Small Number Of Training Samples Research Articles

Short-Term Weather Forecast Skill of Artificial Neural Networks

Abstract We evaluate the short-term weather forecast performance of three flavors of artificial neural networks (NNs): feed forward back propagation, radial basis function, and generalized regression. To prepare the application of the NNs to an operational setting, we tune NN hyperparameters using over two years of historical data. Five objective guidance products serve as predictors to the NNs: North American Mesoscale and Global Forecast System model output statistics (MOS) forecasts, the High-Resolution Rapid Refresh (HRRR) model, National Weather Service forecasts, and the National Blend of Models product. We independently test NN performance using 96 real-time forecasts of temperature, wind, and precipitation across 11 U.S. cities made during the WxChallenge, a weather forecasting competition. We demonstrate that all NNs significantly improve short-range weather forecasts relative to the traditional objective guidance aids used to train the networks. For example, 1-day maximum and minimum temperature forecast error is 20%–30% lower than MOS. However, NN improvement over multiple linear regression for short-term forecasts is not significant. We suggest this may be attributed to the small number of training samples, the operational nature of the experiment, and the short forecast lead times. Regardless, our results are consistent with previous work suggesting that applying NNs to model forecasts can have a positive impact on operational forecast skill and will become valuable tools when integrated into the forecast enterprise. Significance Statement We used approximately two years of historical weather data and objective forecasts for a number of cities to tune a series of artificial neural networks (NNs) to forecast 1-day values of maximum and minimum temperature, maximum sustained wind speed, and quantitative precipitation. We compare forecast error against common objective guidance and multiple linear regression. We found that the NNs exhibit about 25% lower error than common objective guidance for temperature forecasting and 50% lower error for wind speed. Our results suggest that NNs will be a valuable contributor to improving weather forecast skill when adopted into the existing forecast enterprise.

DC-DC Buck circuit fault diagnosis with insufficient state data based on deep model and transfer strategy

As the common power supply of electronic systems, the fault of DC-DC Buck circuit will trigger the faulty power supply which eventually cause serious problems in system tasks. Therefore, it is of great significance to accurately identify the faults of DC-DC Buck circuit. Due to the high cost of collecting circuit fault data set, the lack of effective fault data will lead to the inability to establish an intelligent fault diagnosis model. In this paper, the model of source circuit with sufficient data is used to solve the problem of fault diagnosis of target circuit with lack of data. A deep transfer kernel extreme learning machine auto encoder (DKEA) model is designed. The activation function of Gaussian error linear units (GELU) is used to describe the probability of neuron input, and the kernel extreme learning machine is employed as the classifier to complete the diagnosis task. First, only the output voltage signal of the circuit is collected to extract its primary features in time and frequency domain, and then the primary features are mined and classified by using the deep extreme learning machine auto encoder (ELM-AE) model. Sufficient source domain data is used to train the deep ELM-AE model, and the parameters obtained are used as the initialization parameters of the target domain diagnosis model. With the help of transfer learning, a small number of training samples in the target domain are used to fine-tune the deep ELM-AE model to adapt to the remaining testing samples in the target domain. The experimental results proves that the proposed method can effectively achieve the transfer diagnosis between different DC-DC Buck circuits based on the source domain data.

WBC Image Segmentation Based on Residual Networks and Attentional Mechanisms.

White blood cell (WBC) morphology examination plays a crucial role in diagnosing many diseases. One of the most important steps in WBC morphology analysis is WBC image segmentation, which remains a challenging task. To address the problems of low segmentation accuracy caused by color similarity, uneven brightness, and irregular boundary between WBC regions and the background, a WBC image segmentation network based on U-Net combining residual networks and attention mechanism was proposed. Firstly, the ResNet50 residual block is used to form the main unit of the encoder structure, which helps to overcome the overfitting problem caused by a small number of training samples by improving the network's feature extraction capacity and loading the pretraining weight. Secondly, the SE module is added to the decoder structure to make the model pay more attention to useful features while suppressing useless ones. In addition, atrous convolution is utilized to recover full-resolution feature maps in the decoder structure to increase the receptive field of the convolution layer. Finally, network parameters are optimized using the Adam optimization technique in conjunction with the binary cross-entropy loss function. Experimental results on BCISC and LISC datasets show that the proposed approach has higher segmentation accuracy and robustness.

Contour-Based Wild Animal Instance Segmentation Using a Few-Shot Detector.

Simple SummaryBiodiversity monitoring is one of the primary means of ecological research. With the development of convolutional neural networks (CNNs) in the field of instance segmentation, CNNs are also used for species recognition. Almost all species recognition models apply pixel-based instance segmentation to recognize animal species. However, pixel-based instance segmentation models require a large number of annotations and labels, which makes them time-consuming and unsuitable for small datasets. Therefore, in this paper, we propose a contour-based wild animal instance segmentation model that can reach a balance between accuracy and real-time performance.Camera traps are widely used in wildlife research, conservation, and management, and abundant images are acquired every day. Efficient real-time instance segmentation networks can help ecologists label and study wild animals. However, existing deep convolutional neural networks require a large number of annotations and labels, which makes them unsuitable for small datasets. In this paper, we propose a two-stage method for the instance segmentation of wildlife, including object detection and contour approximation. In the object detection stage, we use FSOD (few-shot object detection) to recognize animal species and detect the initial bounding boxes of animals. In the case of a small wildlife dataset, this method may improve the generalization ability of the wild animal species recognition and even identify new species that only have a small number of training samples. In the second stage, deep snake is used as the contour approximation model for the instance segmentation of wild mammals. The initial bounding boxes generated in the first stage are input to deep snake to approximate the contours of the animal bodies. The model fuses the advantages of detecting new species and real-time instance segmentation. The experimental results show that the proposed method is more suitable for wild animal instance segmentation, in comparison with pixel-wise segmentation methods. In particular, the proposed method shows a better performance when facing challenging images.

Analysis of a deep learning-based method for generation of SPECT projections based on a large Monte Carlo simulated dataset

BackgroundIn recent years, a lot of effort has been put in the enhancement of medical imaging using artificial intelligence. However, limited patient data in combination with the unavailability of a ground truth often pose a challenge to a systematic validation of such methodologies. The goal of this work was to investigate a recently proposed method for an artificial intelligence-based generation of synthetic SPECT projections, for acceleration of the image acquisition process based on a large dataset of realistic SPECT simulations.MethodsA database of 10,000 SPECT projection datasets of heterogeneous activity distributions of randomly placed random shapes was simulated for a clinical SPECT/CT system using the SIMIND Monte Carlo program. Synthetic projections at fixed angular increments from a set of input projections at evenly distributed angles were generated by different u-shaped convolutional neural networks (u-nets). These u-nets differed in noise realization used for the training data, number of input projections, projection angle increment, and number of training/validation datasets. Synthetic projections were generated for 500 test projection datasets for each u-net, and a quantitative analysis was performed using statistical hypothesis tests based on structural similarity index measure and normalized root-mean-squared error. Additional simulations with varying detector orbits were performed on a subset of the dataset to study the effect of the detector orbit on the performance of the methodology. For verification of the results, the u-nets were applied to Jaszczak and NEMA physical phantom data obtained on a clinical SPECT/CT system.ResultsNo statistically significant differences were observed between u-nets trained with different noise realizations. In contrast, a statistically significant deterioration was found for training with a small subset (400 datasets) of the 10,000 simulated projection datasets in comparison with using a large subset (9500 datasets) for training. A good agreement between synthetic (i.e., u-net generated) and simulated projections before adding noise demonstrates a denoising effect. Finally, the physical phantom measurements show that our findings also apply for projections measured on a clinical SPECT/CT system.ConclusionOur study shows the large potential of u-nets for accelerating SPECT/CT imaging. In addition, our analysis numerically reveals a denoising effect when generating synthetic projections with a u-net. Clinically interesting, the methodology has proven robust against camera orbit deviations in a clinically realistic range. Lastly, we found that a small number of training samples (e.g., ~ 400 datasets) may not be sufficient for reliable generalization of the u-net.

Pressing and Rubbing: Physics-Informed Features Facilitate Haptic Terrain Classification for Legged Robots

Non-geometric hazards like sinkage and slipping, correlated to terrain categories, have an apparent effect on the locomotion of legged robots. Tactile-based terrain classification is a more accurate way to distinguish terrains in different properties than the vision, but selecting representative features instead of cumbersome ones in the complex foot-terrain interaction for efficient classification is still a challenge. In this letter, two specific leg motions are designed to inspect terrain bearing and friction properties, and manually designed features are extracted based on the foot-terrain interaction model for classification. These features are physics-informed, tidy and interpretable, and can be used with different classifiers under different foot configurations. Four classic classifiers with physics-informed features are trained for terrain classification and evaluated on our self-developed dataset. At the same time, the proposed method was compared with other two methods: an artificial feature extraction method and a CNN-based method. The results show that our proposed method reaches remarkable precision in terrain classification and can still guarantee a high accuracy under a small number of training samples.

An antagonistic training algorithm for TFT-LCD module mura defect detection

Although the production process of liquid crystal display model has been automated, the quality detection still depends on manual work. Mura defect is one of the common defects appearing in TFT-LCD modules. Since mura defect is not significantly different from the common background, it is difficult to detect. This paper presents a deep channel attention-based classification network (DCANet), which acts as a powerful feature extractor for object detectors, and proposes an antagonistic training algorithm based on convolution neural network. By the proposed training approach, the deep learning-based object detectors can achieve high accuracy even with a small number of training samples of mura defect. The experimental results show that compared to vanilla training method, the deep learning-based detectors trained by our proposed method could significantly improve their performance on mura defect detection with a few training samples. Even trained on only 600 samples, the mistake rate and miss rate are only 8.08% and 0.267% respectively, which can completely fulfill the enterprise’s requirements of 10% and 0.3%.

Intuitiveness in Active Teaching

While machine learning (ML) gives rise to astonishing results in automated systems, it is usually at the cost of large data requirements. This makes many successful algorithms from ML unsuitable for human-machine interaction, where the machine must learn from a small number of training samples that can be provided by a user within a reasonable time frame. Fortunately, the user can tailor the training data they create to be as useful as possible, severely limiting its necessary size—as long as they know about the machine’s requirements and limitations. Of course, acquiring this knowledge can in turn be cumbersome and costly. This raises the question of how easy ML algorithms are to interact with. In this work, we address this issue by analyzing the intuitiveness of certain algorithms when they are actively taught by users. After developing a theoretical framework of intuitiveness as a property of algorithms, we introduce an active teaching paradigm involving a prototypical two-dimensional spatial learning task as a method to judge the efficacy of human-machine interactions. Finally, we present and discuss the results of a large-scale user study into the performance and teaching strategies of 800 users interacting with two prominent ML algorithms in our system, providing first evidence for the role of intuition as an important factor impacting human-machine interaction.

DSNN: A DenseNet-Based SNN for Explainable Brain Disease Classification.

Aims: Brain diseases refer to intracranial tissue and organ inflammation, vascular diseases, tumors, degeneration, malformations, genetic diseases, immune diseases, nutritional and metabolic diseases, poisoning, trauma, parasitic diseases, etc. Taking Alzheimer’s disease (AD) as an example, the number of patients dramatically increases in developed countries. By 2025, the number of elderly patients with AD aged 65 and over will reach 7.1 million, an increase of nearly 29% over the 5.5 million patients of the same age in 2018. Unless medical breakthroughs are made, AD patients may increase from 5.5 million to 13.8 million by 2050, almost three times the original. Researchers have focused on developing complex machine learning (ML) algorithms, i.e., convolutional neural networks (CNNs), containing millions of parameters. However, CNN models need many training samples. A small number of training samples in CNN models may lead to overfitting problems. With the continuous research of CNN, other networks have been proposed, such as randomized neural networks (RNNs). Schmidt neural network (SNN), random vector functional link (RVFL), and extreme learning machine (ELM) are three types of RNNs.Methods: We propose three novel models to classify brain diseases to cope with these problems. The proposed models are DenseNet-based SNN (DSNN), DenseNet-based RVFL (DRVFL), and DenseNet-based ELM (DELM). The backbone of the three proposed models is the pre-trained “customize” DenseNet. The modified DenseNet is fine-tuned on the empirical dataset. Finally, the last five layers of the fine-tuned DenseNet are substituted by SNN, ELM, and RVFL, respectively.Results: Overall, the DSNN gets the best performance among the three proposed models in classification performance. We evaluate the proposed DSNN by five-fold cross-validation. The accuracy, sensitivity, specificity, precision, and F1-score of the proposed DSNN on the test set are 98.46% ± 2.05%, 100.00% ± 0.00%, 85.00% ± 20.00%, 98.36% ± 2.17%, and 99.16% ± 1.11%, respectively. The proposed DSNN is compared with restricted DenseNet, spiking neural network, and other state-of-the-art methods. Finally, our model obtains the best results among all models.Conclusions: DSNN is an effective model for classifying brain diseases.

Automatic Target Recognition of SAR Images Using Collaborative Representation.

Synthetic aperture radar (SAR) automatic target recognition (ATR) is one of the key technologies for SAR image interpretation. This paper proposes a SAR target recognition method based on collaborative representation-based classification (CRC). The collaborative coding adopts the global dictionary constructed by training samples of all categories to optimally reconstruct the test samples and determines the target category according to the reconstruction error of each category. Compared with the sparse representation methods, the collaborative representation strategy can improve the representation ability of a small number of training samples for test samples. For SAR target recognition, the resources of training samples are very limited. Therefore, the collaborative representation is more suitable. Based on the MSTAR dataset, the experiments are carried out under a variety of conditions and the proposed method is compared with other classifiers. Experimental results show that the proposed method can achieve superior recognition performance under the standard operating condition (SOC), configuration variances, depression angle variances, and a small number of training samples, which proves its effectiveness.

Superpixel Nonlocal Weighting Joint Sparse Representation for Hyperspectral Image Classification

Joint sparse representation classification (JSRC) is a representative spectral–spatial classifier for hyperspectral images (HSIs). However, the JSRC is inappropriate for highly heterogeneous areas due to the spatial information being extracted from a fixed-sized neighborhood block, which is often unable to conform to the naturally irregular structure of land cover. To address this problem, a superpixel-based JSRC with nonlocal weighting, i.e., superpixel-based nonlocal weighted JSRC (SNLW-JSRC), is proposed in this paper. In SNLW-JSRC, the superpixel representation of an HSI is first constructed based on an entropy rate segmentation method. This strategy forms homogeneous neighborhoods with naturally irregular structures and alleviates the inclusion of pixels from different classes in the process of spatial information extraction. Afterwards, the superpixel-based nonlocal weighting (SNLW) scheme is built to weigh the superpixel based on its structural and spectral information. In this way, the weight of one specific neighboring pixel is determined by the local structural similarity between the neighboring pixel and the central test pixel. Then, the obtained local weights are used to generate the weighted mean data for each superpixel. Finally, JSRC is used to produce the superpixel-level classification. This speeds up the sparse representation and makes the spatial content more centralized and compact. To verify the proposed SNLW-JSRC method, we conducted experiments on four benchmark hyperspectral datasets, namely Indian Pines, Pavia University, Salinas, and DFC2013. The experimental results suggest that the SNLW-JSRC can achieve better classification results than the other four SRC-based algorithms and the classical support vector machine algorithm. Moreover, the SNLW-JSRC can also outperform the other SRC-based algorithms, even with a small number of training samples.

A Novel Attention-Mechanism Based Cox Survival Model by Exploiting Pan-Cancer Empirical Genomic Information.

Cancer prognosis is an essential goal for early diagnosis, biomarker selection, and medical therapy. In the past decade, deep learning has successfully solved a variety of biomedical problems. However, due to the high dimensional limitation of human cancer transcriptome data and the small number of training samples, there is still no mature deep learning-based survival analysis model that can completely solve problems in the training process like overfitting and accurate prognosis. Given these problems, we introduced a novel framework called SAVAE-Cox for survival analysis of high-dimensional transcriptome data. This model adopts a novel attention mechanism and takes full advantage of the adversarial transfer learning strategy. We trained the model on 16 types of TCGA cancer RNA-seq data sets. Experiments show that our module outperformed state-of-the-art survival analysis models such as the Cox proportional hazard model (Cox-ph), Cox-lasso, Cox-ridge, Cox-nnet, and VAECox on the concordance index. In addition, we carry out some feature analysis experiments. Based on the experimental results, we concluded that our model is helpful for revealing cancer-related genes and biological functions.

Hyperspectral Image Classification Based on Spectral Multiscale Convolutional Neural Network

In recent years, convolutional neural networks (CNNs) have been widely used for hyperspectral image classification, which show good performance. Compared with using sufficient training samples for classification, the classification accuracy of hyperspectral images is easily affected by a small number of samples. Moreover, although CNNs can effectively classify hyperspectral images, due to the rich spatial and spectral information of hyperspectral images, the efficiency of feature extraction still needs to be further improved. In order to solve these problems, a spatial–spectral attention fusion network using four branch multiscale block (FBMB) to extract spectral features and 3D-Softpool to extract spatial features is proposed. The network consists of three main parts. These three parts are connected in turn to fully extract the features of hyperspectral images. In the first part, four different branches are used to fully extract spectral features. The convolution kernel size of each branch is different. Spectral attention block is adopted behind each branch. In the second part, the spectral features are reused through dense connection blocks, and then the spectral attention module is utilized to refine the extracted spectral features. In the third part, it mainly extracts spatial features. The DenseNet module and spatial attention block jointly extract spatial features. The spatial features are fused with the previously extracted spectral features. Experiments are carried out on four commonly used hyperspectral data sets. The experimental results show that the proposed method has better classification performance than some existing classification methods when using a small number of training samples.

Plant Species Classification Based on Hyperspectral Imaging via a Lightweight Convolutional Neural Network Model.

In recent years, many image-based approaches have been proposed to classify plant species. Most methods utilized red green blue (RGB) imaging materials and designed custom features to classify the plant images using machine learning algorithms. Those works primarily focused on analyzing single-leaf images instead of live-crown images. Without considering the additional features of the leaves’ color and spatial pattern, they failed to handle cases that contained leaves similar in appearance due to the limited spectral information of RGB imaging. To tackle this dilemma, this study proposes a novel framework that combines hyperspectral imaging (HSI) and deep learning techniques for plant image classification. We built a plant image dataset containing 1,500 images of 30 different plant species taken by a 470–900 nm hyperspectral camera and designed a lightweight conventional neural network (CNN) model (LtCNN) to perform image classification. Several state-of-art CNN classifiers are chosen for comparison. The impact of using different band combinations as the network input is also investigated. Results show that using simulated RGB images achieves a kappa coefficient of nearly 0.90 while using the combination of 3-band RGB and 3-band near-infrared images can improve to 0.95. It is also found that the proposed LtCNN can obtain a satisfactory performance of plant classification (kappa = 0.95) using critical spectral features of the green edge (591 nm), red-edge (682 nm), and near-infrared (762 nm) bands. This study also demonstrates the excellent adaptability of the LtCNN model in recognizing leaf features of plant live-crown images while using a relatively smaller number of training samples than complex CNN models such as AlexNet, GoogLeNet, and VGGNet.

A Novel Deep Reinforcement Approach for IIoT Microgrid Energy Management Systems

Introducing Deep Learning in the Industrial Internet of Things (IIoT) brings many benefits, such as network resilience and bandwidth usage reduction. In this work, we propose an innovative reinforcement learning architecture to implement distributed energy management systems for microgrids. The architecture is based on novel reinforcement learning and on time series prediction. The designed reinforcement learning uses classical recurrent neural networks instead of the habitual SAR (State Action Reward) method that most of the recent bibliography considers. We applied various techniques (Exact resolution, Rule-Based, Q-Learning, and our designed reinforcement learning) on a distributed IIoT energy control architecture. The proposed method has shown better results compared to the exact resolution and the Q-Learning algorithm. It results in fast learning systems with a small number of training samples. We identified and tested several management strategies. Integer Linear Programming (ILP) optimal expressions and strategy-based implementations are derived. We utilize the obtained results to train the recurrent neural network. Comparative results are very encouraging and prone to a generalization of our approach instead of the classical methods.

Predicting microbiome compositions from species assemblages through deep learning.

Microbes can form complex communities that perform critical functions in maintaining the integrity of their environment or their hosts' well-being. Rationally managing these microbial communities requires improving our ability to predict how different species assemblages affect the final species composition of the community. However, making such a prediction remains challenging because of our limited knowledge of the diverse physical, biochemical, and ecological processes governing microbial dynamics. To overcome this challenge, we present a deep learning framework that automatically learns the map between species assemblages and community compositions from training data only, without knowing any of the above processes. First, we systematically validate our framework using synthetic data generated by classical population dynamics models. Then, we apply our framework to data from in vitro and in vivo microbial communities, including ocean and soil microbiota, Drosophila melanogaster gut microbiota, and human gut and oral microbiota. We find that our framework learns to perform accurate out-of-sample predictions of complex community compositions from a small number of training samples. Our results demonstrate how deep learning can enable us to understand better and potentially manage complex microbial communities.

Single-Sample Face Recognition Based on Shared Generative Adversarial Network

Single-sample face recognition is a very challenging problem, where each person has only one labeled training sample. It is difficult to describe unknown facial variations. In this paper, we propose a shared generative adversarial network (SharedGAN) to expand the gallery dataset. Benefiting from the shared decoding network, SharedGAN requires only a small number of training samples. After obtaining the generated samples, we join them into a large public dataset. Then, a deep convolutional neural network is trained on the new dataset. We use the well-trained model for feature extraction. With the deep convolutional features, a simple softmax classifier is trained. Our method has been evaluated on AR, CMU-PIE, and FERET datasets. Experimental results demonstrate the effectiveness of SharedGAN and show its robustness for single sample face recognition.

Effects of Training Parameter Concept and Sample Size in Possibilistic c-Means Classifier for Pigeon Pea Specific Crop Mapping

This research work aims to study the effect of training parameter concept and sample size in the process of classification by using a fuzzy Possibilistic c-Means (PCM) approach for Pigeon Pea specific crop mapping. For specific class extraction, the “mean” of the training data is considered as a training parameter of the classification algorithm. In this study, we proposed an “Individual Sample as Mean” (ISM) approach where the individual training sample is accounted as a mean parameter for the fuzzy PCM classifier. In order to avoid the spectral overlap of target Pigeon pea crop with other crops in the study area, a temporal indices database was generated from Sentinel 2A/2B satellite images acquired during the 2019–2020 Pigeon Pea crop cycle. The spectral dimensionality of temporal data was reduced to extract the required bands to achieve maximum enhancement of the target crop class in the temporal data. Further, the training sample size was increased to study the heterogeneity within the class in the classified output. The proposed ISM approach delivered a higher mean membership difference (MMD) between the Pigeon Pea crop and the co-cultivated Cotton crop as compared to the conventional mean method. This indicated that a better separation was achieved between the target crop and the spectrally similar crop grown, that were cultivated in the same study area. When the sample size was gradually increased from 5 to 60, the MMD values within the Pigeon Pea test fields remained in the range 0.013–0.02, thereby implying that the proposed algorithm works better even with a small number of training samples. The heterogeneity was better handled using the proposed ISM approach since the variance obtained within Pigeon Pea field was only 0.008, as compared to that of 0.02 achieved using the conventional mean approach.

Rock mass structural surface trace extraction based on transfer learning

Abstract To solve engineering geological problems, including water conservancy, transportation, and mining, it is necessary to obtain information on rock mass structures, such as slopes, foundation pits, and tunnels, in time. The traditional method for obtaining structural information requires manual measurement, which is time consuming and labor intensive. Because geological information is complicated and diverse, it is not practical for general deep learning methods to obtain full-scale structural surface trace images to prepare training samples. Transfer learning can abstract high-level features from low-level features with a small number of training samples, which can automatically express the inherent characteristics of objects. This article proposed a rock mass structural surface trace extraction method based on the transfer learning technique that considers the attention mechanism and shape constraints. For the general test set, the accuracy of rock mass structural surface trace recognition with the proposed method can reach 87.2%. Experimental results showed that the proposed method has advantages in extracting complicated geological structure information and is valuable for providing technical support for the extraction of geological information in the construction of water conservancy, transportation, mining, and related projects.

Combined Spatial-Spectral Schroedinger Eigenmaps with Multiple Kernel Learning for Hyperspectral Image Classification Using a Low Number of Training Samples

The classification of hyperspectral images is one of the most popular fields in remote sensing applications. It should be noted that spectral and spatial features have critical roles in this research area. This paper proposes a method based on spatial-spectral Schroedinger eigenmaps (SSSE) and multiple kernel learning (MKL) to classify hyperspectral images more efficiently while using a low number of training samples. In the proposed method, first SSSE is applied to spectral domain in order to extract significant features and reduce dimension of the original image. Then MKL is utilized to enhance the feature learning process and obtain an optimum combination of some specified kernels. Finally, the classification is carried out by substituting the optimal kernel in support vector machine (SVM) algorithm. Experimental results show that the proposed method improves classification accuracy significantly and provides highly efficient results in the case of a small number of training samples. Furthermore, the computation time of the proposed method is much lower than the state-of-the-art MKL methods.