Most machine learning classifiers only concern classification accuracy, while certain applications (such as medical diagnosis, meteorological forecasting, and computation advertising) require the model to predict the true probability, known as a calibrated estimate. In previous work, researchers have developed several calibration methods to post-process the outputs of a predictor to obtain calibrated values, such as binning and scaling methods. Compared with scaling, binning methods are shown to have distribution-free theoretical guarantees, which motivates us to prefer binning methods for calibration. However, we notice that existing binning methods have several drawbacks: (a) the binning scheme only considers the original prediction values, thus limiting the calibration performance; and (b) the binning approach is non-individual, mapping multiple samples in a bin to the same value, and thus is not suitable for order-sensitive applications. In this paper, we propose a feature-aware binning framework, called Multiple Boosting Calibration Trees (MBCT), along with a multi-view calibration loss to tackle the above issues. Our MBCT optimizes the binning scheme by the tree structures of features, and adopts a linear function in a tree node to achieve individual calibration. Our MBCT is non-monotonic, and has the potential to improve order accuracy, due to its learnable binning scheme and the individual calibration. We conduct comprehensive experiments on three datasets in different fields. Results show that our method outperforms all competing models in terms of both calibration error and order accuracy. We also conduct simulation experiments, justifying that the proposed multi-view calibration loss is a better metric in modeling calibration error. In addition, our approach is deployed in a real-world online advertising platform; an A/B test over two weeks further demonstrates the effectiveness and great business value of our approach.

BACKGROUND: The study of microRNAs (miRNAs) is attracting great considerations. Recent studies revealed that miRNAs play as important regulators of gene expression and some even as cancer players or inhibitors. Many studies try to discover new miRNAs and reveal the miRNA expression profile in cancer using a SAGE-based total RNA clone method. However, the data processing of this method is labor-intensive with several different biological databases and more than ten data processing steps involved. RESULTS: With miRAS, miRNAs and possible miRNA candidates contained in the submitted sequencing data were obtained together with their expression profile. The functions of known and predicted miRNAs were then analyzed by miRNA target prediction followed by target gene annotations. Finally, to extract the biological significance of the miRNAs in the samples, further annotations of the known miRNA and target genes were performed by collecting the public expression datasets of miRNA and target genes in normal and cancer tissues. CONCLUSION: We introduce a web-based analysis platform called miRNA Analysis System (miRAS), for processing and analyzing of the sequence data obtained from the total RNA clone method. The system was built on generalizing the study of a liver cancer cell line total RNA sequencing project. miRAS is freely available on the web.