The use of expert systems and artificial intelligence techniques in disease diagnosis has been increasing gradually. Artificial Immune Recognition System (AIRS) is one of the methods used in medical classification problems. AIRS2 is a more efficient version of the AIRS algorithm. In this paper, we used a modified AIRS2 called MAIRS2 where we replace the K- nearest neighbors algorithm with the fuzzy K-nearest neighbors to improve the diagnostic accuracy of diabetes diseases. The diabetes disease dataset used in our work is retrieved from UCI machine learning repository. The performances of the AIRS2 and MAIRS2 are evaluated regarding classification accuracy, sensitivity and specificity values. The highest classification accuracy obtained when applying the AIRS2 and MAIRS2 using 10-fold cross-validation was, respectively 82.69% and 89.10%.

-This work is builds on the study of the 10 top data mining algorithms identified by the IEEE International Conference on Data Mining (ICDM) community in December 2006. We address the same study, but with the application of statistical tests to establish, a more appropriate and justified ranking classifier for classification tasks. Current studies and practices on theoretical and empirical comparison of several methods, approaches, advocated tests that are more appropriate. Thereby, recent studies recommend a set of simple and robust non-parametric tests for statistical comparisons classifiers. In this paper, we propose to perform non-parametric statistical tests by the Friedman test with post-hoc tests corresponding to the comparison of several classifiers on multiple data sets. The tests provide a better judge for the relevance of these algorithms.

International audienc

Automated classification of medical images is an increasingly important tool for physicians in their daily activities. However, due to its computational complexity, this task is one of the major current challenges in the field of content-based image retrieval (CBIR). In this paper, a medical image classification approach is proposed. This method is composed of two main phases. The first step consists of a pre-processing, where a texture and shape based features vector is extracted. Also, a feature selection approach was applied by using a Genetic Algorithm (GA). The proposed GA uses a kNN based classification error as fitness function, which enables the GA to obtain a combinatorial set of feature giving rise to optimal accuracy. In the second phase, a classification process is achieved by using random Forest classifier and a supervised multi-class classifier based on the support vector machine (SVM) for classifying X-ray images.

Random Forest RF is a successful technique of ensemble prediction that uses the majority voting or an average depending on the combination. However, it is clear that each tree in a random forest can have different contribution to the treatment of some instance. In this paper, we show that the prediction performance of RF's can still be improved by replacing the GINI index with another index (twoing or deviance). Our experiments also indicate that weighted voting gives better results compared to the majority vote.

Opinion mining from medical forums such as health check-ups is sparking growing interest and a stimulating area for natural language processing. This allows for a better understanding of patient health status and drug reactions while generating new knowledge for health care professionals and drug manufacturers, which helps improve the quality of service and produce more effective treatments. In this paper, the researchers present a framework of opinions classification of drug reviews. The objective of this work is to find the best model for analyzing patients’ emotions about drugs. In this sense, the researchers oppose classical text vectorization methods (bag of words, term frequency-inverse document frequency) and word embedding methods (Word2vec, GloVe) for classical opinion mining face to modern machine learning tools with the Convolutional Neural Network (CNN), the Recurrent Neural Networks (Long Short-term Memory and Bidirectional Long Short-Term Memory). Experiments results show that the best model for drug reviews was achieved by CNN based on the Skip-gram model (85% accuracy). Experiments have led to conclude that the performance of a given model will depend on the type of dataset used, on feature representation and better collaboration between classifiers and feature extraction methods.

No abstract

-In this paper, we consider the problem of stabilizing network using a new proportional-integral (PI) based congestion controller in active queue management (AQM) router; with appropriate model approximation in the first order delay systems, we seek a stability region of the controller by using the HermiteBiehler theorem, which isapplicable to quasipolynomials. A Genetic Algorithm technique is employed to derive optimal or near optimal PI controller parameters.