Home About us Contact
|
Home About us Contact | ||
|
|
||
Information Gain (information + gain)
Selected AbstractsEvaluation Metrics in Classification: A Quantification of Distance-BiasCOMPUTATIONAL INTELLIGENCE, Issue 3 2003Ricardo Vilalta This article provides a characterization of bias for evaluation metrics in classification (e.g., Information Gain, Gini, ,2, etc.). Our characterization provides a uniform representation for all traditional evaluation metrics. Such representation leads naturally to a measure for the distance between the bias of two evaluation metrics. We give a practical value to our measure by observing the distance between the bias of two evaluation metrics and its correlation with differences in predictive accuracy when we compare two versions of the same learning algorithm that differ in the evaluation metric only. Experiments on real-world domains show how the expectations on accuracy differences generated by the distance-bias measure correlate with actual differences when the learning algorithm is simple (e.g., search for the best single feature or the best single rule). The correlation, however, weakens with more complex algorithms (e.g., learning decision trees). Our results show how interaction among learning components is a key factor to understand learning performance. [source] Behavioral phenotyping enhanced , beyond (environmental) standardizationGENES, BRAIN AND BEHAVIOR, Issue 1 2002H. Würbel It is basic biology that the phenotype of an animal is the product of a complex and dynamic interplay between nature (genotype) and nurture (environment). It is far less clear, however, how this might translate into experimental design and the interpretation of animal experiments. Animal experiments are a compromise between modelling real world phenomena with maximal validity (complexity) and designing practicable research projects (abstraction). Textbooks on laboratory animal science generally favour abstraction over complexity. Depending on the area of research, however, abstraction can seriously compromise information gain, with respect to the real world phenomena an experiment is designed to model. Behavioral phenotyping of mouse mutants often deals with particularly complex manifestations of life, such as learning, memory or anxiety, that are strongly modulated by environmental factors. A growing body of evidence indicates that current approaches to behavioral phenotyping might often produce results that are idiosyncratic to the study in which they were obtained, because the interactive nature of genotype-environment relationships underlying behavioral phenotypes was not taken into account. This paper argues that systematic variation of genetic and environmental backgrounds, instead of excessive standardization, is needed to control the robustness of the results and to detect biologically relevant interactions between the mutation and the genetic and environmental background of the animals. [source] Assessment of four modifications of a novel indexing technique for case-based reasoningINTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS, Issue 4 2007Mykola Galushka In this article, we investigate four variations (D-HSM, D-HSW, D-HSE, and D-HSEW) of a novel indexing technique called D-HS designed for use in case-based reasoning (CBR) systems. All D-HS modifications are based on a matrix of cases indexed by their discretized attribute values. The main differences between them are in their attribute discretization stratagem and similarity determination metric. D-HSM uses a fixed number of intervals and simple intersection as a similarity metric; D-HSW uses the same discretization approach and a weighted intersection; D-HSE uses information gain to define the intervals and simple intersection as similarity metric; D-HSEW is a combination of D-HSE and D-HSW. Benefits of using D-HS include ease of case and similarity knowledge maintenance, simplicity, accuracy, and speed in comparison to conventional approaches widely used in CBR. We present results from the analysis of 20 case bases for classification problems and 15 case bases for regression problems. We demonstrate the improvements in accuracy and/or efficiency of each D-HS modification in comparison to traditional k -NN, R-tree, C4,5, and M5 techniques and show it to be a very attractive approach for indexing case bases. We also illuminate potential areas for further improvement of the D-HS approach. © 2007 Wiley Periodicals, Inc. Int J Int Syst 22: 353,383, 2007. [source] Internet networking for pharmacists: an evaluation of a mailing list for UK pharmacistsINTERNATIONAL JOURNAL OF PHARMACY PRACTICE, Issue 1 2003Mr. Anthony R. Cox teaching fellow Objective To analyse the content of messages to an internet mailing list for UK pharmacists and to ascertain if the list was performing a continuing professional development (CPD) function. Method For one month all messages to the main list were categorised by topic; details of the gender of the correspondent and their sector of the profession were noted. Members were surveyed using an internet questionnaire. Setting The population of subscribers to the mailing list at http:www.private-rx.com Key findings The top three categories of e-mails posted to the list were clinical pharmacy (20%), pharmacy politics (18%) and non-pharmacy chat (14%). Other subjects included legal issues, the Drug Tariff, government policy, business, risk management and e-mails of a personal and supportive nature. The survey obtained a 46% response rate. Ninety-eight per cent of respondents found the list valuable. Respondents reported increased face to face and Internet contact with other pharmacists after joining the list. Forty-four per cent of respondents said their practice had changed as a result of information gained from the mailing list. Qualitative data self-reported by respondents indicated increased self-perceived competence, confidence, knowledge and skills. Approaches to CPD had also been re-examined. Listening to peers' views and overcoming isolation was seen as important. Conclusion Private-Rx provided pharmacists with a rapid route for information gain, had perceived benefits and appeared to have brought about changes in practice. Internet discussion enables CPD without the restriction of time or place and reaches pharmacists who are under-represented in formal education programmes. [source] Pharmacokinetic/pharmacodynamic studies in drug product developmentJOURNAL OF PHARMACEUTICAL SCIENCES, Issue 1 2002Bernd Meibohm Abstract In the quest of ways for rationalizing and accelerating drug product development, integrated pharmacokinetic/pharmacodynamic (PK/PD) concepts provide a highly promising tool. PK/PD modeling concepts can be applied in all stages of preclinical and clinical drug development, and their benefits are multifold. At the preclinical stage, potential applications might comprise the evaluation of in vivo potency and intrinsic activity, the identification of bio-/surrogate markers, as well as dosage form and regimen selection and optimization. At the clinical stage, analytical PK/PD applications include characterization of the dose,concentration,effect/toxicity relationship, evaluation of food, age and gender effects, drug/drug and drug/disease interactions, tolerance development, and inter- and intraindividual variability in response. Predictive PK/PD applications can also involve extrapolation from preclinical data, simulation of drug responses, as well as clinical trial forecasting. Rigorous implementation of the PK/PD concepts in drug product development provides a rationale, scientifically based framework for efficient decision making regarding the selection of potential drug candidates, for maximum information gain from the performed experiments and studies, and for conducting fewer, more focused clinical trials with improved efficiency and cost effectiveness. Thus, PK/PD concepts are believed to play a pivotal role in streamlining the drug development process of the future. © 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 91:18,31, 2002 [source] EFFICIENT MARKOV NETWORK DISCOVERY USING PARTICLE FILTERSCOMPUTATIONAL INTELLIGENCE, Issue 4 2009Dimitris Margaritis In this paper, we introduce an efficient independence-based algorithm for the induction of the Markov network (MN) structure of a domain from the outcomes of independence test conducted on data. Our algorithm utilizes a particle filter (sequential Monte Carlo) method to maintain a population of MN structures that represent the posterior probability distribution over structures, given the outcomes of the tests performed. This enables us to select, at each step, the maximally informative test to conduct next from a pool of candidates according to information gain, which minimizes the cost of the statistical tests conducted on data. This makes our approach useful in domains where independence tests are expensive, such as cases of very large data sets and/or distributed data. In addition, our method maintains multiple candidate structures weighed by posterior probability, which allows flexibility in the presence of potential errors in the test outcomes. [source] | ||||||||||