Search

Distributed machine learning (ML) is a modern computation paradigm that divides its workload into independent tasks that can be simultaneously achieved by multiple machines (i.e., agents) for better scalability. However, a typical distributed system is usually implemented with a central server that collects data statistics from multiple independent machines operating on different subsets of data to build a global analytic model. This centralized communication architecture however exposes a single choke point for operational failure and places severe bottlenecks on the server's communication and computation capacities as it has to process a growing volume of communication from a crowd of learning agents. To mitigate these bottlenecks, this paper introduces a novel Collective Online Learning Gaussian Process framework for massive distributed systems that allows each agent to build its local model, which can be exchanged and combined efficiently with others via peer-to-peer communication to converge on a global model of higher quality. Finally, our empirical results consistently demonstrate the efficiency of our framework on both synthetic and real-world datasets., Extended version with proofs

Introduction We aimed to clarify the effect of vein of Marshall (VOM) ethanol infusion for treating VOM triggers and/or mitral flutter after first-attempt endocardial ablation in patients with nonparoxysmal atrial fibrillation (AF). Methods and results Of the 254 consecutive patients (age, 56 ± 10 years; 221 male) undergoing catheter ablation for drug-refractory nonparoxysmal AF, 32 (12.6%) received VOM ethanol infusion. The patients were stratified into group 1 (pulmonary vein isolation [PVI], substrate modification, VOM ethanol infusion), group 2 (PVI, substrate modification), and group 3 (PVI alone). Propensity-matched analysis (N = 128) of long-term outcomes (3.9 ± 0.5 years) revealed a higher AF recurrence risk in group 2 (hazard ratio [HR], 4.17; 95% confidence interval [95% CI], 1.63-10.69; P = .003) and group 3 (HR, 1.82; 95% CI, 1.09-3.04; P = .021) than in group 1, as well as a higher atrial arrhythmia recurrence risk in group 2 than in group 1 (HR, 2.42; 95% CI, 1.16-5.03; P = .018). A higher procedural termination rate was observed in group 1 than groups 2 and 3 (41.7% vs 17.2% vs 18.8%; P = .042). On multivariate analysis, VOM ethanol injection was an independent predictor of freedom from recurrence of AF (HR, 0.20; 95% CI, 0.08-0.52; P = .001) and atrial arrhythmia (HR, 0.35; 95% CI, 0.17-0.74; P = .005), whereas a left atrial diameter >45 mm and hypertension were independent risk factors for recurrence. Periprocedural complications rates were comparable among the groups. Conclusion Adjunctive VOM ethanol infusion is effective and safe for treating nonparoxysmal AF in patients with VOM triggers and/or refractory mitral flutter, providing good long-term freedom from AF and atrial arrhythmia.

Background Supervised learning from high-throughput sequencing data presents many challenges. For one, the curse of dimensionality often leads to overfitting as well as issues with scalability. This can bring about inaccurate models or those that require extensive compute time and resources. Additionally, variant calls may not be the optimal encoding for a given learning task, which also contributes to poor predictive capabilities. To address these issues, we present Harvestman, a method that takes advantage of hierarchical relationships among the possible biological interpretations and representations of genomic variants to perform automatic feature learning, feature selection, and model building. Results We demonstrate that Harvestman scales to thousands of genomes comprising more than 84 million variants by processing phase 3 data from the 1000 Genomes Project, one of the largest publicly available collection of whole genome sequences. Using breast cancer data from The Cancer Genome Atlas, we show that Harvestman selects a rich combination of representations that are adapted to the learning task, and performs better than a binary representation of SNPs alone. We compare Harvestman to existing feature selection methods and demonstrate that our method is more parsimonious—it selects smaller and less redundant feature subsets while maintaining accuracy of the resulting classifier. Conclusion Harvestman is a hierarchical feature selection approach for supervised model building from variant call data. By building a knowledge graph over genomic variants and solving an integer linear program , Harvestman automatically and optimally finds the right encoding for genomic variants. Compared to other hierarchical feature selection methods, Harvestman is faster and selects features more parsimoniously.

Rough set theory is useful mathematical tool developed to deal with vagueness and uncertainty. As an important concept of rough set theory, an attribute reduct is a subset of attributes that are jointly sufficient and individually necessary for preserving a particular property of the given information table. Rough set theory is also the most popular for generating decision rules from decision table. In this paper, we propose an algorithm finding object reduct of consistent decsion table. On the other hand, we also show an algorithm to find some attribute reducts and the correctness of our algorithms is proof-theoretically. These our algorithms have polynomial time complexity. Our finding object reduct helps other algorithms of finding attribute reducts become more effectively, especially as working with huge consistent decision table.

Aims Most left atrial tachycardia (LAT) is associated with atrial fibrillation (AF). The clinical and electrophysiological characteristics and outcomes of LAT without AF have not been investigated. This study sought to determine the long-term ablation outcomes and predictors of recurrence of isolated LAT. Methods This is a single-center study of consecutive patients with isolated LAT. Atrial arrhythmia recurrence was determined from follow-up records of patients who underwent LAT ablation from 2008 to 2017. Clinical and electrophysiologic characteristics associated with atrial arrhythmia recurrence were identified. Results A total of 50 patients (53 ± 19 years, 46% male) with 59 LAT (1.16 ± 0.47 per patient) were enrolled. Over a mean follow-up of 37 ± 33 months, atrial arrhythmia recurrence occurred in 22 (44%) patients, 11 with atrial tachycardia (AT) only, five with AF only, and six with concurrent AT and AF. The incidence of pulmonary vein (PV) origins increased significantly in the repeat procedure (P = 0.036). Multivariate analysis identified left ventricular ejection fraction (LVEF) as the only predictor of any atrial arrhythmia recurrence and LAT recurrence, while smoking and identified macroreentrant LAT in the index procedure predicted AF recurrence. Conclusion This study demonstrated a higher rate of atrial arrhythmia recurrence, including AF, among patients with initially isolated LAT. A lower LVEF predicted any atrial arrhythmia and LAT recurrence, whereas smoking and index macroreentrant AT mechanism predicted long-term AF. PV ATs were frequently observed in recurrent patients irrespective of index procedure origin.