Flexible Metric Nearest Neighbor Classi ̄cation
نویسنده
چکیده
The K-nearest-neighbor decision rule assigns an object of unknown class to the plurality class among the K labeled \training" objects that are closest to it. Closeness is usually de ̄ned in terms of a metric distance on the Euclidean space with the input measurement variables as axes. The metric chosen to de ̄ne this distance can strongly e®ect performance. An optimal choice depends on the problem at hand as characterized by the respective class distributions on the input measurement space, and within a given problem, on the location of the unknown object in that space. In this paper new types of K-nearest-neighbor procedures are described that estimate the local relevance of each input variable, or their linear combinations, for each individual point to be classi ̄ed. This information is then used to separately customize the metric used to de ̄ne distance from that object in ̄nding its nearest neighbors. These procedures are a hybrid between regular K-nearest-neighbor methods and tree-structured recursive partitioning techniques popular in statistics and machine learning.
منابع مشابه
Symbolic Nearest Mean Classiiers
Piew Datta and Dennis Kibler Department of Information and Computer Science University of California Irvine, CA 92717 fpdatta, [email protected] Abstract The minimum-distance classi er summarizes each class with a prototype and then uses a nearest neighbor approach for classi cation. Three drawbacks of the minimum-distance classi er are its inability to work with symbolic attributes, weigh at...
متن کاملTemporal and Frequential Metric Learning for Time Series kNN Classification
This work proposes a temporal and frequential metric learning framework for a time series nearest neighbor classi cation. For that, time series are embedded into a pairwise space where a combination function is learned based on a maximum margin optimization process. A wide range of experiments are conducted to evaluate the ability of the learned metric on time series kNN classi cation.
متن کاملApplication of k Nearest Neighbor on Feature Projections Classi er to Text Categorization
This paper presents the results of the application of an instance based learning algorithm k Nearest Neighbor Method on Fea ture Projections k NNFP to text categorization and compares it with k Nearest Neighbor Classi er k NN k NNFP is similar to k NN ex cept it nds the nearest neighbors according to each feature separately Then it combines these predictions using a majority voting This prop er...
متن کاملAdaptive Nearest Neighbor Classi cation using
The nearest neighbor technique is a simple and appealing method to address classi-cation problems. It relies on the assumption of locally constant class conditional probabilities. This assumption becomes invalid in high dimensions with a nite number of examples due to the curse of dimensionality. Severe bias can be introduced under these conditions when using the nearest neighbor rule. The empl...
متن کاملMetrics and Models for Handwritten Character Recognition
A digitized handwritten numeral can be represented as a binary or greyscale image. An important pattern recognition task that has received much attention lately is to automatically determine the digit, given the image. While many di erent techniques have been pushed very hard to solve this task, the most successful and intuitively appropriate is due to Simard (Simard, LeCun & Denker 1993). Thei...
متن کامل