Using the unlabeled data in the semi-Supervised learning can significantly improve the accuracy of Supervised Classification. But in some cases, it may dramatically reduce the accuracy of the Classification. The reason of such degradation is incorrect labeling of unlabeled data. In this article, we propose the method for high confidence labeling of unlabeled data. The base classifier in the proposed algorithm is the support vector machine. In this method, the labeling is performed only on the set of the unlabeled data that is closer to the decision boundary from the threshold. This data is called informative data. the adding informative data to the training set has a great effect to achieve the optimal decision boundary if the predicted label is correctly. The Epsilon-neighborhood Algorithm (DBSCAN) is used to discover the labeling structure in the data space. The comparative experiments on the UCI dataset show that the proposed method outperforms than some of the previous work to achieve greater accuracy of the self-training semi-Supervised Classification.

Data stream is a sequence of data generated from various information sources at a high speed and high volume. Classifying data streams faces the three challenges of unlimited length, online processing, and concept drift. In related research, to meet the challenge of unlimited stream length, commonly the stream is divided into fixed size windows or gradual forgetting is used. Concept drift refers to changes in the statistical properties of data, and is divided into four categories: sudden, gradual, incremental, and recurring. Concept drift is generally dealt with by periodically updating the classifier, or employing an explicit change detector to determine the update time. These approaches are based on the assumption that the true labels are available for all data samples. Nevertheless, due to the cost of labeling instances, access to a partial labeling is more realistic. In a number of studies that have used semi-supervisory learning, the labels are received from the user to update the models in form of active learning. The purpose of this study is to classify samples in an unlimited data stream in presence of concept drift, using only a limited set of initial labeled data. To this end, a semi-Supervised ensemble learning algorithm for data stream is proposed, which uses entropy variation to detect concept drift and is applicable for sudden and gradual drifts. The proposed model is trained with a limited initial labeled set. In occurrence of concept drift, the unlabeled data is used to update the ensemble model. It does not require receiving the labels from the user. In contrast to many of the current studies, the proposed algorithm uses an ensemble of K-NN classifiers. It constructs a group of clustering-based Classification models, each of which is trained on a batch of data. On receiving each new sample, first it is determined whether the data sample is an outlier or not. If the data is included in a cluster, the sample class is determined by majority voting. When a window of the stream is received, the possibility of concept drift is examined based on entropy variation, and the classifier is updated by a semi-Supervised approach if necessary. The model itself determines the required data labels. The proposed method is capable of detecting concept drift in data, and improving its accuracy via updating the learning model with appropriate samples received from the stream. Therefore, the proposed method only requires a small initial labeled data. Experiments are performed using five real and synthetic datasets, and the model performance is compared to three other approaches. The results show that the proposed method is superior in terms of precision, recall and F1 score compared to other studies.

Nowadays, satellite data are considered as one of the most important sources for production of landuse/cover maps. Several digital Classification methods have been applied and evaluation and comparison of the methods for Classification has a great degree of importance tasks. In this research, some Classification methods, e.g. K-Nearest Neighbour, Maximum Likelihood and Box Classification were used for evaluating land use maps in Kavar area which is located in the south of Shiraz. Also, the GIS were applied in this research for correcting the wrong pixels of some classes. Evaluations of the accuracy were performed using over-all accuracy and KAPPA index which are based on analysis of error matrix. According to these results, the over-all accuracy for Maximum Likelihood, K-Nearest Neighbour and Box Classification has been reached to 94.7, 61.7 and 60.4 percent respectively. Results of this research show that the capability of Maximum Likelihood is higher than the other methods.

In this research, the application of Artificial Neural Network or MLP method in the process Assignment of relevé-groups/ plant communities allocation was evaluated using Buxus hyrcana forests database. For this purpose, firstly, the ecological and sociological groups of B. hyrcana were determined using TWINSPAN and Braun-Blanquet method, respectively. The results of both numerical and expert based Classification dendrogram of the B. hyrcana communities, which included seven levels of Classification as primary groups/plant communities, were introduced to MLP. Then, with assignments in three sets of training (70%), test (15%) and validation (15%), the MLP Classification was performed on each level of the two dendrograms. The results showed that by increasing the level of Classification, the degree of adaptation of the MLP result with primary ecological and sociological groups of TWINSPAN (99% to 60%) and Braun-Blanquet (98% to 68%) decreased from the cutoff level of 1 to 7. Results of sensitivity and kappa cross tab coefficients, except in 7 cut level, imply that the quality of MLP groups based on TWINSPAN primary ecological group is better than the primary Braun-Blanquet groups. The MLP results in Buxus hyrcana plant communities Classification were consistent with the results of TWINSPAN (90%) and Braun-Blanquet (89%) ecological/syntaxa groups at the fifth cut level of both dendrograms concluding the reliability of MLP application for Classification of plant communities. So, our result confirms that MLP can be introduced as a suitable method for the assignment of releves to plant communities.

Automatic modulation Classification (AMC) in detected signals is an intermediate step between signal detection and demodulation, and is also an essential task for an intelligent receiver in various civil and military applications. In this paper, we propose a semi-Supervised Large margin AMC and evaluate it on tracking the received signal to noise ratio (SNR) changes to classify most popular single carrier modulations in non-stationary environments. To achieve this objective, two structures for self-training of large margin classifiers were developed in additive white Gaussian noise (AWGN) channels with priori unknown SNR. A suitable combination of the higher order statistics (HOS) and instantaneous characteristics of digital modulation are selected as effective features. We investigated the robustness of the proposed classifiers with respect to different SNRs of the received signals via simulation results and we have shown that adding unlabeled input samples to the training set, improve the tracking capacity of the presented system to robust against environmental SNR changes. The performance of the automatic modulation classifier is presented in the form of k-fold cross-validation test, Classification accuracy and confusion matrix methods. Simulation results show that the proposed approach is capable to classify the modulation class in unknown variable noise environment at even low SNRs.