Wrapper–Filter Feature Selection Algorithm Using a Memetic Framework

This correspondence presents a novel hybrid wrapper and filter feature selection algorithm for a classification problem using a memetic framework. It incorporates a filter ranking method in the traditional genetic algorithm to improve classification performance and accelerate the search in identifying the core feature subsets. Particularly, the method adds or deletes a feature from a candidate feature subset based on the univariate feature ranking information. This empirical study on commonly used data sets from the University of California, Irvine repository and microarray data sets shows that the proposed method outperforms existing methods in terms of classification accuracy, number of selected features, and computational efficiency. Furthermore, we investigate several major issues of memetic algorithm (MA) to identify a good balance between local search and genetic search so as to maximize search quality and efficiency in the hybrid filter and wrapper MA     

Intelligent system based on local linear wavelet neural network and recursive least square approach for breast cancer classification

A new learning technique for local linear wavelet neural network (LLWNN) is presented in this paper. The difference of the network with conventional wavelet neural network (WNN) is that the connection weights between the hidden layer and output layer of conventional WNN are replaced by a local linear model. A hybrid training algorithm of Error Back propagation and Recursive Least Square (RLS) is introduced for training the parameters of LLWNN. The variance and centers of LLWNN are updated using back propagation and weights are updated using Recursive Least Square (RLS). Results on extracted breast cancer data from University of Wisconsin Hospital Madison show that the proposed approach is very robust, effective and gives better classification.     

Local linear wavelet neural network based breast tumor classification using firefly algorithm

Breast cancer is the major cause of cancer deaths in women today and it is the most common type of cancer in women. This paper presents some experiments for classifying breast cancer tumor and proposes the use of firefly algorithm (FA) to improve the performance of Local linear wavelet neural network. This work in fact uses FA to optimize the parameters of local linear wavelet neural network. The experiments were conducted on extracted breast cancer data from University of Winconsin Hospital, Madison. The result has been compared with a wide range of classifiers to evaluate its performance. The evaluations show that the proposed approach is very robust, effective and gives better correct classification as compared to other classifiers.     

Phenology of vegetation in Southern England from Envisat MERIS terrestrial chlorophyll index (MTCI) data

Given the close association between climate change and vegetation response, there is a pressing requirement to monitor the phenology of vegetation and understand further how its metrics vary over space and time. This article explores the use of the Envisat MERIS terrestrial chlorophyll index (MTCI) data set for monitoring vegetation phenology, via its estimates of chlorophyll content. The MTCI was used to construct the phenological profile of and extract key phenological event dates from woodland and grass/heath land in Southern England as these represented a range of chlorophyll contents and different phenological cycles. The period 2003–2008 was selected as this was known to be a period with temperature and phenological anomalies. Comparisons of the MTCI-derived phenology data were made with ground indicators and climatic proxy of phenology and with other vegetation indices: MERIS global vegetation index (MGVI), MODIS normalized difference vegetation index (NDVI) and MODIS enhanced vegetation index (EVI). Close correspondence between MTCI and canopy phenology as indicated by ground observations and climatic proxy was evident. Also observed was a difference between MTCI-derived phenological profile curves and key event dates (e.g. green-up, season length) and those derived from MERIS MGVI, MODIS NDVI and MODIS EVI. The research presented in this article supports the use of the Envisat MTCI for monitoring vegetation phenology, principally due to its sensitivity to canopy chlorophyll content, a vegetation property that is a useful proxy for the canopy physical and chemical alterations associated with phenological change.     

Simultaneous MSMR and SSM/I observations and analysis of sea‐ice characteristics over the Antarctic region

Polar sea ice has been monitored quasi‐continuously over the last 30 years using passive microwave radiometers onboard three satellites in polar orbit, namely Nimbus‐5, Nimbus‐7 and Defense Meteorological Satellite Program (DMSP) series. A short overlap between Scanning Multichannel Microwave Radiometer (SMMR) on Nimbus‐7 and Special Sensor Microwave Imager (SSM/I) onboard DMSP allowed inter‐calibration of the two sensors leading to a consistent series of long‐term sea‐ice measurements since 1978. With the launch of Multifrequency Scanning Microwave Radiometer (MSMR) onboard OCEANSAT‐1 in the polar sun‐synchronous orbit during 1999, India developed the capability to monitor the polar sea ice on a regular basis. The concurrent availability of SSM/I and MSMR over the last few years presents a valuable opportunity to attempt an inter‐comparison of MSMR with SSM/I measurements and derived sea‐ice parameters.

In this paper, we present an indirect validation of the brightness temperatures (T _b) observed by MSMR with near‐simultaneous measurements from SSM/I over the Antarctic and Southern Polar Ocean regions. Simultaneous MSMR and SSM/I data from two contrasting seasons—summer and winter—for the 1999–2000 period have been used. Analysis includes a comparison of T _b scatterograms to achieve confidence in the quantitative use of the T _b data to derive various geophysical parameters, e.g. sea‐ice concentration and extent. Additionally, the T _b images produced by the two sensors are compared to establish the capability of MSMR in reliable two‐dimensional portrayal of all the sea and continental ice features over the Antarctic Region. Based on a regression analysis between MSMR observed T _b in different frequency channels and polarizations, and SSM/I‐derived sea‐ice concentration (SIC) values, we have developed algorithms to estimate SIC over the Southern Polar Ocean from MSMR data. The MSMR algorithms allow estimation of SIC with better than 10% rms error. MSMR SIC images faithfully capture the observed distribution of sea ice in all the sectors of the Southern Ocean both during summer and winter periods. Using the MSMR‐derived SIC, we have also derived monthly sea‐ice extent (SIE) estimates for a period extending for about 20 months from the beginning of the launch of MSMR. These estimates show excellent agreement with values derived from SSM/I. These analyses bring out the very high level of compatibility in the measurements and derived sea‐ice parameters produced by the two sensors.