SPL, and IPL, occipital lobe, lentiform nucleus; isolateral temporal pole, hippocampus, insula and thalamus; contralateral head of nucleus caudate, corpus callosum, ACG and PCG. Acupuncture at Liv3 resulted in activation of visual area, limbic system and subcortical gray structures, which was considered as the specific central nervous response within the brain to acupuncture at Liv3. Moreover, the activation still existed during PE of RA. fMRI provides an objective evidence for post-effect existence, which will establish the foundations of later scientific design in acupuncture experiments.
To conduct data mining including mining on the web data, we often need to collect data from various parties. Privacy concerns may prevent the parties from directly sharing the data and some types of information about the data. How multiple parties collaboratively conduct data mining without breaching data privacy presents a challenge. In this paper, we present a solution for privacy-preserving k-Medoids clustering which is one of data mining tasks. The solution is based on the cryptography technology.
Organizations gain competitive advantages and benefits through e-Business Intelligence (e-BI) technologies at all levels of business operations. E-BI gathers, processes, and analyzes tremendous relevant data to help enterprises make better decisions. Data mining, which utilizes methods and tools from various fields to extract useful knowledge from large amount of data, provides significant support to e-BI/BI applications. This paper presents an overview of a data mining approach: Multiple Criteria Mathematical Programming (MCMP); describes a real-life application using MCMP; and explains how business users at different levels can benefit from the results of MCMP. Then three application models were presented for efficient implementation of e-BI/BI by MCMP models.
Memory is very vivid and long-lasting, particularly when we have strong emotions, such as pleasure, love, fun, surprise, fear, anxiety and anger. This fact leads us to imagine that memory storage in the hippocampus can be switched according to the emotion-related brain state. In this paper, we first review the two classical concepts for the hippocampal network. One is the trisynaptic circuit connecting the entorhinal cortex, the dentate gyrus, the CA3 field and the CA1 field. The other is the lamellar hypothesis which proposes that the lamellar organization of the trisynaptic circuit. We then summarize our studies highlighting a role of the hippocampal CA2 field in memory formation. Finally we propose a novel circuit in the hippocampal network, which contains the CA3-CA2-CA1 pathway in which the CA2 field functions as a gate controlled by the activity of the supramammillary nucleus to control signal propagation between hippocampal lamellar organizations.
We discus the Wisdom Granular Computing (WGC) as a basic methodology for Perception Based Computing (PBC). By wisdom, we understand an adaptive ability to make judgements correctly to a satisfactory degree (in particular, correct decisions) having in mind real-life constraints. We propose Rough-Granular Computing (RGC) as the basis for WGC.
Due to the uncertainty in accessing Web pages, analysis of Web logs faces some challenges. Several rough k-means cluster algorithms have been proposed and successfully applied to Web usage mining. However, they did not explain why rough approximations of these cluster algorithms were introduced. This paper analyzes the characteristics of the data in the boundary areas of clusters, and then a rough k-means cluster algorithm based on a reasonable rough approximation (RKMrra) is proposed. Finally RKMrra is applied to Web access logs. In the experiments RKMrra compares to Lingras and West algorithm and Peters algorithm with respect to five characteristics. The results show that RKMrra discovers meaningful clusters of Web users and its rough approximation is more reasonable.
We take it for granted that computers hold answers to our questions, our information requirements, our needs over the past twenty five years we have learned much about language, about databases, and about how people interact with computers; researchers have made great strides in the construction of human computer interfaces which (relatively) seamlessly integrate modalities, for example, speech and written language, natural language and menu systems, and so on. The next generation of interfaces and browsers, in order to be considered successful, must do more: they must individualize frameworks of meaning in order to provide relevant timely responses to information requests. I want to make several points, perhaps circuitously, but directed as examining some basic tenets regarding our faith in machines. I direct your attention to several problems inherent in representation(s) required to place information into machines for easy (individualized) access, followed up by some larger questions about the inherent capabilities of machines (versus humans).
