the process of developing applications of machinelearning and dataminingthat employ supervised classification algorithms includes the important step of knowledge verification. Interpretable output is presented to a...
详细信息
the process of developing applications of machinelearning and dataminingthat employ supervised classification algorithms includes the important step of knowledge verification. Interpretable output is presented to a user so that they can verify that the knowledge contained in the output makes sense for the given application. As the development of an application is an iterative process it is quite likely that a user would wish to compare models constructed at various times or stages. One crucial stage where comparison of models is important is when the accuracy of a model is being estimated, typically using some form of cross-validation. this stage is used to establish an estimate of how well a model will perform on unseen data. this is vital information to present to a user, but it is also important to show the degree of variation between models obtained from the entire dataset and models obtained during cross-validation. In this way it can be verified that the cross-validation models are at least structurally aligned withthe model garnered from the entire dataset. this paper presents a diagnostic tool for the comparison of tree-based supervised classification models. the method is adapted from work on approximate tree matching and applied to decision trees. the tool is described together with experimental results on standard datasets.
Addresses the extraction of knowledge from recurrent neural networks trained to behave like deterministic finite-state automata (DFAs). To date, methods used to extract knowledge from such networks have relied on the ...
详细信息
Addresses the extraction of knowledge from recurrent neural networks trained to behave like deterministic finite-state automata (DFAs). To date, methods used to extract knowledge from such networks have relied on the hypothesis that network states tend to cluster and that clusters of network states correspond to DFA states. the computational complexity of such a cluster analysis has led to heuristics which either limit the number of clusters that may form during training or limit the exploration of the output space of hidden recurrent state neurons. these limitations, while necessary, may lead to reduced fidelity, i.e. the extracted knowledge may not model the true behavior of a trained network, perhaps not even for the training set. the method proposed uses a polynomial-time symbolic learning algorithm to infer DFAs solely from the observation of a trained network's input/output behavior. thus, this method has the potential to increase the fidelity of the extracted knowledge.
the proceedings contains 78 papers from the 1997 IEEE internationalconference on Tools with Artificial Intelligence. Topics discussed include: neural networks;knowledge representation and reasoning;artificial intelli...
详细信息
the proceedings contains 78 papers from the 1997 IEEE internationalconference on Tools with Artificial Intelligence. Topics discussed include: neural networks;knowledge representation and reasoning;artificial intelligence;software engineering;genetic algorithms;logic based reasoning systems;natural language processing;vision and patternrecognition;optimization problem solving tools;evolutionary computation;object-oriented methodologies;intelligent agents;knowledge based systems;intelligent user interfaces;datamining;and machinelearning.
Automatic statistical clustering techniques have been applied to implement different multiple prototype classifiers. Multiple prototyping offers an optimised solution to cases where there is significant variability in...
详细信息
ISBN:
(纸本)085296692X
Automatic statistical clustering techniques have been applied to implement different multiple prototype classifiers. Multiple prototyping offers an optimised solution to cases where there is significant variability in the training data. A typical application area is the recognition of handwritten characters. Once a set of features has been extracted, different statistical clustering techniques can be implemented to achieve multi-dimensional clustering in the feature space. Building of prototypes from these clusters is straight-forward, the success of the multi-prototyping depends on the efficiency of the statistical clustering techniques. In this paper, different clustering techniques have been used in conjunction withthe use of different approaches to the formation of prototypes and the relative performance enhancements are reported.
An object tracking system based on a template matching approach is demonstrated. the system identifies the target and tracks a sequence of video-recorded images without losing the object by using a genetic algorithm (...
详细信息
ISBN:
(纸本)085296692X
An object tracking system based on a template matching approach is demonstrated. the system identifies the target and tracks a sequence of video-recorded images without losing the object by using a genetic algorithm (GA)-based learning to adapt template matching processes to environmental conditions. the adaptive GA generates templates better than fixed, random or indexed template generation techniques in terms of boththe cross-correlation score and time necessary to choose the template model. While the overload using the GA is minimal if it is not necessary to change template model, in all other cases the GA offers a better solution to the search the associated search and optimization problem than usual algorithms.
the backpropagation neural network is applied to three gauging fringe analysis applications: classification of five spherical surfaces of differing radii;classification of five real objects with surfaces of different ...
详细信息
ISBN:
(纸本)085296692X
the backpropagation neural network is applied to three gauging fringe analysis applications: classification of five spherical surfaces of differing radii;classification of five real objects with surfaces of different radii of curvature;and identification of eggs according to their given commercial grades. three methods for creating test and training vectors are used, of which the fast Fourier transform demonstrated drastic reduction in the network size, while applications with relatively noise-free data are indicated for mean/standard deviation driven networks.
the paper proposes a general framework for shape detection based on supervised symbolic learning. Differently from other visual systems exploiting machinelearning, the proposed architecture does not follow the object...
详细信息
this paper describes how techniques from the discipline of neuro-fuzzy and soft computing techniques can be used, in conjunction with methodologies from patternrecognition and digital signal processing, to effectivel...
详细信息
this paper describes how techniques from the discipline of neuro-fuzzy and soft computing techniques can be used, in conjunction with methodologies from patternrecognition and digital signal processing, to effectively perform speech data classification. In particular, we have applied the proposed method to automatic speaker recognition and achieved satisfactory results.
In this paper a method for knowledge acquisition in diagnosis problems is presented. this method results in a zero-order Sugeno rule base where the combinatorial explosion of rules is solved by a decomposition scheme....
详细信息
In this paper a method for knowledge acquisition in diagnosis problems is presented. this method results in a zero-order Sugeno rule base where the combinatorial explosion of rules is solved by a decomposition scheme. this approach allows a unified representation, where the knowledge obtained from data by a supervised learning algorithm can be directly confronted withthe knowledge elicited from the experts. the supervised learning algorithm is rested upon some classification problems found in literature.
An efficient corner matching algorithm is developed to minimize the amount of calculation. To reduce the amount of calculation, all available information from a corner detector is used to generate the model. this info...
详细信息
ISBN:
(纸本)0780337964
An efficient corner matching algorithm is developed to minimize the amount of calculation. To reduce the amount of calculation, all available information from a corner detector is used to generate the model. this information has uncertainties due to discretization noise and geometric distortion, and this is represented by fuzzy rule base which can represent and handle the uncertainties. From fuzzy inference procedure, a matched segment list is extracted, and resulted segment list is used to calculate the transformation between object of model and scene. An auto-tuning scheme of the fuzzy rule base is developed to find out the uncertainties of features from recognized results automatically. To show the effectiveness of the developed algorithm, experiments are conducted for synthetic images and images of real electronic components.
暂无评论