Database Integration Using Neural Networks: Implementation and Experiences

Applications in a wide variety of industries require access to multiple heterogeneous distributed databases. One step in heterogeneous database integration is semantic integration: identifying corresponding attributes in different databases that represent the same real world concept. The rules of semantic integration can not be ‘pre-programmed’ since the information to be accessed is heterogeneous and attribute correspondences could be fuzzy. Manually comparing all possible pairs of attributes is an unreasonably large task. We have applied artificial neural networks (ANNs) to this problem. Metadata describing attributes is automatically extracted from a database to represent their ‘signatures’. The metadata is used to train neural networks to find similar patterns of metadata describing corresponding attributes from other databases. In our system, the rules to determine corresponding attributes are discovered through machine learning. This paper describes how we applied neural network techniques in a database integration problem and how we represent an attribute with its metadata as discriminators. This paper focuses on our experiments on effectiveness of neural networks and each discriminator. We also discuss difficulties of using neural networks for this problem and our wish list for the Machine Learning community. Received 18 February 1999 / Revised 22 April 1999 / Accepted in revised form 20 November 1999     

磨粒图象纹理分形特征的研究

机械设备在运转过程中,发生各种各样形式的摩擦磨损,其磨损颗粒的数量、尺寸大小都随着运转过程而逐渐增加,在颗粒表面留下了与摩擦磨损机理相关的痕迹。这对磨粒的表面纹理特征进行分析,通过求取特征值,作为判断出机械装置的寿命、故障机理等指标的重要参数。     

On-line tool wear monitoring in turning using neural networks

The on-line supervision of a tool's wear is the most difficult task in the context of tool monitoring. Based on an in-process acquisition of signals with multi-sensor systems, it is possible to estimate or classify wear parameters by means of neural networks. This article demonstrates that solutions can be improved significantly by using available secondary information about physical models of the cutting process and about the temporal development of wear. Process models describing the influence of process parameters are used for a dedicated pre-processing of the sensor signals. The essential signal behaviour in a certain time window is described by means of polynomial coefficients. These coefficients are used as inputs for feedforward networks considering the temporal development of wear (multilayer perceptrons with a sliding window technique and time-delay neural networks). With a combination of the proposed measures it is possible to obtain remarkable improvements of both tool wear estimation and classification.     

Tool condition classification using Hidden Markov Model based on fractal analysis of machined surface textures

The texture of a machined surface generated by a cutting tool, with geometrically well-defined cutting edges, carries essential information regarding the extent of tool wear. There is a strong relationship between the degree of wear of the cutting tool and the geometry imparted by the tool on to the workpiece surface. The monitoring of a tool’s condition in production environments can easily be accomplished by analyzing the surface texture and how it is altered by a cutting edge experiencing progressive wear and micro-fractures. This paper discusses our work which involves fractal analysis of the texture of surfaces that have been subjected to machining operations. Two characteristics of the texture, high directionality and self-affinity, are dealt with by extracting the fractal features from images of surfaces machined with tools with different levels of tool wear. The Hidden Markov Model is used to classify the various states of tool wear. In this paper, we show that fractal features are closely related to tool condition and HMM-based analysis provides reliable means of tool condition prediction.     

Tool wear monitoring using a fast Hough transform of images of machined surfaces

The texture of machined surfaces provides reliable information regarding the extent of tool wear. In this paper, we propose a structure-based approach to analyzing machined surfaces. The original surface images are first preprocessed by a Canny edge detector. A new connectivity-oriented fast Hough transform is then applied to the edge image to detect all the line segments. The distributions of the orientations and lengths of the line segments are used to determine tool wear. Through our experiments, we found a strong correlation between tool wear and features. The computational complexity of the fast Hough transform is also analyzed.Received: 6 November 2002, Accepted: 18 December 2003, Published online: 13 May 2004 Correspondence to: A.A. Kassim     

Evolutionary multi-feature construction for data reduction: A case study

Real-world data are often prepared for purposes other than data mining and machine learning and, therefore, are represented by primitive attributes. When data representation is primitive, preprocessing data before looking for patterns becomes necessary. The low-level primitive representation of real-world problems facilitates the existence of complex interactions among attributes. If lack of domain experts prevents traditional methods to uncover patterns in data due to complex attribute interactions, then the use of soft computing techniques such as genetic algorithms becomes necessary. This article introduces MFE3/GA^DR, a data reduction method derived from the learning preprocessing system MFE3/GA. The method restructures the primitive data representation by capturing and compacting hidden information into new features in order to highlight regularities to the learner. We thoroughly analyze the empirical results obtained on the poker hand data set. The results show that this approach successfully compacts the set of low-level primitive attributes into a smaller set of highly informative features which outline patterns to the learner; thus, the new approach provides data reduction and yields learning a smaller and more accurate classifier.     

Collision: Modeling,simulation and identification of robotic manipulators interacting with environments

The performance of many robotic tasks depends greatly on their dynamic collision behavior. This article presents a simple method for modeling and simulating collision behavior in manipulators. The main goal in this task is to provide informative contact models. The proposed models encompasscollision attributes which comprise not only (local) contact surface properties but also structural properties of the environmental object and the manipulator. With this method, the entire dynamic and interactive motion of the manipulator with the environmental object can be simulated effectively. This is verified by our simulation results. To facilitate our investigation, a 2 DOF planar elbow manipulator with PD control is considered in the simulations as well as theoretical analysis. The simulation results are used to highlight the collision attributes which affect collision behavior and to study the effects of these attributes on the manipulator-work environment safety and performance. On the other hand, the reliable operation of intelligent robotic systems in unstructured environments requires the estimation of collision attributes before the prediction of the collision behavior can be completed. For this purpose, we introduce the notion ofcollision identification. The present paper introduces a framework for collision identification in robotic tasks. The proposed framework is based on Artificial Neural Networks (ANNs) and provides fast and relatively reliable identification of the collision attributes. The simulation results are used to generate training data for the set of ANNs. A modularized ANN-based architecture is also developed to reduce the training effort and to increase the accuracy of ANNs. The test results indicate the satisfactory performance of the proposed collision identification system.     

Reconstruction of Patterns from Noisy Inputs Using Morphological Associative Memories

Morphological neural networks are based on a new paradigm for neural computing. Instead of adding the products of neural values and corresponding synaptic weights, the basic neural computation in a morphological neuron takes the maximum or minimum of the sums of neural values and their corresponding synaptic weights. By taking the maximum (or minimum) of sums instead of the sum of products, morphological neuron computation is nonlinear before thresholding. As a consequence, the properties of morphological neural networks are drastically different than those of traditional neural network models. In this paper we restrict our attention to morphological associative memories. After a brief review of morphological neural computing and a short discussion about the properties of morphological associative memories, we present new methodologies and associated theorems for retrieving complete stored patterns from noisy or incomplete patterns using morphological associative memories. These methodologies are derived from the notions of morphological independence, strong independence, minimal representations of patterns vectors, and kernels. Several examples are provided in order to illuminate these novel concepts.     

Intelligent stock trading system with price trend prediction and reversal recognition using dual-module neural networks

This article presents an intelligent stock trading system that can generate timely stock trading suggestions according to the prediction of short-term trends of price movement using dual-module neural networks(dual net). Retrospective technical indicators extracted from raw price and volume time series data gathered from the market are used as independent variables for neural modeling. Both neural network modules of thedual net learn the correlation between the trends of price movement and the retrospective technical indicators by use of a modified back-propagation learning algorithm. Reinforcing the temporary correlation between the neural weights and the training patterns, dual modules of neural networks are respectively trained on a short-term and a long-term moving-window of training patterns. An adaptive reversal recognition mechanism that can self-tune thresholds for identification of the timing for buying or selling stocks has also been developed in our system. It is shown that the proposeddual net architecture generalizes better than one single-module neural network. According to the features of acceptable rate of returns and consistent quality of trading suggestions shown in the performance evaluation, an intelligent stock trading system with price trend prediction and reversal recognition can be realized using the proposed dual-module neural networks.     

Extracting structural attributes from IKONOS imagery for Eucalyptus plantation forests in KwaZulu-Natal,South Africa,using image texture analysis and artificial neural networks

The suitability of optical IKONOS satellite data (multispectral and panchromatic) for the estimation of forest structural attributes – for example, stems per hectare (SPHA), diameter at breast height (DBH), mean tree height (MTH), basal area (BA) and volume in plantation forest environments – was assessed in this study. The relationships of these forest structural attributes to statistical image texture from IKONOS imagery were analysed. The coefficients of determination (R ²) of multilinear regression models developed for the estimation of SPHA, DBH, MTH, BA and volume using statistical texture features from multispectral data were 0.63, 0.68, 0.81, 0.86 and 0.86, respectively. When the statistical texture features from panchromatic data were applied, the R ² for the respective forest structural attributes increased by 25%, 31%, 6%, 0.2% and 0.2%, respectively. Artificial neural network (ANN) models produced strong and significant relationships between estimated and actual measures of SPHA, DBH, MTH, BA and volume with an R ² of 0.83, 0.83, 0.90, 0.90 and 0.92, respectively, based on multispectral IKONOS data. Based on panchromatic IKONOS imagery, the R ² for the respective forest structural attributes increased by 18%, 12%, 5%, 3% and 6%, respectively. Results such as these bode well for the application of high spatial resolution imagery to forest structural assessment.     

Image segmentation with the SOLNN unsupervised logic neural network

In this article, an image segmentation method based on the SOLNN self-organising logic neural network is studied. The input image is initially processed using the TCS texture-highlighting technique and is then presented to the SOLNN network which segments it. The SOLNN is characterised by a variable sensitivity which enables it to be fine-tuned to detect different sub-textures within each texture to the desired degree of detail. The experimental results reported here illustrate the fact that the SOLNN indeed clusters accurately the textural information so that each cluster represents a single texture even for images which are objectively very difficult to segment. Thus, it is supported that the proposed approach leads to the design of an effective texture-based image-segmentation system.     

Morphogenetic approach to system identification

In this paper, we propose a novel approach to system identification based on morphogenetic theory (MT). Given a context H defined by a set of M objects, each described by a set of N attributes, and a vector X of desired outputs for each object, MT combines notions from formal concept analysis and tensor calculus so as to generate a morphogenetic system (MS). The MS is defined by a set of weights s¹, …, s^N, one for each attribute. Given H and X, weights are computed so as to generate the projection Y of X on the space of the attributes with the minimum distance between Y and X. An MS can be represented as a neuron, morphogenetic neuron, with a number of synapses equal to the number of attributes and synaptic weights equal to s¹, …, s^N. Unlike traditional neural network paradigm, which adopts an iterative process to determine synaptic weights, in MT, weights are computed at once. We introduce a method to generate a morphogenetic neural network (MNN) for identification problems. The method is based on extending appropriately and iteratively the attribute space so as to reduce the error between desired output and computed output. By using four well‐known datasets, we show that an MNN can identify an unknown system with a precision comparable with classical multilayer perceptron with complexity similar to the MNN but reducing drastically the time needed to generate the neural network. Furthermore, the structure of the MNN is generated automatically by the method and does not require a trial‐and‐error approach often applied in classical neural networks. © 2009 Wiley Periodicals, Inc.     

Incremental Learning with Respect to New Incoming Input Attributes

Neural networks are generally exposed to a dynamic environment where the training patterns or the input attributes (features) will likely be introduced into the current domain incrementally. This Letter considers the situation where a new set of input attributes must be considered and added into the existing neural network. The conventional method is to discard the existing network and redesign one from scratch. This approach wastes the old knowledge and the previous effort. In order to reduce computational time, improve generalization accuracy, and enhance intelligence of the learned models, we present ILIA algorithms (namely ILIA1, ILIA2, ILIA3, ILIA4 and ILIA5) capable of Incremental Learning in terms of Input Attributes. Using the ILIA algorithms, when new input attributes are introduced into the original problem, the existing neural network can be retained and a new sub-network is constructed and trained incrementally. The new sub-network and the old one are merged later to form a new network for the changed problem. In addition, ILIA algorithms have the ability to decide whether the new incoming input attributes are relevant to the output and consistent with the existing input attributes or not and suggest to accept or reject them. Experimental results show that the ILIA algorithms are efficient and effective both for the classification and regression problems.     

Model-based sensory pattern fusion

This article describes an evidential pattern classifier for the combination of data from physically different sensors. We assume that the sensory evidence is multiresolutional, incomplete, imprecise, and possibly inconsistent. Our focus is on two types of sensory information patterns: visual and tacticle. We develop a logical sensing scheme by using a model-based representation of prototypical 3-D surfaces. Each surface represents a class of topological patterns described by shape and curvature features. The sensory evidence is classified by using a conductivity measure to determine which prototypical surface best matches the evidence.     

联合形态学滤波和卷积稀疏编码的图像去雨

目的 图像去雨技术是对雨天拍摄图像中雨纹信息进行检测和去除,恢复目标场景的细节信息,从而获得清晰的无雨图像。针对现有方法对雨纹信息检测不完全、去除不彻底的问题,提出一种联合自适应形态学滤波和多尺度卷积稀疏编码(multi-scale convolution sparse coding, MS-CSC)的单幅图像去雨方法。方法 考虑雨纹信息的形状结构特点,构造一种自适应形态学滤波器来滤除有雨图像中的雨纹信息,获得包含图像自身纹理的低频成分;利用全变分模型正则化方法来增强低频成分的纹理信息,并利用有雨图像减去低频成分获得包含雨纹信息的高频成分;针对高频成分,根据雨纹的方向性提出一种基于方向梯度正则化的MS-CSC方法来重构高频成分,并通过迭代求解获得包含精确雨纹的高频成分,即雨层;利用有雨图像减去雨层得到最终的去雨图像。结果 为验证本文方法的有效性,与一些主流的去雨方法进行实验比较。实验结果表明,本文方法在模拟数据集上的平均峰值信噪比(peak signal-to-noise ratio, PSNR)和平均结构相似度(structural similarity, SSIM)指标分别提高了0...     

Building a perceptual visualization architecture

Scientific datasets are often difficult to analyse or visualize, due to their large size and high dimensionality. A multistep approach to address this problem is proposed. Data management techniques are used to identify areas of interest within the dataset. This allows the reduction of a dataset's size and dimensionality, and the estimation of missing values or correction of erroneous entries. The results are displayed using visualization techniques based on perceptual rules. The visualization tools are designed to exploit the power of the low-level human visual system. The result is a set of displays that allow users to perform rapid and accurate exploratory data analysis. In order to demonstrate the techniques, an environmental dataset being used to model salmon growth and migration patterns was visualized. Data mining was used to identify significant attributes and to provide accurate estimates of plankton density. Colour and texture were used to visualize the significant attributes and estimated plankton densities for each month for the years 1956-1964. Experiments run in the laboratory showed that the chosen colours and textures support rapid and accurate element identification, boundary detection, region tracking and estimation. The result is a visualization tool that allows users to quickly locate specific plankton densities and the boundaries they form. Users can compare plankton densities to other environmental conditions like sea surface temperature and current strength. Finally, users can track changes in any of the dataset's attributes on a monthly or yearly basis.     

Predicting old-growth tropical forest attributes from very high resolution (VHR)-derived surface metrics

Old-growth tropical forests are increasingly vanishing worldwide. Although the accurate quantification of tropical old-growth forests attributes is essential to understand, manage, and conserve their high diversity and biomass, conducting this task over large areas and at fine detail is not only expensive and time consuming, but also often practically impossible. This calls for the search for more efficient alternatives, particularly those based on remote sensing. In this study, we evaluate the potential of several surface metrics (tone and texture) extracted from very high resolution (VHR) satellite imagery to model the structural and diversity attributes of a tropical dry forest (TDF) in southern Mexico. We constructed simple linear models that used each forest attribute as dependent variables, and the tone and texture metrics extracted from several bands, the panchromatic (resolution = 0.5 m), red (R), infrared, and two vegetation indices (normalized difference vegetation index (NDVI), enhanced vegetation index (EVI); resolution = 2 m), of a VHR image (GeoEye-1) as predictive variables. The significance of the models including one, two, two and its interaction, and three image metrics was evaluated by comparing them with null models. The structural characteristics of the TDF (basal area (BA), mean height, stem density) showed the highest modelling potential, with the goodness-of-fit (R²) values ranging from 0.58 to 0.66. Conversely, no significant models were obtained for total crown area (TCA) and all diversity attributes. Our results show that remote-sensing metrics detect the spatial variation in the structural attributes of this old-growth TDF better than they detect the variation in its diversity. Our ability to model forest attributes at large scales at fine detail (sampling plots <0.2 ha) can be much improved by combining the use of VHR imagery with an array as wide as possible of the image surface metrics, including both tone and texture.     

False positive reduction in urinary particle recognition

The identification of non-cell objects in biological images is not a trivial task largely due to the difficulty in describing their characteristics in recognition systems. In order to better reduce the false positive rate caused by the presence of non-cell particles, we propose a novel approach using a local jet context features scheme combined with a two-tier object classification system. The newly proposed feature scheme, namely local jet context feature, integrates part of global features with the “local jet” features. The scheme aims to effectively describe the particle characteristics that are invariant to shift and rotation, and hence help to retain the critical shape information. The proposed two-tier particle classification strategy consists of a pre-recognition stage first and later a further filtering phase. Using the local jet context features coupled with a multi-class SVM classifier, the pre-recognition stage intends to assign the particles to their corresponding classes as many as possible. To further reduce the false positive particles, next a decision tree classifier based on shape-centered features is applied. Our experimental study shows that through the proposed two-tier classification strategy, we are able to achieve 85% of identification accuracy and 80% of F₁ value in urinary particle recognition. The experiment results demonstrate that the proposed local jet context features are capable to discriminate particles in terms of shape and texture characteristics. Overall, the two-tier classification stage is found to be effective in reducing the false positive rate caused by non-cell particles.     

Survey of Texture Mapping

Texture mapping is one of the most successful new techniques in high-quality image synthesis. It can enchance the visual richness of raster-scan images immensely while entailing only a relatively smann increase in computation. The technique has been applied to a number of surface attributes: surface color, surface normal, specularity, transparency, illumination, and surface displacement?to name a few. Although the list is potentially endless, the techniques of texture mapping are essentially the same in all cases. This article surveys the fundamentals of texture mapping, which can be spilt into two topics: the geometric mapping that warps a texture onto a surface, and the filtering necessary to avoid aliasing. An extensive bibliography is included.