Training algorithm

Chau KW (2006) Particle swarm optimization training algorithm for ANNs in stage prediction of Shing Mun River. J Hydrol 329 363-367... 

A number of approaches to predict ionization based on structure have been published (for a review, see [53]) and some of these are commercially available. Predictions tend to be good for structures with already known and measured functional groups. However, predictions can be poor for new innovative structures. Nevertheless, pfCa predictions can still be used to drive a project in the desired direction and the rank order of the compounds is often correct. More recently training algorithms have also become available which use in-house data to improve the predictions. This is obviously the way forward. 

It follows that the most important aspect to train an ANN is to determine how the weights are modified in an iterative, automatic process that is, the criterion to change the weights. How this is done determines the difierent learning (training) algorithms. Diverse learning methods exist [30,32] of which some typical examples are cited first, then we will concentrate on the most common one. 

Properties such as thermodynamic values, sequence asymmetry, and polymorphisms that contribute to RNA duplex stability are taken into account by these databases (Pei and Tuschl 2006). In addition, artificial neural networks have been utilized to train algorithms based on the analysis of randomly selected siRNAs (Huesken et al. 2005). These programs siphon significant trends from large sets of RNA sequences whose efficacies are known and validated. Certain base pair (bp) positions have a tendency to possess distinct nucleotides (Figure 9.2). In effective nucleotides, position 1 is preferentially an adenosine (A) or uracil (U), and many strands are enriched with these nucleotides along the first 6 to 7 bps of sequence (Pei and Tuschl 2006). The conserved RISC cleavage site at nucleotide position 10 favors an adenosine, which may be important, while other nucleotides are... 

After the Minsky and Papert book in 1969 (Minsky Papert, 1969) which clarified the linearity restrictions of perceptrons, little work was done with perceptrons. However, in 1986 McClelland and Rumelhart (McClelland Rumelhart, 1986) revived the field with multilayer perceptrons and an intuitive training algorithm called back-propagation (discussed in Chapter 5). 

Radial basis function networks and multilayer perceptions have similar functions but their training algorithms are dramatically different. Training radial basis function networks proceeds in two steps. First the hidden layer parameters are determined as a function of the input data and then the weights between the hidden and output layers are determined from the output of the hidden layer and the target data. 

In addition to the described training algorithms for supervised learning, several practical problems need to be addressed during training. One serious problem with multilayer... 

Prechelt, L. (1994). PROBEN1 - A set of benchmarks and benchmarking rules for neural network training algorithms. Fakultat fur Informatik, Universitat Karlsruhe, Karlsruhe, Germany. 

There is a need to design a fuzzy training algorithm by a hierarchical procedure. 

Lennox, B. Montague, G.A. Frith, A.M. Gent, C. Bevan, V. Industrial apphcations of neural networks—an investigation. J. Process Control. 2001, 11, 497-507. Murtoniemi, E. Merkku, P. Yliruusi, J. Comparison of four different neural network training algorithms in modelling the fluidized bed granulation process. Lab. Microcom-put. 1993, 12, 69-76. 

Furthermore, in the multivariable problem, while three to five variables can be handled relatively easily, one reaches a computational bottleneck for larger problems. This can be possibly resolved by considering some of the new developments in HMM training algorithms [254, 71],... 

The optimum number of neurons in the hidden layer varied with the type of training algorithm. If a Bayesian regulation backpropagation algorithm is used, the optimum number of... 

All algorithms are to some extent data-driven even hand written rules use some data either explicitly or in a mental representation where the developer ean imagine examples and how they should be dealt with. The difference between hand written rules and data driven teehniques lies not in whether one uses data or not, but how the data is used. Most data-driven teehniques have an automatic training algorithm such that they can be trained on the data without the need for human intervention. 

---