فهرست مطالب mahdi eftekhari

Today, the great performance of Deep Neural Networks(DNN) has been proven in various fields. One of its most attractive applications is to produce artistic designs. A carpet that is known as a piece of art is one of the most important items in a house, which has many enthusiasts all over the world. The first stage of producing a carpet is to prepare its map, which is a difficult, time-consuming, and expensive task. In this research work, our purpose is to use DNN for generating a Modern Persian Carpet Map. To reach this aim, three different DNN style transfer methods are proposed and compared against each other. In the proposed methods, the Style-Swap method is utilized to create the initial carpet map, and in the following, to generate more diverse designs, methods Clip-Styler, Gatys, and Style-Swap are used separately. In addition, some methods are examined and introduced for coloring the produced carpet maps. The designed maps are evaluated via the results of filled questionnaires where the outcomes of user evaluations confirm the popularity of generated carpet maps. Eventually, for the first time, intelligent methods are used in producing carpet maps, and it reduces human intervention. The proposed methods can successfully produce diverse carpet designs, and at a higher speed than traditional ways.

A new image encryption scheme using the advanced encryption standard (AES), a chaotic map, a genetic operator, and a fuzzy inference system is proposed in this paper. In this work, plain images were used as input, and the required security level was achieved. Security criteria were computed after running a proposed encryption process. Then an adaptive fuzzy system decided whether to repeat the encryption process, terminate it, or run the next stage based on the achieved results and user demand. The SHA-512 hash function was employed to increase key sensitivity. Security analysis was conducted to evaluate the security of the proposed scheme, which showed it had high security and all the criteria necessary for a good and efficient encryption algorithm were met. Simulation results and the comparison of similar works showed the proposed encryptor had a pseudo-noise output and was strongly dependent upon the changing key and plain image.

Medical image segmentation is one of the most important steps in medical image analysis to improve diagnosis and findings. One of the most common segmentation methods in deep learning is the use of Unet networks. The presence of overlapping layers in the Unet decoder does not allow extracting information from deeper layers. Also, due to the limited range of the received field of convolution cores, long-range information and dependencies are not considered well. In this article, our goal is to place a structure in the area between encoder and decoder in the Unet model in order to fill the semantic gap between the encoder and decoder area and better extract features by paying attention to local and global features. This model makes the target region more prominent in different medical datasets. We have conducted our experiment on 6 medical data sets, and the results obtained in two evaluation criteria, Dice and Iou, show that our proposed model has better results than Unet and based methods.

Feature selection methods are known to be effective in improving the learning process. The purpose of a feature selection method is to identify relevant features and remove irrelevant features in order to obtain a suitable subset of features, so that the redundancy between the selected features is minimized. In multi-label data, if there is a correlation between features, it is possible that the amount of redundancy in the feature set is increased. The existence of redundancy between features along with the challenge of high dimensions of multi-label data can grow the computational calculations, decrease the accuracy and finally increase the probability of errors in the prediction and classification of multi-label data. In this article, with the aim of minimizing the redundancy of features, a multi-label feature selection algorithm is proposed considering the least squares regression model and sparse regularization. Finally, using a number of well-known multi-label data sets, the efficiency of the proposed method is verified and the results are compared with some common multi-label feature selection methods.

Intercropping is an important strategy for enhancing the sustainability of agriculture in the world by increasing food production and land use per unit area. Intercropping, which is defined as growing two or more species simultaneously in the same field during a growing season, is considered as an important strategy in developing sustainable production systems. In most cases, one of the plants introduced in intercropping is a plant from the legumes family. Chickpea is a plant that plays an important role in sustainable agriculture and soil fertility.

In order to evaluation effect of chickpea-barley intercropping on component yield, grain yield and weed control, an experiment was conducted at the Jahad Keshavarzi Shazand Research Station during 2019-2020 growing season. Experiment was conducted as a randomized complete block design with three replications. Different experimental treatment were: 1- chickpea sole cropping without weeding, 2- chickpea sole cropping with weeding, 3- barley sole cropping (normal weed control by farmers), 4- additive intercropping of 100% chickpea + 15% barley, 5- additive intercropping of 100% chickpea + 30% barley, 6- additive intercropping of 100% chickpea + 45% barley, 7- additive intercropping of 100% chickpea + 60% barley and 8- additive intercropping of 100% chickpea + 75% barley. Intercropping systems was evaluated by using land equivalent ratio (LER) index.

The results showed that most of the evaluated properties were affected by experimental treatments. The highest chickpea biological yield and grain yield (334 and 99 g m-2, respectively) belonged to the chickpea sole cropping with weeding treatment. Also, the highest barley biological yield and grain yield (423 and 195 g m-2, respectively) were obtained in its sole cropping treatment. In intercropping treatments, the highest grain yield of chickpea and barley (94 and 190 g m-2, respectively) were obtained in additive intercropping of 100% chickpea + 15% barley and additive intercropping of 100% chickpea + 75% barley treatments, respectively. Also, intercropping treatments significantly reduced the weed density and total dry weight. Thus, all intercropping treatments had less weed density and biomass in comparison to chickpea sole cropping without weeding. The lowest weed density and biomass (7.3 plant per m2 and 60 g m-2, respectively) belonged to the additive intercropping of 100% chickpea + 75% barley treatment. In addition, in all intercropping treatments, the land equivalent ratio and relative value indices were higher than one, which indicates the usefulness of intercropping.

The results of this study showed that, chickpea-barley intercropping systems decreased chickpea grain yield in comparison with sole cropping with weeding treatment but, in general, in the most of the intercropping treatments, was better than their sole cropping and associated with improving economic yield and land use efficiency. In general, it can be said that chickpea and barley intercropping, while better controlling chickpea weeds, has a higher total yield than sole cropping of both chickpea and barley species and increases land use efficiency.

In machine learning, transferring and generalizing the knowledge learned from one domain to another is one of the important and basic capabilities. Since supervised learning is not complete, the use of other methods, such as self-supervised learning methods, can be very helpful in domain generalization. In this paper, we present a method that, in addition to classify original images in order to learn data labels in a supervised process, attempts to classify images resulting from the application of discrete wavelet transform on the original images by generating pseudo-labels for them. This extra work as a self-supervision task can lead to learn useful features and a general image representation for images of different domains, which can greatly help to improve the problem of domain generalization. In the following, by combining self-supervised methods such as jigsaw puzzles and guessing the rotation angle with discrete wavelet transform, we show that this combination can improve the results for the domain generalization problem. In this paper, we used the well-known PACS, VLCS and office-Home datasets to perform experiments, and the results show that our proposed method can work better than advanced and state-of-the-art domain generalization methods.

The purpose of this study was to compare feedlot performance and carcass characteristics in Lori-bakhtiari and Romanov-Lori bakhtiari crossbred male lambs. In this opinion, 14 Lori bakhtiari and 14 Romanov- Lori bakhtiari male lambs were selected and their feedlot performance was studied in a completely random design during 100 days of trial. Lambs were fed by a total mixed ration (TMR) with forage to concentrate ratio of 30:70 in diets. Feed intake was recorded daily and the lambs were weighted monthly. At the end of the experiment, lambs were slaughtered and carcass analysis was done. The results of feedlot performances showed that crossbreeding had no significant effect on dry matter intake, final live weight, average daily gain and feed conversion ratio. Studying the carcass characteristics showed that crossbreeding had no significant effect on the warm carcass weight, empty rumen, lung and liver percent and carcass length (p>0.05), but the percent of heart, kidney and the internal fat weight were significantly affected by crossbreeding (p<0.05). The effect of genotype on the longissimus dorsi muscle area was significant also (p<0.05). Crossbreeding had a significant effect on the percentage of shank, round, brisket, flank and sirioin (P<0.05) and increased them. In general, although the effect of crossbreeding was not significant on the feedlot performance parameters, but improved the proportion of carcass components.

By the rapid progress of deep learning and its use in a variety of applications, however, deep networks have shown that they are vulnerable to adversarial examples. Recently developed researches show that using self-supervised learning (SSL) in various ways results in increasing network robustness. This paper examines the effect of a particular type of Contrastive SelfSupervised learning (CSSL) called Momentum Contrast (MoCo) on increasing network robustness to adversarial examples. For this purpose, MoCo is employed as a pre-text task and a deep network is pre-trained for this task. Then fine-tuning will cause to increase the robustness of the network against adversarial attacks examples. A new attack method is introduced based on MoCo and one of the Projected Gradient Descent (PGD) or Fast Gradient Sign (FGSM) methods that do not require any labeled data. Using this corrupted data and adversarial training method, a deep network is pre-trained and the representation provided by it is used to fine-tune downstream tasks that results in increasing network robustness. For an instance, the setup including Resnet50 structure, PGD attack, and MoCo-v1 shows 2.79%, 2%, and 1.35% of improvements comparing to the Jigsaw, Rotation, Selfie, respectively. More details of experiments and the improvements raised by MoCo are given in the results part and show the superiority of MoCo based models on CIFAR-10 and CIFAR-10-C datasets. Also, the obtained results for validating the robustness of proposed models against various noises with different corruption strengths, confirm the resistance of the proposed methods.

The purpose of this study was to investigate reproductive parameters in industrial dairy herds in Qazvin province. To conduct this study Herds were categorized into five capacity categories of dairy cows including herds by capacity less than 100, 100-199, 200-499, 500-999 and more than 1000 heads, and the data were analyzed in a completely unbalanced randomized design. The results of this study showed that there was no significant difference among treatments in terms of mean day of first insemination after calving. The dose of sperm consumed per each pregnancy in heifers was significantly different among the different category of capacities. (P≤ 0.05), It was better than others in treatments 1 and 2 (1.23 and 1.40%, respectively). The open days in the studied herds were 144.55 ±14.57 days on average and there was no significant difference among different treatments. Although there was no significant difference among the herds in terms of cattle fertility but fertility of heifers differed significantly among different treatment groups (P≤ 0.05), and treatments 1 and 2 were better than other capacity categories. Mean dry days in herds were 75.91 ± 0.31 days but there was no difference between treatments. Abortion rate was significantly different among treatments and treatment 1 had the lowest abortion rate (P≤ 0.05).  The results of this study showed that in general with increasing capacity of dairy farms, the reproductive parameters are negatively affected and the incidence of abortion increases.

One of the most important and complex management decisions in dairy farms that greatly affects the economy and profitability of the dairy herd is the issue of culling. Regarding the importance of the subject, this study aimed to investigate the culling status and causes of Holstein dairy cattle in Qazvin province. For this research, the culling data of 35 industrial dairy herds under the cover of Qazvin Farmerschr('39') Union were collected and analyzed during one year. Herds were categorized into five categories(treatments) of dairy cows including herds by capacity less than 100 (low), 100-200 (relatively low), 200-500 (average), 1000-500 (relatively high) and more than 1000 heads (high)., The culling data were analyzed in a completely randomized design in two main categories of forced and optional removal. The results showed that there was a significant difference in mean parity of the cows between the different treatments and mean parity decreased with increasing capacity of the studied cows. The mean parity of culling cows in the studied herds in Qazvin province was 3.92. There was no significant difference between different treatments in the percentage of fresh cows, primiparous cows as well as the percentage of productive cows (lactating and dry). Clinical mastitis accounted for 5.08% of total culling in dairy herds and there was no significant difference between treatments for this disease. BLV disease accounted for 1.25% of the total culling in the dairy herds studied and the difference between relatively high and high treatments were significant (P≤0.05). There were no significant differences between treatments for acidosis, abomasal displacement, ketosis / fatty liver, milk fever and bloat. The results of this study showed that in high capacity dairy herds (more than 500 heads), cows are culled earlier from the herd. The voluntary removal ratio in industrial dairy herds in Qazvin province was 27.67%. The most important factor for removal in these dairy farms was infertility and reproductive malformations with the ratio of 31.85%.

Recently many neural network methods have been proposed for multi-label classification in the literature. One of these recent researches is the multi-layer extreme learning machines (ML-ELM) in which stack auto encoders have been used for tuning the weights. However, ML-ELM suffers from three primary drawbacks: First, input weights and biases are chosen randomly. Second, the pseudo-inverse solution for calculating output weights will cause to increase the reconstruction error. Third, memory and execution time of transformation matrices are proportional to the number of hidden layers. In this paper multi-layer kernel extreme learning machine, that uses a linear combination of base kernels in each layer, is proposed for multi-label classification. The proposed approach effectively addresses the above-mentioned drawbacks. Furthermore, multi-label classification data inherently have multi-modal aspects due to the variety of labels assigned to each instance. Applying a combination of different kernels is the added advantage of ML-CK-ELM that leads to implicitly assess the inherent multi-modal aspects of multi-label data; each kernel can be effectively used to cover one of the modals better than the other kernels. The empirical study indicates that ML-CK-ELM represents competitive performance against other state-of-the-art methods, and experimental results over multi-label datasets verify the feasibility of ML-CK-ELM.

Recently, a number of Extreme Learning Machine (ELM) based training algorithms have been introduced for training deep neural network structures. ELM based Auto-Encoder (ELM-AE) is one such algorithm that has been used for making multilayer structures and tuning parameters of each layer. In a simple ELM-AE training algorithm, the weights of the first layer are initialized randomly. This issue is a leading factor in producing reconstruction error. The frequent use of ELM-AE in deep network layers results in propagating such errors through deep structures and in decreasing performance as a consequent. In this paper, we introduce a multilayer structure and a new learning algorithm to train it that prevents error propagation. In order to boost the performance of the model, the parameters in the first layer are initialized by a novel type of ELM-AE called Repeated-AE (RAE) rather than by a random selection method. This RAE-based technique determines the parameters in the first layer far better than do the other ELM-AE existed methods. Next, a single hidden layer ELM is applied for handling the classification task. Experimental results for data classification show that the proposed method outperforms some other methods in terms of the average accuracy over all datasets by amounts of 4%, 26%, 17% and 31%. Eventually, so as to verify the performance of the proposed multilayer ELM-AE in application, we used this model to reconstruct images. The reconstructed images obtained by our approach appeared visually a lot better compared to those obtained by the other methods do.

In this paper, an effective method for regression is presented in which a variety of fuzzy clustering methods and concepts of Hesitant fuzzy sets are used. First, the fuzzy clustering algorithm is applied to the data, and after projecting the cluster membership function on different features, the number of clusters of fuzzy sets is obtained on each dimension (or feature). We then consider these fuzzy sets as a hesitant fuzzy set on each feature, and we obtain the Hesitant Fuzzy Correlation Coefficient Matrix (HFCCM) for the attributes. Subsequently, a nonlinear mapping based on the principal components analysis of the HFCCM is used to convert the dataset's features into new features. Finally, the new extracted features are assigned to the fuzzy clustering algorithm and a Sugeno fuzzy regression system is fitted. The proposed method was compared with some other methods to several regression datasets. The results of the experiments indicate that the proposed method is successful in extracting and reducing the characteristics, as well as increasing the regression accuracy. Also, the number of rules of the fuzzy regression model in the proposed method is fairly low.

DNA micro-array datasets play crucial role in machine learning and recognition of various kinds of cancer structures. Micro-array datasets are typically characterized by the high number of features and the small number of samples. Such problems may result in overfitting and low prediction accuracy of classifiers due to the irrelevant features, and therefore, they are considered as a challenging task in machine learning. The direct way to deal with such challenges is dimensionality reduction of data. In this regard, feature selection method acts as an effective solution for dimensinality reduction and increasing efficiency of learning algorithms. In this paper, by using the concept of “the basis for the DNA micro-array datasets”, a new feature selection method is introduced. To be more specific, rather than utilizing the entire micro-array dataset for tackling the problem of feature selection, a basis that is a much more smaller subset of the micro-array dataset is used. This method is based on subspace learning and matrix factorization. Finally, by making use of the DNA micro-array datasets, the effectiveness of the proposed method is evaluated, and the obtained results are compared with some state-of-the-art supervised feature selection methods.

Software defects detection is one of the most important challenges of software development and it is the most prohibitive process in software development. The early detection of fault-prone modules helps software project managers to allocate the limited cost, time, and effort of developers for testing the defect-prone modules more intensively.  In this paper, according to the importance of software defects detection, a method based on fuzzy sets and evolutionary algorithms is proposed. Due to the imbalanced nature of software defect detection datasets, benefits of fuzzy clustering algorithms were used to data sampling and more attention to the minority class. This method is a combined algorithm which, firstly has used fuzzy c-mean clustering as weighted bootstrap sampling. Weight of data (their membership’s degrees) increases for minority class. In the next step, the subtractive clustering algorithm is applied to produce the classifier which was trained by produced data in the previous step. The binary genetic algorithm was utilized to select appropriate features. The results and also comparisons with eight popular methods in software defect detection literature, show an acceptable performance of the proposed method. The experiments were performed on ten real-world datasets with a wide range of data sizes and imbalance rates. Also T-test is used as the statistical significance test for pair wise comparison of our proposed method against the others. The final results of T-test are shown in tables for three performance measures (G-mean, AUC and Balanced) over various datasets. (As the obtained results apparently show our proposed method has the ability to improve three aforementioned performance criteria simultaneously). Some methods just have improved the G-mean measure while the AUC and Balance criteria have lower values than the others. Securing a high level of three performance measures simultaneously illustrates the ability of our proposed algorithm for handling the imbalance problem of software defects detection datasets.

High dimensional microarray datasets are difficult to classify since they have many features with small number ofinstances and imbalanced distribution of classes. This paper proposes a filter-based feature selection method to improvethe classification performance of microarray datasets by selecting the significant features. Combining the concepts ofrough sets, weighted rough set, fuzzy rough set and hesitant fuzzy sets for developing an effective algorithm is the maincontribution of this paper. The mentioned method has two steps, in the first step, four discretization approaches areapplied to discretize continuous datasets and selects a primary subset of features by combining of weighted rough setdependency degree and information gain via hesitant fuzzy aggregation approach. In the second step, a significancemeasure of features (defined by fuzzy rough concepts) is employed to remove redundant features from primary set.The Wilcoxon Signed Ranked tes (A Non-parametric statistical test) is conducted for comparing the presented methodwith ten feature selection methods across seven datasets. The results of experiments show that the proposed methodis able to select a significant subset of features and it is an effective method in the literature in terms of classificationperformance and simplicity.

Nanotechnology is a new technology that is used in various fields of science such as agriculture. One of the sectors related to agriculture that this technique taken a step forward is livestock husbandry and their nutrition. Minerals are One of the most important and concerned area in livestock nutrition. Zinc is an important micromineral and have been conducted a lot of research on the use of nanotechnology in livestock nutrition on it. Particle size of nanoparticles of zinc oxide is between 1 to 100 nm. It promote growth, increases production, and is effective on immunity and reproduction. Because of the high bioavailability of zinc oxide nanoparticles, these supplements compared with the usual sources can be used in small amounts in animal rations and reduces the concerns about environmental pollution. However determination of the consumption level in livestock diets need to more complementary studies. Material presented in this paper Surveys the feasibility of using the Zn nanoparticles as a dietary supplement in livestock nutrition.

In this paper, a robust linear quadratic regulator (LQR) based Reinforcement learning method is designed for a four degree of freedom inverted pendulum. The considered system contains a four degree of freedom inverted pendulum with a concentrated mass at the tip of it. The bottom of inverted pendulum is moved in x-y plane in x and y directions. For tracking control of two angles of inverted pendulum, two plane forces are applied in x and y directions at the bottom of pendulum. The governing equations of the system are derived using the Lagrange method and then a robust linear quadratic regulator (LQR) based Reinforcement learning controller is designed. The inverted pendulum is learned for a range of different angles, different lengths and different masses. The parametric uncertainties are defined as various lengths and masses of inverted pendulum and the disturbances are defined as impact and continuous forces which are applied on the inverted pendulum. After learning, the controller can learn online the system for any arbitrary angle, length, mass or disturbance which are not learned in the defined range. Numerical results show that the good performance of the reinforcement learning controller for the inverted pendulum in the presence of structural and parametric uncertainties, impact and continuous disturbances and sensor noises.

Cutaneous Leishmaniasis is considered as an important health problem in Iran. This disease begins as small papules and then they gradually grow and turn into wounds. Since the epidemiological study of this problem is effective in preventing and controlling it, the current research was conducted on epidemiological pattern of Cutaneous Leishmaniasis in the province of Fars, Iran (2010-2014)
This cross-sectional study was conducted on the registered information (20601 patients) and documents of the patients having problem in Fars province remedial and health care centers over the last 5 years. First, the registered demographic and epidemiological data of patients were extracted and then analyzed through by using SPSS software.
A total of 20601 patients including 10607 males (51%) and 9994 females (49%) participated this study. Patients were in the age range of 20-29 years old and minority of them was 5-9 years old. Most lesions were on the hand and leg of the patients. It was also found that there was a significant relationship between age groups and types of Leishmaniasis (P According to this is investigation, there was an ascending trend in disease frequency which was based on the descending process. This indicates promotion of educational- hygiene status as well as observance of personal hygiene principals.

In classification problems, we often encounter datasets with different percentage of patterns (i.e. classes with a high pattern percentage and classes with a low pattern percentage). These problems are called “classification Problems with imbalanced data-sets”. Fuzzy rule based classification systems are the most popular fuzzy modeling systems used in pattern classification problems. Rule weights have been usually used to improve the classification accuracy and fuzzy versions of confidence and support merits have been widely used for rules weighting in fuzzy rule based classifiers. In this paper, we propose an evolutionary approach based on genetic programming to generate weighting expressions. For producing expressions confidence, support, lift and recall merits are used as terminals of genetic programming. Experiments are performed over 20 imbalanced KEEL's datasets and the results are analyzed using statistical tests. The results show that the proposed method improves the classification accuracy of FRBCS.

Bacterial meningitis is a known infectious disease which occurs at early ages and should be promptly diagnosed and treated. Bacterial and aseptic meningitis are hard to be distinguished. Therefore, physicians should be highly informed and experienced in this area. The main aim of this study was to suggest a system for distinguishing between bacterial and aseptic meningitis, using fuzzy logic. In the first step, proper attributes were selected using Weka 3.6.7 software. Six attributes were selected using Attribute Evaluator, InfoGainAttributeEval, and Ranker search method items. Then, a fuzzy inference engine was designed using MATLAB software, based on Mamdani’s fuzzy logic method with max-min composition, prod-probor, and centroid defuzzification. The rule base consisted of eight rules, based on the experience of three specialists and information extracted from textbooks. Data were extracted from 106 records of patients with meningitis (42 cases with bacterial meningitis) in order to evaluate the proposed system. The system accuracy, specificity, and sensitivity were 89%, 92 %, and 97%, respectively. The area under the ROC curve was 0.93, and Kappa test revealed a good level of agreement (k=0.84, P<0.0005). According to the results, the suggested fuzzy system showed a good agreement and high efficiency in terms of distinguishing between bacterial and aseptic meningitis. To avoid unnecessary antibiotic treatments, patient hospitalization, and misdiagnosis of bacterial meningitis, such systems are useful and highly recommended. However, no system has been yet introduced with 100% correct output and further research is required to improve the results.

In this study، 32 multiparous non lactating Holstein cows were used to investigate the effects of chromium-l-methionine (Cr-Met) supplementation and dietary energy source on production and nutrient digestibility during the periparturient period. Cows were fed a total mixed ration from 28 days before anticipated calving through 28 days after calving. The study was designed as a randomized complete block design with 2 (energy sources) × 2 (Cr-Met levels) factorial arrangement. Energy sources in the prepartum were fish oil and grain، and were fish oil and palm fat powder in the postpartum. The Cr-Met was supplemented at dosages of 0 or 0. 08 mg of Cr/kg of metabolic body weight. Fish oil Supplementation had no significant effect on prepartum dry matter intake (DMI) but tended to decrease postpartum DMI (P< 0. 1). Supplemental Cr-Met increased prepartum and postpartum DMI significantly (P<0. 05). The effect of treatments on milk production was not significant. No differences were detected in milk composition except for fat percentage in which cows were fed the fish oil had lower percentage of fat compare to others. Supplemental Cr-Met had no significant effect on nutrient digestibility in the prepartum and postpartum. Use of fish oil had no effect on nutrient digestibility in the prepartum، but decreased digestibility of dry matter، organic matter، and crude protein in the postpartum (P<0. 05). Results of this study indicate no significant interaction between chromium supplementation and energy source.

During the last decade, applying feature selection methods in bioinformatics has become an essential necessity for model building. This is due to the high dimensional nature of many modeling tasks in bioinformatics of them being Single Nucleotide Polymorphisms (SNPs) selection. In this paper, we propose three hybrid feature selection methods named CNNFS, Ck-NNFS, and CRRFS, which are combinations of filter and wrapper techniques. In our methods, filter techniques were applied to remove the irrelevant/redundant features as the first step. Then in the second step, wrapper techniques were exploited to refine the primary feature subset obtained from the first step. Neural Network, k-Nearest Neighbor, and Ridge Regression were injected in the wrapper phase as induction algorithms. Since pure wrapper methods take a long time to run on high dimensional data, we compared our methods with three well-known filter methods, and skipped the wrappers. The results vividly show the performance of hybrid methods in addition to their dimensionality reduction ability in SNPs selection. The CRRFS algorithm brought the most satisfactory results regarding to the precision of recognizing candidate SNPs, and the recall of them in the final SNPs subset.

we propose in this article a new hybrid method for modeling accurate fuzzy rule based classification systems. The new method is a combination of manifold based data mapping method, a heuristic fuzzy rule based construction method and an evolutionary based rule weighting approach. Manifold based data mapping method considers the intricate geometric relationships that may exist among the data and computes a new representation of data that optimally preserves local neighborhood information in a certain sense. Although this new representation does not secure the interpret ability of obtained fuzzy models, the main intention of this research is to improve the classification accuracy significantly. Experiments on some well-known datasets are performed to show the performance of the new proposed approach. Some nonparametric statistical tests are used to analysis the results obtained in experiments. Experimental results confirm the effectiveness of our proposed method in improvement of the classification accuracy.

Fuzzy Decision Tree (FDT) classifiers combine decision trees with approximate reasoning offered by fuzzy representation to deal with language and measurement uncertainties. When a FDT induction algorithm utilizes stopping criteria for early stopping of the tree''s growth, threshold values of stopping criteria will control the number of nodes. Finding a proper threshold value for a stopping criterion is one of the greatest challenges to be faced in FDT induction. In this paper, we propose a new method named Iterative Deepening Fuzzy ID3 (IDFID3) for FDT induction that has the ability of controlling the tree’s growth via dynamically setting the threshold value of stopping criterion in an iterative procedure. The final FDT induced by IDFID3 and the one obtained by common FID3 are the same when the numbers of nodes of induced FDTs are equal, but our main intention for introducing IDFID3 is the comparison of different stopping criteria through this algorithm. Therefore, a new stopping criterion named Normalized Maximum fuzzy information Gain multiplied by Number of Instances (NMGNI) is proposed and IDFID3 is used for comparing it against the other stopping criteria. Generally speaking, this paper presents a method to compare different stopping criteria independent of their threshold values utilizing IDFID3. The comparison results show that FDTs induced by the proposed stopping criterion in most situations are superior to the others and number of instances stopping criterion performs better than fuzzy information gain stopping criterion in terms of complexity (i.e. number of nodes) and classification accuracy. Also, both tree depth and fuzzy information gain stopping criteria, outperform fuzzy entropy, accuracy and number of instances in terms of mean depth of generated FDTs.