No abstract

No abstract

Knee osteoarthritis (OA) detection is an important area of research in health informatics that aims to improve the accuracy of diagnosing this debilitating condition. In this paper, we investigate the ability of DenseNet169, a deep convolutional neural network architecture, for knee osteoarthritis detection using X-ray images. We focus on the use of the DenseNet169 architecture and propose an adaptive early stopping technique that utilizes gradual cross-entropy loss estimation. The proposed approach allows for the efficient selection of the optimal number of training epochs, thus preventing overfitting. To achieve the goal of this study, the adaptive early stopping mechanism that observes the validation accuracy as a threshold was designed. Then, the gradual cross-entropy (GCE) loss estimation technique was developed and integrated to the epoch training mechanism. Both adaptive early stopping and GCE were incorporated into the DenseNet169 for the OA detection model. The performance of the model was measured using several metrics including accuracy, precision, and recall. The obtained results were compared with those obtained from the existing works. The comparison shows that the proposed model outperformed the existing solutions in terms of accuracy, precision, recall, and loss performance, which indicates that the adaptive early stopping coupled with GCE improved the ability of DenseNet169 to accurately detect knee OA.

Interactive data exploration platforms in Web, business and scientific domains are becoming increasingly popular. Typically, users without prior knowledge of data interact with these platforms in an exploratory manner hoping they might retrieve the results they are looking for. One way to explore large-volume data is by posing aggregate queries which group values of multiple rows by an aggregate operator to form a single value: an aggregated value. Though, when a query fails, i.e., returns undesired aggregated value, users will have to undertake a frustrating trial-and-error process to refine their queries, until a desired result is attained. This data exploration process, however, is growing rather difficult as the underlying data is typically of large-volume and high-dimensionality. While heuristic-based techniques are fairly successful in generating refined queries that meet specified requirements on the aggregated values, they are rather oblivious to the (dis)similarity between the input query and its corresponding refined version. Meanwhile, enforcing a similarity-aware query refinement is rather a non-trivial challenge, as it requires a careful examination of the query space while maintaining a low processing cost. To address this challenge, we propose an innovative scheme for efficient Similarity-Aware Refinement of Aggregation Queries called (EAGER) which aims to balance the tradeoff between satisfying the aggregate and similarity constraints imposed on the refined query to maximize its overall benefit to the user. To achieve that goal, EAGER implements efficient strategies to minimize the costs incurred in exploring the available search space by utilizing similaritybased and monotonic-based pruning techniques to bound the search space and quickly find a refined query that meets users' expectations. Our extensive experiments show the scalability exhibited by EAGER under various workload settings, and the significant benefits it provides.