Knowledge Granulation Based Roughness Measure for Neighborhood Rough Sets

Yang, Chengdong; Qiu, Jianlong; Zhang, Wenyin

doi:10.1109/isdea.2012.218

Cited by 3 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This section presents the fundamental concept of four forward greedy feature selection techniques based on NRS [31], VPNRS [34], CMNRS [35] and GKNRS [36]. The detail implementation of each of the technique for HSI band selection is presented in the following subsections.…”

Section: Hsi Band Selection Modelsmentioning

confidence: 99%

“…Definition 13 Given a neighbourhood information system

false(U, C \cup D, N false)

B \subseteq C

. The granulation of knowledge of B is defined as

G K_{B} false(δ false) = \frac{1}{{(||, U)}^{2}} {false\sum}_{i = 1}^{|U|} |δ_{B} (x_{i})|

Yang et al [36] proposed a new roughness measure by combining classical roughness

ρ_{B} false(D false)

and granulation knowledge

G K_{B} false(δ false)

.Definition 14 Given a neighbourhood decision table

false(U, C \cup D, N false)

B \subseteq C

and

δ > 0

. The roughness

r_{B}^{δ} false(D false)

and accuracy

A C C_{B}^{δ} false(D false)

of decision D with respect to attribute B are defined as

r_{B}^{δ} false(D false) = ρ_{B} false(D false) \times G K_{B} false(δ false)

A C C_{B}^{δ} false(D false) = 1 - r_{B}^{δ} false(D false)

The roughness or accuracy is an important measure for finding out informative attributes B from C .…”

Section: Hsi Band Selection Modelsmentioning

confidence: 99%

“…NRS uses the concept of neighbourhood granule to gather neighbourhood information. In this paper, we have implemented and compared four forward greedy hyperspectral band selection algorithms using (i) NRS [31], (ii) variable precision NRS (VPNRS) [34], (iii) consistency measure of NRS (CMNRS) [35] and (iv) granulation knowledge based NRS (GKNRS) [36]. In the experiments, the effectiveness of these techniques are analysed by using three real hyperspectral data sets.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations