Fuzzy ARTMAP: A Neural Network Architecture for Incremental Supervised Learning of Analog Multidimensional Maps

A new neural network architecture is introduced for incremental supervised learning of recognition categories and multidimensional maps in response to arbitrary sequences of analog or binary input vectors. The architecture, called Fuzzy ARTMAP, achieves a synthesis of fuzzy logic and Adaptive Resonance Theory (ART) neural networks by exploiting a close formal similarity between the computations of fuzzy subsethood and ART category choice, resonance, and learning. Fuzzy ARTMAP also realizes a new Minimax Learning Rule that conjointly minimizes predictive error and maximizes code compression, or generalization. This is achieved by a match tracking process that increases the ART vigilance parameter by the minimum amount needed to correct a predictive error. As a result, the system automatically learns a minimal number of recognition categories, or "hidden units", to met accuracy criteria. Category proliferation is prevented by normalizing input vectors at a preprocessing stage. A normalization procedure called complement coding leads to a symmetric theory in which the MIN operator (Λ) and the MAX operator (v) of fuzzy logic play complementary roles. Complement coding uses on-cells and off-cells to represent the input pattern, and preserves individual feature amplitudes while normalizing the total on-cell/off-cell vector. Learning is stable because all adaptive weights can only decrease in time. Decreasing weights correspond to increasing sizes of category "boxes". Smaller vigilance values lead to larger category boxes. Improved prediction is achieved by training the system several times using different orderings of the input set. This voting strategy can also be used to assign probability estimates to competing predictions given small, noisy, or incomplete training sets. Four classes of simulations illustrate Fuzzy ARTMAP performance as compared to benchmark back propagation and genetic algorithm systems. These simulations include (i) finding points inside vs. outside a circle; (ii) learning to tell two spirals apart; (iii) incremental approximation of a piecewise continuous function; and (iv) a letter recognition database. The Fuzzy ARTMAP system is also compared to Salzberg's NGE system and to Simpson's FMMC system.

Learning soft computing control strategies in a modular neural network architecture

Modelling and control of nonlinear dynamical systems is a challenging problem since the dynamics of such systems change over their parameter space. Conventional methodologies for designing nonlinear control laws, such as gain scheduling, are effective because the designer partitions the overall complex control into a number of simpler sub-tasks. This paper describes a new genetic algorithm based method for the design of a modular neural network (MNN) control architecture that learns such partitions of an overall complex control task. Here a chromosome represents both the structure and parameters of an individual neural network in the MNN controller and a hierarchical fuzzy approach is used to select the chromosomes required to accomplish a given control task. This new strategy is applied to the end-point tracking of a single-link flexible manipulator modelled from experimental data. Results show that the MNN controller is simple to design and produces superior performance compared to a single neural network (SNN) controller which is theoretically capable of achieving the desired trajectory. (C) 2003 Elsevier Ltd. All rights reserved.

Image indexing and retrieval using an ART-2A neural network architecture

Traditional content-based image retrieval (CBIR) systems use low-level features such as colors, shapes, and textures of images. Although, users make queries based on semantics, which are not easily related to such low-level characteristics. Recent works on CBIR confirm that researchers have been trying to map visual low-level characteristics and high-level semantics. The relation between low-level characteristics and image textual information has motivated this article which proposes a model for automatic classification and categorization of words associated to images. This proposal considers a self-organizing neural network architecture, which classifies textual information without previous learning. Experimental results compare the performance results of the text-based approach to an image retrieval system based on low-level features. (c) 2008 Wiley Periodicals, Inc.

Multi-level virtual ring : a foundation network architecture to support peer-to-peer application in wireless sensor network

Two main problems prevent the deployment of peer-to-peer application in a wireless sensor network: the index table, which should be distributed stored rather than uses a central server as the director; the unique node identifier, which cannot use the global addresses. This paper presents a multi-level virtual ring (MVR) structure to solve these two problems.

The index table in MVR is distributed stored by using the DHT technique. MVR is constructed decentralized and runs on mobile nodes themselves, requiring no central server or interruption. Naming system in MVR uses natural names rather than global addresses to identify sensor nodes. The MVR can route directly on the name identifiers of the sensor nodes without being aware the location. Some sensor nodes are selected as the backbone nodes by the backbone selection algorithm and are placed on the different levels of the virtual rings. MVR hashes nodes’ identifiers on the virtual ring, and stores them at the backbone nodes. Furthermore, MVR adopts cross-level routing to improve the routing efficiency.

Experiments using ns2 simulator for up to 200 nodes show that the storage and bandwidth requirements of MVR grow slowly with the size of the network. Furthermore, MVR has demonstrated as self-administrating, fault-tolerant, and resilient under the different workloads.

An FPGA network architecture for accelerating 3DES - CBC

This paper presents a DES/3DES core that will support cipher block chaining (CBC) and also has a built in keygen that together take up about 10% of the resources in a Xilinx Virtex II 1000-4. The core will achieve up to 200Mbit/s of encryption or decryption. Also presented is a network architecture that will allow these CBC capable 3DES cores to perform their processing in parallel.

Toward blueprints for network architecture, biophysical dynamics, and signal transduction

We review mathematical aspects of biophysical dynamics, signal transduction and network architecture that have been used to uncover functionally significant relations between the dynamics of single neurons and the networks they compose. We focus on examples that combine insights from these three areas to expand our understanding of systems neuroscience. These range from single neuron coding to models of decision making and electrosensory discrimination by networks and populations, as well as coincidence detection in pairs of dendrites and the dynamics of large networks of excitable dendritic spines. We conclude by describing some of the challenges that lie ahead as the applied mathematics community seeks to provide the tools that will ultimately underpin systems neuroscience.

Measuring QoS in an aeronautical opportunistic network architecture with limited access to a satellite communications backhaul

We measure quality of service (QoS) in a wireless network architecture of transoceanic aircraft. A distinguishing characteristic of the network scheme we analyze is that it mixes the concept of Delay Tolerant Networking (DTN) through the exploitation of opportunistic contacts, together with direct satellite access in a limited number of the nodes. We provide a graph sparsification technique for deriving a network model that satisfies the key properties of a real aeronautical opportunistic network while enabling scalable simulation. This reduced model allows us to analyze the impact regarding QoS of introducing Internet-like traffic in the form of outgoing data from passengers. Promoting QoS in DTNs is usually really challenging due to their long delays and scarce resources. The availability of satellite communication links offers a chance to provide an improved degree of service regarding a pure opportunistic approach, and therefore it needs to be properly measured and quantified. Our analysis focuses on several QoS indicators such as delivery time, delivery ratio, and bandwidth allocation fairness. Obtained results show significant improvements in all metric indicators regarding QoS, not usually achievable on the field of DTNs.

SAT-GRD: An ID/Loc split network architecture interconnecting satellite and ground networks

Since the satellite network plays an irreplaceable role in many fields, how to interconnect it with the ground network has received an unprecedented attention. However, with much more requirements imposed to the current terrestrial network, many serious problems caused by the IP dual-role exposed. In this context, their direct interconnection seems not the most appropriate way. Thus, in this paper, SAT-GRD, an incrementally deployable ID/Loc split network architecture is proposed, aiming to integrate the satellite and ground networks efficiently. Specifically, SAT-GRD separates the identity of both the host and network from the location. Then, it isolates the host from the network, and further divides the whole network into core and edge networks. These make SAT-GRD much more flexible and scalable to achieve heterogeneous network convergence and avoid problems resulting from the overloaded semantics of IP addresses. In addition, much work has been done to implement the proof-of-concept prototype of SAT-GRD, and experimental results prove its feasibility.

Architecture of macromolecular network of soft functional materials : from structure to function

An enhanced macromolecular nanofiber network and its implications have been developed by employing the understanding of its formation with an emphasis on its topological aspect. Using agarose aqueous solution as a typical example, the macromolecular nanofiber network of soft functional materials has been clearly visualized for the first time using the developed technique of field emission scanning electronic microscopy coupled with flash-freeze-drying. Both the systematic kinetic study and the image evidence indicates that the nanofiber network in soft functional materials such as agarose turns out to form through a self-expitaxial nucleation-controlled process. This new understanding enables us to engineer ultra functions of soft materials via nanofiber network architecture, which in turn opens up a new direction in nano fabrication.