Label Noise Cleaning with an Adaptive Ensemble Method Based on Noise Detection Metric

Real-world datasets are often contaminated with label noise; labeling is not a clear-cut process and reliable methods tend to be expensive or time-consuming. Depending on the learning technique used, such label noise is potentially harmful, requiring an increased size of the training set, making the trained model more complex and more prone to overfitting and yielding less accurate prediction. This work proposes a cleaning technique called the ensemble method based on the noise detection metric (ENDM). From the corrupted training set, an ensemble classifier is first learned and used to derive four metrics assessing the likelihood for a sample to be mislabeled. For each metric, three thresholds are set to maximize the classifying performance on a corrupted validation dataset when using three different ensemble classifiers, namely Bagging, AdaBoost and k-nearest neighbor (k-NN). These thresholds are used to identify and then either remove or correct the corrupted samples. The effectiveness of the ENDM is demonstrated in performing the classification of 15 public datasets. A comparative analysis is conducted concerning the homogeneous-ensembles-based majority vote method and consensus vote method, two popular ensemble-based label noise filters.

Twitter Comment Predictions on Dues Changes BPJS Health In 2020

The Social Security Administering Body (BPJS) is a facility established by The government in providing services to citizens in The field of health welfare. The Spirit of cooperation in the utilization of health services which is very much currently a constraint in the budget is still insufficient in covering health services as a whole. For this reason, government policy is following with PERPRES No. 75 in 2019, the Government officially raised the BPJS Health contributions for 2020. The increase in BPJS Health contributions certainly caused a lot of comments. Namely Twitter, one of the social media that is used by the public to express disapproval or support for this government policy. This study, testing was carried out related to the prediction of comments from social media on community responses to the increase in BPJS Health contributions taken by the government. In the test carried out 3 (three) input algorithms. For every single algorithm including getting results through the K-NN method with an accuracy of 71.83% and AUC value of 0812, for the Naïve Bayes method produces an accuracy of 81.63% and AUC value of 0586. As for the C 4.5 method, the accuracy is 65.37% and the AUC value is 0628. While testing conducted through the Ensembles Vote method which combines the 3 algorithms above gives the best results with an accuracy of 80.10% and AUC value is 0871 for Twitter comment predictions.

The Impact of Cross-Species Gene Flow on Species Tree Estimation

Recent analyses of genomic sequence data suggest cross-species gene flow is common in both plants and animals, posing challenges to species tree estimation. We examine the levels of gene flow needed to mislead species tree estimation with three species and either episodic introgressive hybridization or continuous migration between an outgroup and one ingroup species. Several species tree estimation methods are examined, including the majority-vote method based on the most common gene tree topology (with either the true or reconstructed gene trees used), the UPGMA method based on the average sequence distances (or average coalescent times) between species, and the full-likelihood method based on multilocus sequence data. Our results suggest that the majority-vote method based on gene tree topologies is more robust to gene flow than the UPGMA method based on coalescent times and both are more robust than likelihood assuming a multispecies coalescent (MSC) model with no cross-species gene flow. Comparison of the continuous migration model with the episodic introgression model suggests that a small amount of gene flow per generation can cause drastic changes to the genetic history of the species and mislead species tree methods, especially if the species diverged through radiative speciation events. Estimates of parameters under the MSC with gene flow suggest that African mosquito species in the Anopheles gambiae species complex constitute such an example of extreme impact of gene flow on species phylogeny. [IM; introgression; migration; MSci; multispecies coalescent; species tree.]

The Impact of Cross-Species Gene Flow on Species Tree Estimation

ABSTRACTRecent analyses of genomic sequence data suggest cross-species gene flow is common in both plants and animals, posing challenges to species tree inference. We examine the levels of gene flow needed to mislead species tree estimation with three species and either episodic introgressive hybridization or continuous migration between an outgroup and one ingroup species. Several species tree estimation methods are examined, including the majority-vote method based on the most common gene tree topology (with either the true or reconstructed gene trees used), the UPGMA method based on the average sequence distances (or average coalescent times) between species, and the full-likelihood method based on multi-locus sequence data. Our results suggest that the majority-vote method is more robust to gene flow than the UPGMA method and both are more robust than likelihood assuming a multispecies coalescent (MSC) model with no cross-species gene flow. A small amount of introgression or migration can mislead species tree methods if the species diverged through speciation events separated by short time intervals. Estimates of parameters under the MSC with gene flow suggest the Anopheles gambia African mosquito species complex is an example where gene flow greatly impacts species phylogeny.