Learning method

In the case of chemoinformatics this process of abstraction will be performed mostly to gain knowledge about the properties of compounds. Physical, chemical, or biological data of compounds will be associated with each other or with data on the structure of a compound. These pieces of information wQl then be analyzed by inductive learning methods to obtain a model that allows one to make predictions. 

Inductive methods for establishing a correlation between chemical compounds and their properties are the theme of Chapter 9. In many cases, the structure of chemical compounds has to be pre-processed in order to make it amenable to inductive learning methods. This is usually achieved by means of structure descriptors, methods for the calculation of which are outlined in Chapter 8. 

Let us start with a classic example. We had a dataset of 31 steroids. The spatial autocorrelation vector (more about autocorrelation vectors can be found in Chapter 8) stood as the set of molecular descriptors. The task was to model the Corticosteroid Ringing Globulin (CBG) affinity of the steroids. A feed-forward multilayer neural network trained with the back-propagation learning rule was employed as the learning method. The dataset itself was available in electronic form. More details can be found in Ref. [2]. 

An observation of the results of cross-validation revealed that all but one of the compounds in the dataset had been modeled pretty well. The last (31st) compound behaved weirdly. When we looked at its chemical structure, we saw that it was the only compound in the dataset which contained a fluorine atom. What would happen if we removed the compound from the dataset The quahty ofleaming became essentially improved. It is sufficient to say that the cross-vahdation coefficient in-CTeased from 0.82 to 0.92, while the error decreased from 0.65 to 0.44. Another learning method, the Kohonen s Self-Organizing Map, also failed to classify this 31st compound correctly. Hence, we had to conclude that the compound containing a fluorine atom was an obvious outlier of the dataset. 

It extends the usage of statistical methods and combines it with machine learning methods and the application of expert systems. The visualization of the results of data mining is an important task as it facilitates an interpretation of the results. Figure 9-32 plots the different disciplines which contribute to data mining. 

On top of that, reaction databases can also be used to derive knowledge on chemical reactions which can then be used for reaction prediction, The huge amount of information in reaction databases can be processed by inductive learning methods in order to condense these individual pieces of information into essential features... 

The main characteristics of the method, developed in our group for reaction classification arc 1) the representation of a reaction by physicochemical values calculated for the bonds being broken and made during the reaction, and 2 use of the unsupervised learning method of a self-organi2ing neural network for the perception of similarity of chemical reactions [3, 4],... 

Thi.s could provide a much richer source of information on chemical reactions and thus build a better basis for automatic learning methods. 

The understanding and simulation of chemical reactions is one of the great challenges of chemoinformatics. Each day millions of reactions are performed, sometimes with rather poor results because of our limited understanding of chemical reactivity and the influence of solvents, catalysts, temperature, etc. This problem has to be tackled by both deductive and inductive learning methods. 

Multiple linear regression is strictly a parametric supervised learning technique. A parametric technique is one which assumes that the variables conform to some distribution (often the Gaussian distribution) the properties of the distribution are assumed in the underlying statistical method. A non-parametric technique does not rely upon the assumption of any particular distribution. A supervised learning method is one which uses information about the dependent variable to derive the model. An unsupervised learning method does not. Thus cluster analysis, principal components analysis and factor analysis are all examples of unsupervised learning techniques. 

The Back-Propagation Algorithm (BPA) is a supervised learning method for training ANNs, and is one of the most common forms of training techniques. It uses a gradient-descent optimization method, also referred to as the delta rule when applied to feedforward networks. A feedforward network that has employed the delta rule for training, is called a Multi-Layer Perceptron (MLP). 

The general principle behind most commonly used back-propagation learning methods is the delta rule, by which an objective function involving squares of the output errors from the network is minimized. The delta rule requires that the sigmoidal function used at each neuron be continuously differentiable. This methods identifies an error associated with each neuron for each iteration involving a cause-effect pattern. Therefore, the error for each neuron in the output layer can be represented as ... 

Eormal education and induction training alone do not fully prepare individuals for the world of work continuous learning by way of a variety of learning methods is the only way to achieve maximum performance. 

To further clarify how mentoring differs from training, management and other one-to-one development approaches, take a look at Table 1.2. This also illustrates the circumstances under which mentoring should be chosen as an employee learning method as opposed to any of the other ones. 

Mentoring relationships provide an effective method of learning by avoiding the limitations of traditional teaching and learning methods. The unique positive characteristics associated with mentoring as a development method are as follows ... 

In the case where an employee does not perceive mentoring as an appropriate learning method for him or herself, choosing a... 

Active and cooperative learning methods are consistent with social/cultural constructivism, provide a better learning enviromnent and contribute to deeper understanding and development of learning skills (Duncan-Hewitt, Mount, Apple, 1995 Johnson, Johnson, Smith, 1991). This form of learning is traditionally used in laboratory work. It was also used in the project-emiched physical chemistry laboratoiy described above (Tsaparlis Gorezi, 2005 2007). 

This is the framework of teaching and learning methods that are nsed to communicate the substantive and syntactical stmcture of a snbject. This framework has been... 

Cristianini N, Shawe-Taylor J. An introduction to support vector machines and other kernel-based learning methods. Cambridge, UK Cambridge University Press, 2000. 

Li H, Yap CW, Ung CY, Xue Y, Cao ZW and Chen YZ Effect of selection of molecular descriptors on the prediction of blood-brain barrier penetrating and nonpenetrating agents by statistical learning methods. J Chem Inf Model 2005 45 1376-1384. 

Finally, having set up the learning problem, we need to employ a learning method that will guarantee preservation of the correctness of the branch-and-bound algorithm and make useful additions to the control information we have about the problem (Section V). 

Yap, C. W U, Z. R Chen, Y. Z. Quantitative strucmre-pharmacokinetic relationships for drug clearance by using statistical learning methods. /. Mol. Graph. Model. 2006, 24, 383-395. 

Muller, K.-R., Ratsch, G., Sonnenburg, S., Mika, S., Grimm, M., Heinrich, N. Classifying drug-likeness with kernel-based learning methods. J. Chem. Inf. Model. 2005, 45, 249-253. 

A machine-learning method was proposed by Klon et al. [104] as an alternative form of consensus scoring. The method proved unsuccessful for PKB, but showed promise for the phosphatase PTPIB (protein tyrosine phosphatase IB). In this approach, compounds were first docked into the receptor and scored using conventional means. The top scoring compounds were then assumed to be active and used to build a naive Bayes classification model, all compounds were subsequently re-scored and ranked using the model. The method is heavily dependent upon predicting accurate binding... 

Genetic programming, a specific form of evolutionary computing, has recently been used for predicting oral bioavailability [23], The results show a slight improvement compared with the ORMUCS Yoshida-Topliss approach. This supervised learning method and other described methods demonstrate that at least qualitative (binned) predictions of oral bioavailability seem tractable directly from the structure. 

Learning methods for food preparation that incorporate lower amounts of oil and fat (for example, using margarine instead of butter or oil to prepare fried chicken, or baking instead of frying). 

---