In the real world, a realistic setting for computer vision or multimedia recognition problems is that we have some classes containing lots of training data and many classes contain a small amount of training data. Therefore, how to use frequent classes to help learning rare classes for which it is harder to collect the training data is an open question. Learning with Shared Information is an emerging topic in machine learning, computer vision and multimedia analysis. There are different level of components that can be shared during concept modeling and machine learning stages, such as sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, etc. Regarding the specific methods, multi-task learning, transfer learning and deep learning can be seen as using different strategies to share information. These learning with shared information methods are very effective in solving real-world large-scale problems. This special issue aims at gathering the recent advances in learning with shared information methods and their applications in computer vision and multimedia analysis. Both state-of-the-art works, as well as literature reviews, are welcome for submission. Papers addressing interesting real-world computer vision and multimedia applications are especially encouraged. Topics of interest include, but are not limited to:  • Multi-task learning or transfer learning for large-scale computer vision and multimedia analysis • Deep learning for large-scale computer vision and multimedia analysis • Multi-modal approach for large-scale computer vision and multimedia analysis • Different sharing strategies, e.g., sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, • Real-world computer vision and multimedia applications based on learning with shared information, e.g., event detection, object recognition, object detection, action recognition, human head pose estimation, object tracking, location-based services, semantic indexing. • New datasets and metrics to evaluate the benefit of the proposed sharing ability for the specific computer vision or multimedia problem. • Survey papers regarding the topic of learning with shared information.  Authors who are unsure whether their planned submission is in scope may contact the guest editors prior to the submission deadline with an abstract, in order to receive feedback.

IEEE transactions on pattern analysis and machine intelligence

Advancement and emergence of newer technologies such as analytics, artificial intelligence, machine learning have impacted many sectors such as health care, automotive etc. In the healthcare sector, these technologies resulted in various benefits such as clinical decision support, better care coordination, improving patient wellness etc. The World over, Coronary Heart Disease (CHD) is affecting millions of people. Various machine learning techniques include ensemble classifiers can be used in healthcare for improving prediction accuracy. This paper analyses various ensemble methods (Bagged Tree, Random Forest, and AdaBoost) along with Feature subset selection method — Particle Swarm Optimization (PSO), to accurately predict the occurrence of heart disease for a particular patient. Experimental results show that Bagged Tree and PSO achieved the highest accuracy.

Prediction of heart disease using ensemble learning and Particle Swarm Optimization

Earlier risk assessment and identification of different diseases is the most crucial issue for avoiding and lowering their progression. The researchers typically used the statistical comparative analysis or step-by-step methods of feature selection using regression techniques to estimate the risk factors of diseases. The outcomes from these methods emphasized on individual risk factors separately. A combination of factors, however, is likely to affect the development of disease rather than just anyone alone. Genetic algorithms (GA) can be useful and efficient for searching a combination of factors for the fast diagnosis with best accuracies, especially for a large number of complex and poorly understood factors, as in the case in the prediction of disease development. Our proposed model shows the potential for the application of GA in diagnosing disease and predicting accuracy. Our proposed model demonstrated that the amalgamation of a small subset of input features produces the optimum performance than the use of all the single significant features individually. This model not only predicts the best feature sets and accuracy but also overcome the problem of missing values present in the dataset. Variables more frequently selected by LR might be more important for the prediction of disease development and accuracies by GA.

Variable Selection in Genetic Algorithm Model with Logistic Regression for Prediction of Progression to Diseases

Fuzzy clustering is one of the most widely used and sensitive algorithms However, one of its fatal weaknesses is that it is very sensitive to initialization and easy to get into local minima WWO (The water wave optimization algorithm) is a widely used global optimization method Its main advantage is simple, general and suitable for parallel processing Thus combining optimization algorithm in waves and fuzzy clustering, can play a water wave algorithm for global optimization ability It can give attention to both local optimization ability and improve the convergence speed at the same time, to better solve the problem of clustering The algorithm using water wave optimization algorithm to find the optimal solution as the initial clustering center of fuzzy clustering And then, using the fuzzy clustering to initialize clustering center Finally obtain the global optimal solution, thus overcomes the shortcomings of fuzzy clustering

Fuzzy Clustering Based on Water Wave Optimization

This paper presents a rough kernelized fuzzy c-means clustering (RKFCM) based medical image segmentation algorithm. It is a combination of rough set and kernelized FCM clustering (KFCM). KFCM introduced new technique of clustering using kernel induced distance and improved its robustness towards noise. However, it is failed to remove the vagueness and uncertainty of the clustering technique. In this paper, we use rough set with KFCM for removal of uncertainty by introduction of higher and lower estimation of rough set theory. The objective function derived from KFCM is merged with rough set to get better segmentation results. Experiments performed on numerous medical image data sets and its resulting validity index values have proved this algorithm to be more efficient in comparison to existing algorithms.

Rough Kernelized Fuzzy C-Means Based Medical Image Segmentation

Imprecision based data clustering algorithms have gained a lot of importance these days because of the imprecise character of modern day databases. Some such algorithms are the rough c-means (RCM), fuzzy c-means (FCM) and their hybrid versions. Li et al used the decision theoretic rough set (DTRS) model by the way improving the RCM. In their approach they have used a notion called loss function to limit the information lost due to neighbours. The method of allocation using decision-theoretic rough sets model deals with potentially high computational cost. It has been observed that hybrid models are better than individual models. Keeping this in view, here we develop a clustering algorithm using DTRS and fuzzy sets in view called the decision-theoretic rough fuzzy c-means (DTRFCM). Experiments carried out show that our approach is more efficient than the DTRS algorithm. For this purpose we used several well-known data sets and parameters like the DB-index, D-index and Accuracy measure.

A decision theoretic rough fuzzy c-means algorithm

The RCM algorithm may lead to undesirable solutions in practice because the points close to data points being assigned are neglected which led to the development of the decision theoretic rough set (DTRS) model and the decision theoretic rough C-means algorithm (DTRCM). It was recently improved further with the introduction of decision theoretic rough fuzzy C-means (DTRFCM). Here, we present a further improved algorithm called the decision theoretic intuitionistic fuzzy rough C-means (DTRIFCM) and provide a comparative performance analysis of DTRCM, DTRFCM and DTRIFCM through experiments and efficiency measuring indices DB, D and Acc. According to DB and D indexes DTRIFCM is better than the other two, where as far as the accuracy is concerned DTRFCM is better. We have chosen the data sets Iris, Wine, WDBC and Glass from UCI repository as input for the experimental purpose.

Sresht Agrawal

Papers

A decision theoretic rough fuzzy c-means algorithm

Decision Theoretic Rough Intuitionistic Fuzzy C-Means Algorithm