No abstract

No abstract

Neural Word Sense Disambiguation (WSD) has recently been shown to benefit from the incorporation of pre-existing knowledge, such as that coming from the WordNet graph. However, state-of-the-art approaches have been successful in exploiting only the local structure of the graph, with only close neighbors of a given synset influencing the prediction. In this work, we improve a classification model by recomputing logits as a function of both the vanilla independently produced logits and the global WordNet graph. We achieve this by incorporating an online neural approximated PageRank, which enables us to refine edge weights as well. This method exploits the global graph structure while keeping space requirements linear in the number of edges. We obtain strong improvements, matching the current state of the art. Code is available at https://github.com/SapienzaNLP/ neural-pagerank-wsd.

Heart diseases are considered one of the leading death rates for humanity in the recent decades. The early diagnosis and prediction of heart disease becomes a critical subject in medical domain. Data mining techniques are usually used for finding anomalies, patterns and correlations within large data sets, thus it's crucial for clinical data analysis and various disease prediction. Ensemble approaches have proven to be quite effective in solving a variety of classification problems. In this research, we propose a hybrid ensemble stacking model with different feature engineering algorithms. The proposed ensemble model is based on five base models: Random Forest, Decision Tree, K-Nearest Neighbour (K-NN), Support Vector Machine (SVM), and Naïve Bayes for heart disease diagnosis. Logistic Regression meta model is used to merge base models predictions. We have examined various feature selection approaches such as: Brute Force, Principal Component Analysis (PCA), Classification and Regression Tree (CART) Feature Importance, and Logistic Regression based Recursive Feature Elimination. The proposed approach has been experimentally validated and evaluated on different dataset : UCI Cleveland and UCI Statlog. A quantitative evaluation shows that the combination of the ensemble model with brute force as feature selection technique yields a top accuracy of 97.8% for heart disease classification. the proposed stacking model has proven it's efficiency and overcomes existing approaches in heart diseases classification

No abstract