Search

Eigenspace estimation is a fundamental tool in data analytics, which has found applications in PCA, dimension reduction, and clustering, among others. The modern machine learning community usually involves data that come from and belong to different organizations. The low communication power and possible data privacy breaches make the eigenspace estimation challenging. To address these issues, we propose a class of algorithms called FedPower within the federated learning (FL) framework. FedPower leverages the well-known power method by alternating multiple local power iterations and a global aggregation step, thus improving communication efficiency. In the aggregation, we propose to weight each local eigenvector matrix with Orthogonal Procrustes Transformation (OPT) for better alignment. We add Gaussian noise in each iteration to ensure strong privacy protection by adopting the notion of differential privacy (DP). We provide convergence bounds for FedPower composed of different interpretable terms corresponding to the effects of Gaussian noise, parallelization, and random sampling of local machines. Additionally, we conduct experiments to demonstrate the effectiveness of our proposed algorithms. [ABSTRACT FROM AUTHOR]

We study the convergence rate of the famous Symmetric Rank-1 (SR1) algorithm, which has wide applications in different scenarios. Although it has been extensively investigated, SR1 still lacks a non-asymptotic superlinear rate compared with other quasi-Newton methods such as DFP and BFGS. In this paper, we address the aforementioned issue to obtain the first explicit non-asymptotic rates of superlinear convergence for the vanilla SR1 methods with a correction strategy that is used to achieve numerical stability. Specifically, the vanilla SR1 with the correction strategy achieves the rate of the form 2 n ln (4 ϰ) k k / 2 for general smooth strongly-convex functions where k is the iteration counter, ϰ is the condition number of the objective function, and n is the dimensionality of the problem. Furthermore, the vanilla SR1 algorithm enjoys a little faster convergence rate and can find the optima of the quadratic objective function at most n steps. [ABSTRACT FROM AUTHOR]

Endometriosis (EMs) is defined as the presence of tissue which somewhat resembles endometrial glands and stroma outside the uterus, and elicits fibrosis. Fibrosis is the main factor resulting in pain and infertility, while the aetiology of endometrial fibrosis is unknown. There is strong evidence from numerous experiments showing that connective tissue growth factor (CCN2) plays a central role in fibrogenesis. Exosomal miR-214-3p can regulate the expression of CCN2 through binding to complementary sites in the 3′ untranslated region. This study aimed to explore the role of exosomal miR-214-3p in endometriosis fibrosis and the relationship between CCN2 and miR-214-3p in endometriosis fibrosis. Our results demonstrated that miR-214-3p was significantly down-regulated and CCN2 was up-regulated in EMs ectopic lesion and stromal cells compared with EMs eutopic and endometrium of patients without endometriosis. Exosomal miR-214-3p can inhibit fibrosis in EMs through targeting CCN2. The results were explored and verified in vitro and in vivo, respectively. Cell co-culture was used to explore the contributions of exosomes to intercellular information transmission of miR-214-3p. The results showed that exosomes play a pivotal role in the transportation of miR-214-3p between cells. Furthermore, level of exosomal miR-214-3p in endometriosis patients' serum was lower than that in patients without endometriosis. In conclusion, exosomal miR-214-3p can inhibit fibrosis in EMs by targeting CCN2. MiR-214-3p may be considered as a bio-marker and has a potential therapeutic effect in EMs. [ABSTRACT FROM AUTHOR]

This paper addresses the learning algorithm on the unit sphere. The main purpose is to present an error analysis for regression generated by regularized least square algorithms with spherical harmonics kernel. The excess error can be estimated by the sum of sample errors and regularization errors. Our study shows that by introducing a suitable spherical harmonics kernel, the regularization parameter can decrease arbitrarily fast with the sample size. [ABSTRACT FROM AUTHOR]

The previous results describing the generalization ability of Empirical Risk Minimization (ERM) algorithm are usually based on the assumption of independent and identically distributed (i.i.d.) samples. In this paper we go far beyond this classical framework by establishing the first exponential bound on the rate of uniform convergence of the ERM algorithm with V-geometrically ergodic Markov chain samples, as the application of the bound on the rate of uniform convergence, we also obtain the generalization bounds of the ERM algorithm with V-geometrically ergodic Markov chain samples and prove that the ERM algorithm with V-geometrically ergodic Markov chain samples is consistent. The main results obtained in this paper extend the previously known results of i.i.d. observations to the case of V-geometrically ergodic Markov chain samples. [ABSTRACT FROM AUTHOR]

In this paper we propose an L regularizer which has a nonconvex penalty. The L regularizer is shown to have many promising properties such as unbiasedness, sparsity and oracle properties. A reweighed iterative algorithm is proposed so that the solution of the L regularizer can be solved through transforming it into the solution of a series of L regularizers. The solution of the L regularizer is more sparse than that of the L regularizer, while solving the L regularizer is much simpler than solving the L regularizer. The experiments show that the L regularizer is very useful and efficient, and can be taken as a representative of the L (0 > p > 1)regularizer. [ABSTRACT FROM AUTHOR]

To meet the growing electricity demand, China's power generation sector has become an increasingly large source of air pollutants. Specific control policymaking needs an inventory reflecting the overall, heterogeneous, time-varying features of power plant emissions. Due to the lack of comprehensive real measurements, existing inventories rely on average emission factors that suffer from many assumptions and high uncertainty. This study is the first to develop an inventory of particulate matter (PM), SO2 and NOX emissions from power plants using systematic actual measurements monitored by China's continuous emission monitoring systems (CEMS) network over 96–98% of the total thermal power capacity. With nationwide, source-level, real-time CEMS-monitored data, this study directly estimates emission factors and absolute emissions, avoiding the use of indirect average emission factors, thereby reducing the level of uncertainty. This dataset provides plant-level information on absolute emissions, fuel uses, generating capacities, geographic locations, etc. The dataset facilitates power emission characterization and clean air policy-making, and the CEMS-based estimation method can be employed by other countries seeking to regulate their power emissions. Measurement(s) air pollution Technology Type(s) computational modeling technique Sample Characteristic - Environment fossil fuel power plant Sample Characteristic - Location China Machine-accessible metadata file describing the reported data: 10.6084/m9.figshare.12813917 [ABSTRACT FROM AUTHOR]