科研之友
微信
新浪微博
Facebook
分享链接
摘要
Lazy learning is a memory-based learning technique that uses a query-based approach to estimate the best local model configuration by selecting neighbors of the query. The bottleneck of this algorithm is management of large datasets and searching for the relevant neighbors. This paper aims to deal with this issue and the kappa-Means cluster algorithm is cast into the lazy learning framework. By using this strategy, the nearest neighbors searching process can be converted into a hierarchical searching process with two levels and a lot of searching time for lazy learning can be saved. A novel criterion between two samples is proposed, and based on this criterion, the kappa- Vector Nearest Neighbors (kappa-VNN) is used to find the neighbors of the query, The database can quickly be updated without the need for any further computation. This updating strategy can save a lot of memory space and decrease the neighbor-searching time. These techniques have been successfully applied to estimate a nonlinear function.
