Проблема описания системы функционирования английских артиклей в дидактических целях
The paper makes a brief introduction into multiple classifier systems and describes a particular algorithm which improves classification accuracy by making a recommendation of an algorithm to an object. This recommendation is done under a hypothesis that a classifier is likely to predict the label of the object correctly if it has correctly classified its neighbors. The process of assigning a classifier to each object involves here the apparatus of Formal Concept Analysis. We explain the principle of the algorithm on a toy example and describe experiments with real-world datasets.
The effect of conceptual flexibility involves inclusion of attributes that are irrelevant to the formed category in the concept and their further handling where required. The previous studies show that the conceptual flexibility effect arises while performing feature inference tasks and doesn’t arise while performing classification tasks. In the last case attention becomes too focused on one attribute. In the study the hypothesis according to which the conceptual flexibility effect may arise while performing classification tasks is tested on a sample of students (N=60). As this take place objects with attributes that are functionally connected and potentially related to semantic knowledge of the students are used as stimuli.
Symbolic classifiers allow for solving classification task and provide the reason for the classifier decision. Such classifiers were studied by a large number of researchers and known under a number of names including tests, JSM-hypotheses, version spaces, emerging patterns, proper predictors of a target class, representative sets etc. Here we consider such classifiers with restriction on counter-examples and discuss them in terms of pattern structures. We show how such classifiers are related. In particular, we discuss the equivalence between good maximally redundant tests and minimal JSM-hyposethes and between minimal representations of version spaces and good irredundant tests.
This book constitutes the refereed proceedings of the 6th IAPR TC3 International Workshop on Artificial Neural Networks in Pattern Recognition, ANNPR 2014, held in Montreal, QC, Canada, in October 2014. The 24 revised full papers presented were carefully reviewed and selected from 37 submissions for inclusion in this volume. They cover a large range of topics in the field of learning algorithms and architectures and discussing the latest research, results, and ideas in these areas.
In this paper, we use robust optimization models to formulate the support vector machines (SVMs) with polyhedral uncertainties of the input data points. The formulations in our models are nonlinear and we use Lagrange multipliers to give the first-order optimality conditions and reformulation methods to solve these problems. In addition, we have proposed the models for transductive SVMs with input uncertainties.
The material of the present paper is grounded on the holist algebraic method (Q-analysis) proposed by English mathematician and physicist R.H.Atkin. At its core, the approach is aimed at both analysis of systems structures (in the form of simplicial complexes K, which is formed by a set of properly adjoined objects called simplexes) and calculation of numeric estimates of structural complexity of systems based on the results of such analysis.
Turning complexity estimate of system’s structure into a real number creates additional difficulties in the comparison of two different complexes because there is no real verbal scale, which would have been accustomed to human beings and would allow a group of experts to express opinions and draw easily conclusions about degree of complexity of K at each particular dimensional level of its analysis. Therefore, the present paper deals with consideration of the approach that is more focused on human perception of characteristics obtained, mental comprehension and formation (comparison) of personal constructs in psychological space (or, P-space) – modified structural complexity estimate is based right on notions of distance and similarity within psychological space.
Our experimental study looked into the way existing knowledge influences the way subjects con- struct the rules of categorization and modify them as they are applied. We modified the experiment of E. Wisniewski and D. Medina (1994) by asking the respondents not only to create a categorization rule, but also to use it to categorize new images, and we looked at the frequency and type of subsequent rule modification. The respondents, 114 university students, were given a set of images drawn by children and asked to identify their common features under one of the four conditions: relevant prior knowledge (participants were told that the drawings had been made by children with high and low creativity), stan- dard condition (participants were told the drawings had been made by children from groups A and B), standard condition with examples (one sample of drawings from each group was shown), and irrelevant knowledge. We found that under the relevant prior knowledge condition, compared to the other three conditions, the respondents tended to construct more complex and abstract rules and to change them more frequently when they categorized new objects. We also found that rule modifications during usage led to more complex and abstract rules under all four conditions. We interpret the findings as evidence for two stages of categorization, the first stage involving search for existing generalizations in semantic memory, and the second stage involving adaptation of prior knowledge to current conditions.