In this paper we study random forests through their connection with a new framework of adaptive nearest neighbor methods. We first introduce a concept of potential nearest neighbors (k-PNN’s) and show that random forests can be seen as adaptively weighted k-PNN methods. Various aspects of random forests are then studied from this perspective. We investigate the effect of terminal node sizes and...

The clustering over various granularities for high dimensional data in arbitrary shape is a challenge in data mining. In this paper Nearest Neighbors Absorbed First (NNAF) clustering algorithm is proposed to solve the problem based on the idea that the objects in the same cluster must be near. The main contribution includes:(1) A theorem of searching nearest neighbors (SNN) is proved. Based on ...

2 KNN Algorithm 2 2.

The growing information infrastructure in a variety of disciplines involves an increasing requirement for efficient data mining techniques. Fast dimensionality reduction methods are important for understanding and processing of large data sets of high-dimensional patterns. In this work, unsupervised nearest neighbors (UNN), an efficient iterative method for dimensionality reduction, is presente...

A key problem in time series prediction using autoregressive models is to fix the model order, namely the number of past samples required to model the time series adequately. The estimation of the model order using cross-validation is a long process. In this paper we explore faster alternative to cross-validation, based on nonlinear dynamics methods, namely Grassberger-Procaccia, Kégl and False...

The method of false nearest neighbors (FNN) is presented here as a tool for analyzing the \dimensionali-ty" of a nonlinear input/output system directly from data. A unique feature of this method is that no speciic model structure is assumed, the dimensional determination is made directly from topological considerations. An extension to the FNN algorithm is given to consider the problem of infer...

In this paper, we test a constructive methodology for shaping neural networks models of non-linear dynamic systems. The method is supported by results and prescriptions related to the Takens-Mañé theorem, and is based on the measurement of the first minimum of the mutual information of the output signal, and in the application of the method of global false nearest neighbors to determine the emb...

In this paper, the necessary embedding dimension for electroencephologram(EEG) data is calculated. The tools used include a 1981 theorem from Floris Takens, and a subsequent extension of that theorem in 1991 given by Casdagli, Sauer and Yorke. These theorems give a method of reconstructing the phase space of the original system up to diffeomorphism using time-delays. If d is the original dimens...

Fractional derivatives are applied in the reconstruction, from a single observable, of the dynamics of a Duffing oscillator and a two-well experiment. The fractional derivatives of time series data are obtained in the frequency domain. The derivative fraction is evaluated using the average mutual information between the observable and its fractional derivative. The ability of this reconstructio...

In order to deliver the promise of Moore’s Law to the end user, compilers must make decisions that are intimately tied to a specific target architecture. As engineers add architectural features to increase performance, systems become harder to model, and thus, it becomes harder for a compiler to make effective decisions. Machine-learning techniques may be able to help compiler writers model mod...