An emerging technique to improve classification performance is to build several different CLASSIFIERs, and then to combine them, known as MULTIPLE CLASSIFIER SYSTEMS or ensemble classification SYSTEMS. The design process of an ensemble system generally involves two steps: the collection of an ensemble of CLASSIFIERs and the design of the combination rule. Researchers in various fields including pattern recognition, machine learning and statistics have examined the use of ensemble SYSTEMS. Nabavi-Kerizi and Kabir provided a review of ensemble classification, where combining techniques have been mainly considered. However, the trend of recent papers in this active field shows that the ensemble SYSTEMS have focused on different ways to design the ensemble of CLASSIFIERs. In this paper, first we aim to establish a framework for different approaches. Based on this architecture, each approach has been introduced in details. Combination methods are then described in brief. At the end, active research areas in the field of ensemble learning are presented.

Optical and polarimetric synthetic aperture radar (PolSAR) earth observations offer valuable sources of information for agricultural applications and crop mapping. Various spectral features، vegetation indices and textural indicators can be extracted from optical data. These features contain information about the reflectance and spatial arrangement of crop types. By contrast، PolSAR data provide quad-polarization backscattering observations and target decompositions، which give information about the structural properties and scattering mechanisms of different crop types. Combining these two sources of information can present a complementary data set with a significant number of spectral، textural، and polarimetric features for crop mapping and monitoring. Moreover، a temporal combination of both observations may lead to obtaining more reliable results compared to the use of single-time observations. However، there are several challenges in cropland classification using this large amount of information. The first challenge is the possibility correlation among some optical features or radar features which leads to redundant features. Moreover، some optical or radar features may have a low relevancy with some or all crop types. These two challenges cause to increase complexity and computational load of classification. In addition، when the ratio of number of samples to the number of features is very low، the curse of dimensionality may be occur. Another challenge of classification is the imbalanced distribution among various crop types، the so called imbalanced data. Various CLASSIFIER have been presented for cropland classification from optical and radar data. Among these CLASSIFIERs، the MULTIPLE CLASSIFIER SYSTEMS (MCS) especially the random forest (RF). The main aim of this paper is an alternative to RF which is able to solve these two challenges، the curse of dimensionality and the imbalanced data، simultaneously. The proposed MCSs have other modifications in feature selection and fusion steps of RF. These two methods called as balanced filter forest (BFF) and cost-sensitive filter forest (CFF). The study area of this paper was the southwest district of Winnipeg، Manitoba، Canada، which is covered by various annual crops. The data used in this paper were bi-temporal optical and radar images acquired by RapidEye satellites and the UAVSAR system. RapidEye is a spaceborne satellite، which has five spectral channels: blue (B)، green (G)، red (R)، NIR and RE. In this paper، two optical images were collected on 5 and 14 July 2012. Both these images were orthorectified on the local North American 1983 datum (NAD-83) with a spatial resolution of about 5 m. The UAVSAR sensor is an airborne SAR sensor، which operates in the L-band frequency in full polarization mode (i. e.، HH، HV، VH and VV). The radar images used in this paper were simultaneously acquired with the optical images. They were orthorectified on the World Geodetic System 1984 datum (WGS-84) with an SRTM3 digital elevation model. They were also multilooked by 2 pixels in azimuth and 3 pixels in range directions. Moreover، the de-speckling process، using a 5 × 5 boxcar filter، was applied to the data in order to alleviate the speckle effect. The spatial resolution of these images was then approximately 15 m. The results indicated that the proposed methods could increase the overall accuracy to 10% and the speediness to 6 times more than the classical RF method.

A fully polarimetric synthetic aperture radar (POLSAR) image can provide a high-dimensional data. This large amount of information can increase the overall accuracy of land-cover classification. But increasing the data dimensions if inadequately number of training samples may increase the complexity and cause the curse of dimensionality phenomenon. One of the strategies for solving this problem is the use of MULTIPLE CLASSIFIER SYSTEMS (MCS) that has the capability of divide and conquer to the large data as compared to the individual CLASSIFIERs. In addition, some of MCS methods using the weak and unstable CLASSIFIERs such as decision tree (DT) and neural network (NN) can obtain the high accuracy in high-dimensional data. The objective of this paper is also to use several popular MCS methods such as adaboost, bagging and random forests in order to improve the accuracy of land-cover classification from high-dimensional PolSAR images. The data used in this paper are Radarsat-2 image from San Francisco Bay and AIRSAR image of Flevoland. For classifying two these images, 69 polarimetric features were extracted from them. Two CLASSIFIERs of DT and NN were chosen as the base CLASSIFIERs of adaboost and bagging methods. In the next, the MCS methods were compared with the individual CLASSIFIERs of DT and NN. The results indicated the higher overall accuracy of MCS methods between 5%–8% for classifying first image and 9%–16% for classifying second image. Even, the producer's accuracy and user's accuracy of MCS methods at all classes were more than the those of individual CLASSIFIERs. So that at some classes, the difference was between 20% to even near 50%. These results confirmed that the MCS methods not only can produce higher overall accuracy at land-cover classification, but also they have the higher efficiency and reliability at discriminate individual classes.

One of the tools of artificial intelligence is the adaptive neural-fuzzy inference system (ANFIS), which is used in this article to build an intrusion detection system and we call it the neural-fuzzy CLASSIFIER. The Intrusion Detection System based on ANFIS is an anomaly based intrusion detection system that uses fuzzy logic and neural network to detect if malicious activity is taking place on a network. This paper describes the architecture of the ANFIS and its components. The sample fuzzy rules are developed for some kinds of attacks and the testing results with actual network data are described. Our experiments and evaluations were performed with the NSLKDD intrusion detection dataset which is a version of the KDD Cup99 intrusion detection evaluation dataset prepared and managed by MIT Lincoln Laboratories. Finally, this paper tries to show the efficiency of the designed model by examining the performance of the "neural-fuzzy adaptive inference system" model on a standard and comprehensive set.

Designing an effective criterion for selecting the best rule is a major problem in the process of implementing Fuzzy Learning CLASSIFIER (FLC) SYSTEMS. Conventionally confidence and support or combined measures of these are used as criteria for fuzzy rule evaluation. In this paper new entities namely precision and recall from the field of Information Retrieval (IR) SYSTEMS is adapted as alternative criteria for fuzzy rule evaluation. Several Different combinations of precision and recall are redesigned to produce a metric measure. These newly introduced criteria are utilized as a rule selection mechanism in the method of Iterative Rule Learning (IRL) of FLC. In several experiments, three standard datasets are used (0 compare and contrast the novel IR based criteria with other previously developed measures. Experimental results illustrate the effectiveness of the proposed techniques in terms of classification performance and computational efficiency.

This study was conducted during summer and winter of 2018- 2019 in the agricultural research field of Shahid Chamran University. Experimental design was split- plot based on RCBD with three replications. The main plot was the type of agricultural system in three levels including conventional (Conv), organic (Org) and sustainable (Sust) (integrated between Conv and Org) and sup- plot was the type of pre- cultivated crop in sequence with wheat including cultivation of mung bean (M- W), corn (C- W), sesame (S- W) and fallow (F- W). Yield quantity (yield and its component) and quality (grain protein), an estimate of photosynthesis matter transfer index of wheat and soil organic carbon (SOC) after one double-cropping were measured. The result showed that the highest (545.04 g/m2) and the lowest (409.28 g/m2) seed yields were obtained in Conv and Org respectively. In contract, with the changing type of system from Conv to Org, grain protein was increased significantly (from 8.3 to 9.6 %). In addition, the highest (535.47 g/m2) yield of wheat was obtained from M- W double cropping. On the other hands the highest remobilization and current photosynthesis matter were obtained in the organic agricultural system with M- W and conventional with M- W double cropping. The situation of SOC showed that the highest (33.18 mg/g) SOC was obtained in the organic agricultural system with C- W double cropping. The reason for improving SOC in the organic and sustainable agricultural system was application of organic matter (compost and vermicompost) and crop residue management. Totally, from the crop ecology point of view, sustainable agricultural method with a sequence of M- W was the most desirable system.

One efficient approach in classification is using a set of individual CLASSIFIERs and then combining their outputs, usually knows as ensemble classification or MULTIPLE CLASSIFIER system. In this paper, an ensemble classification system based on the random subspace approach is employed for diagnosis of endometriosis, in which individual CLASSIFIERs of the ensemble system are trained with different feature subsets. Finally, for classifying an unknown test sample, CLASSIFIERs’ outputs are fused using the majority voting combination rule.