Search

Brain image classification is one of the most useful and widely needed processes in the medical system, and it is a highly challenging field. This paper presents a new method for selecting a significant subset of features as the input to the classifier, called mutual information-accelerated singular value decomposition (MI-ASVD). This novel algorithm is exploited to design an intelligent system for classifying MRI brain images into three classes: healthy, high-grade glioma, and low-grade glioma. The proposed system has six stages: pre-processing, clustering, tumour localization, feature extraction, MI-ASVD and classification. First, the MR images are smoothed by using enhancement techniques such as Gaussian kernel filters. Then, local difference in intensity-means (LDI-Means) clustering is employed to segment and detect suspicious regions. The grey-level run-length matrix (GLRLM), texture, and colour intensity features are used for tumour feature extraction. Later, a special method including a summation of feature selection and dimensionality reduction, MI-ASVD, is applied to select the most useful features for the classification process. Finally, the simplified residual neural network technique is implemented to classify the MR brain images. Using MI-ASVD provided accurate and more efficacious results in classification compared with the original feature space and with two other standard dimensionality reduction methods, principal component analysis (PCA) and singular value decomposition (SVD). It achieved a classification accuracy of 94.91%, which is better than the two state-of-the-art techniques as well as methods from similar published studies.

Deep learning models have been very successful in computer vision and image processing applications. Since its inception, Convolutional Neural Network (CNN)-based deep learning models have consistently outperformed other machine learning methods on many significant image processing benchmarks. Many top-performing methods for image segmentation are also based on deep CNN models. However, deep CNN models fail to integrate global and local context alongside visual features despite having complex multi-layer architectures. We propose a novel three-layered deep learning model that learns independently global and local contextual information alongside visual features, and visual feature selection based on a genetic algorithm. The novelty of the proposed model is that One-vs-All binary class-based learners are introduced to learn Genetic Algorithm (GA) optimized features in the visual layer, followed by the contextual layer that learns global and local contexts of an image, and finally the third layer integrates all the information optimally to obtain the final class label. Stanford Background and CamVid benchmark image parsing datasets were used for our model evaluation, and our model shows promising results. The empirical analysis reveals that optimized visual features with global and local contextual information play a significant role to improve accuracy and produce stable predictions comparable to state-of-the-art deep CNN models.

In this project, a sinonasal diseases dataset is created with the help of Ear, Nose, and Throat (ENT) specialists at King Abdullah University Hospital (KAUH), Jordan. This dataset is then used to experiment with different features extraction and selection methods and different machine learning classification methods. The work can be summarized as follows. We start by selecting the participating patients according to their history information that is acquired from the patients using a questionnaire. Then, we construct a tool that employs image processing techniques to process the selected patients’ CT scan images to extract useful information from them. After that, different filters and wrapper selection methods with various machine learning techniques are utilized to classify the cases in the dataset. The results show that the performance of wrapper feature selection (using PART classifier with best first search) with all used classifiers produces better results compared with the other feature selection methods.

Deep learning has brought a series of breakthroughs in image processing. Specifically, there are significant improvements in the application of food image classification using deep learning techniques. However, very little work has been studied for the classification of food ingredients. Therefore, this paper proposes a new framework, called DeepFood which not only extracts rich and effective features from a dataset of food ingredient images using deep learning but also improves the average accuracy of multi-class classification by applying advanced machine learning techniques. First, a set of transfer learning algorithms based on Convolutional Neural Networks (CNNs) are leveraged for deep feature extraction. Then, a multi-class classification algorithm is exploited based on the performance of the classifiers on each deep feature set. The DeepFood framework is evaluated on a multi-class dataset that includes 41 classes of food ingredients and 100 images for each class. Experimental results illustrate the effectiveness of the DeepFood framework for multi-class classification of food ingredients. This model that integrates ResNet deep feature sets, Information Gain (IG) feature selection, and the SMO classifier has shown its supremacy for foodingredients recognition compared to several existing work in this area.

We present a benchmarking framework to design multi-spectral systems working in the NIR range for multiple purposes. This framework is composed of a hyperspectral imaging hardware and an ad-hoc software that performs pattern recognition experiments (image acquisition, segmentation, feature extraction, feature selection, classification and evaluation steps) comparing different algorithms in every step. For each experiment, we obtain a solution using a generic hyperspectral system, but we also obtain enough data to design a specific multi-spectral system in order to decrease the overall execution time. This improvement is based in the feature selection step, that provides the most relevant wavelengths for the problem. The framework has been tested for detecting internal and external features in potatoes, determining the origin of honey, and studying fecundity parameters in hen eggs.

AdaBoost is an efficient method for producing a highly accurate learning algorithm by assembling multiple classifiers, but it is also widely known for its long duration of off-line learning. Especially, when it is applied for feature selection for object detection, its learning process is to exhaustively evaluate every feature in a large set. With the increasing of image resolution and complexity of feature transformation approaches, the computational time will be extremely long, which makes the large scale AdaBoost learning very difficult. In this paper, we have employed Grid Computing to solve the difficulty. The proposed algorithm is to select the most significant features for face recognition. The selection algorithm is derived from multi-class AdaBoost, which exhaustively evaluate every feature from a large set. The deployed Grid Computing system is actually used for High Throughput Computing specialised on advanced resource management. To utilizing Grid Computing on the feature selection process, we have improved multi-class AdaBoost learning algorithm with parallel structure, so that the task of High Performance Computing is accomplished in the environment of High Throughput Computing. With Grid Computing, selecting 200 features from a large set of 30240 features is finished in 20 days, while without Grid Computing the time would be more than two years. It shows that Grid Computing brings vast advantage to computer vision, machine learning, image processing, and pattern recognition.

Cellular image analysis is being attempted by several Image Processing (IP), statistical approaches and machine learning (ML), Neural Networks (NN) techniques. IP based algorithms are getting trapped into subjectivity due to reasons of cell image properties and methods of extractions and classification techniques. ML based methods are also well explored, but it suffer from large training sets and feature selection to reduce dimensionality of neural network. This paper proposes design and integration of Evolutionary methodologies with temporal networks and on-the-fly learning for getting desired confidence interval with minimal training. IP based method is experimented with morphological feature extraction. Results revels that, utility limited to spatial and noise-free data. This emphasizes the need of evolutionary methodologies with temporal learning and for optimizing weights, structure and learning of NN. Paper presents results extracted from IP algorithm with its shortcoming and proposes framework based on Evolutionary ANN to overcome the subjectivity issue reported in text and experimentation.

Most techniques for image analysis consider the image labels fixed and without uncertainty. In this paper, we address the problem of ordinal/rank label prediction based on training samples with uncertain labels. First, the core ranking model is designed as the bilinear fusing of multiple candidate kernels. Then, the parameters for feature selection and kernel selection are learned by maximum a posteriori for given samples and uncertain labels. The convergency provable Expectation-Maximization (EM) method is used for inferring these parameters. The effectiveness of the proposed algorithm is finally validated by the extensive experiments on age ranking task. The FG-NET and Yamaha aging database are used for the experiments, and our algorithm significantly outperforms those state-of-the-art algorithms ever reported in literature.

In pattern recognition problems features plays an important role for classification results. It is very important which features are used and how many features are used for the classification process. Most of the real life classification problem uses different category of features. It is desirable to find the optimal combination of features that improves the performance of the classifier. There exists different selection framework that selects the features. Mostly do not incorporate the impact of one category of features on another. Even if they incorporate, they produce conflict between the categories. In this paper we proposed a restricted crossover selection framework which incorporate the impact of different categories on each other, as well as it restricts the search within the category which searching in the global region of the search space. The results obtained by the proposed framework are promising.

The problem of feature selection is fundamental in a number of different tasks like classification, data mining, image processing, conceptual learning, etc. In recent times, the growing importance of knowledge discovery and data-mining approaches in practical applications has made the feature selection problem a quite hot topic. Choosing a subset of the features may increase accuracy and reduce complexity of the acquired knowledge. This paper, proposes a new feature subset selection approach by using slicing techniques, which was originally proposed for programming languages. Slicing means that we are interested in automatically obtaining that portion 'features' of the case responsible for specific parts of the solution of the case at hand. The paper presented the proposed approach based on slicing techniques improved the feature subset selection for classification task in data mining.