Exploratory Analysis of Metallurgical Process Data with Neural Networks and Related MethodsThis volume is concerned with the analysis and interpretation of multivariate measurements commonly found in the mineral and metallurgical industries, with the emphasis on the use of neural networks. The book is primarily aimed at the practicing metallurgist or process engineer, and a considerable part of it is of necessity devoted to the basic theory which is introduced as briefly as possible within the large scope of the field. Also, although the book focuses on neural networks, they cannot be divorced from their statistical framework and this is discussed in length. The book is therefore a blend of basic theory and some of the most recent advances in the practical application of neural networks. |
Contents
| 1 | |
CHAPTER 2 TRAINING OF NEURAL NETWORKS | 50 |
CHAPTER 3 LATENT VARIABLE METHODS | 74 |
CHAPTER 4 REGRESSION MODELS | 112 |
CHAPTER 5 TOPOGRAPHICAL MAPPINGS WITH NEURAL NETWORKS | 172 |
CHAPTER 6 CLUSTER ANALYSIS | 199 |
CHAPTER 7 EXTRACTION OF RULES FROM DATA WITH NEURAL NETWORKS | 228 |
CHAPTER 8 INTRODUCTION TO THE MODELLING OF DYNAMIC SYSTEMSCHAPTER | 262 |
DYNAMIC SYSTEMS ANALYSIS AND MODELLING | 285 |
CHAPTER 10 EMBEDDING OF MULTIVARIATE DYNAMIC PROCESS SYSTEMS | 299 |
CHAPTER 11 FROM EXPLORATORY DATA ANALYSIS TO DECISION SUPPORT AND PROCESS CONTROL | 313 |
| 333 | |
| 366 | |
DATA FILES | 370 |
Other editions - View all
Exploratory Analysis of Metallurgical Process Data with Neural ..., Volume 1 Chris Aldrich No preview available - 2002 |
Common terms and phrases
activation addition algorithm analysis application approach approximately associated attractor attribute calculated classification cluster coefficients complexity computational considered consists constructed containing continuous correlation curve data set decision defined dependent derived determined dimension direction distance distribution dynamic embedding equation error estimated example exemplars extracted Figure fitted follows fuzzy rules Gaussian given hidden layer indicated individual initial input learning least linear matrix means measure methods mill minimize multivariate neural network nodes noise nonlinear objects observations obtained operator optimal original output parameters pattern performance plant points possible prediction principal component principal component analysis problem projection radial basis function reconstructed region regression represented respectively rules sample scale selected separation shown in Figure similar single space squares statistical step structure Table techniques tree values variables variance vector weight
Popular passages
Page 335 - The sample complexity of pattern classification with neural networks: The size of the weights is more important than the size of the network.
Page 360 - Differential evolution: A simple and efficient adaptive scheme for global optimization over continuous spaces.
Page 338 - Shavlik, JW (1994). Using sampling and queries to extract rules from trained neural networks.
Page 341 - A growing neural gas network learns topologies. In: Tesauro, G., Touretzky, DS, Leen, TK (eds.) Advances in Neural Information Processing Systems, vol.
Page 363 - The GENITOR Algorithm and Selective Pressure: Why Rank-Based Allocation of Reproductive Trials is Best, in Proc.
Page 348 - Kramer, MA (1991). Nonlinear Principal Component Analysis Using Autoassociative Neural Networks.
Page 21 - Each process element in the Kohonen layer measures the Euclidean distance of its weights to the input values (exemplars) fed to the layer. For example, if the input data consist of M-dimensional vectors of the form x = {x\,xi, . . .xM}, then each Kohonen element will have M weight values, which can be denoted by w, = {Wi\,Wii,...WiM}.