Search

In the flourishing live streaming industry, accurate recognition of streamers' emotions has become a critical research focus, with profound implications for audience engagement and content optimization. However, precise emotion coding typically requires manual annotation by trained experts, making it extremely expensive and time-consuming to obtain complete observational data for large-scale studies. Motivated by this challenge in streamer emotion recognition, we develop here a novel imputation method together with a principled statistical inference procedure for analyzing partially observed binary data. Specifically, we assume for each observation an auxiliary feature vector, which is sufficiently cheap to be fully collected for the whole sample. We next assume a small pilot sample with both the target binary covariates (i.e., the emotion status) and the auxiliary features fully observed, of which the size could be considerably smaller than that of the whole sample. Thereafter, a regression model can be constructed for the target binary covariates and the auxiliary features. This enables us to impute the missing binary features using the fully observed auxiliary features for the entire sample. We establish the associated asymptotic theory for principled statistical inference and present extensive simulation experiments, demonstrating the effectiveness and theoretical soundness of our proposed method. Furthermore, we validate our approach using a comprehensive dataset on emotion recognition in live streaming, demonstrating that our imputation method yields smaller standard errors and is more statistically efficient than using pilot data only. Our findings have significant implications for enhancing user experience and optimizing engagement on streaming platforms.

We develop here a novel transfer learning methodology called Profiled Transfer Learning (PTL). The method is based on the \textit{approximate-linear} assumption between the source and target parameters. Compared with the commonly assumed \textit{vanishing-difference} assumption and \textit{low-rank} assumption in the literature, the \textit{approximate-linear} assumption is more flexible and less stringent. Specifically, the PTL estimator is constructed by two major steps. Firstly, we regress the response on the transferred feature, leading to the profiled responses. Subsequently, we learn the regression relationship between profiled responses and the covariates on the target data. The final estimator is then assembled based on the \textit{approximate-linear} relationship. To theoretically support the PTL estimator, we derive the non-asymptotic upper bound and minimax lower bound. We find that the PTL estimator is minimax optimal under appropriate regularity conditions. Extensive simulation studies are presented to demonstrate the finite sample performance of the new method. A real data example about sentence prediction is also presented with very encouraging results.

We study here a Gaussian Mixture Model (GMM) with rare events data. In this case, the commonly used Expectation-Maximization (EM) algorithm exhibits extremely slow numerical convergence rate. To theoretically understand this phenomenon, we formulate the numerical convergence problem of the EM algorithm with rare events data as a problem about a contraction operator. Theoretical analysis reveals that the spectral radius of the contraction operator in this case could be arbitrarily close to 1 asymptotically. This theoretical finding explains the empirical slow numerical convergence of the EM algorithm with rare events data. To overcome this challenge, a Mixed EM (MEM) algorithm is developed, which utilizes the information provided by partially labeled data. As compared with the standard EM algorithm, the key feature of the MEM algorithm is that it requires additionally labeled data. We find that MEM algorithm significantly improves the numerical convergence rate as compared with the standard EM algorithm. The finite sample performance of the proposed method is illustrated by both simulation studies and a real-world dataset of Swedish traffic signs.

In this study, we develop a latent factor model for analysing high-dimensional binary data. Specifically, a standard probit model is used to describe the regression relationship between the observed binary data and the continuous latent variables. Our method assumes that the dependency structure of the observed binary data can be fully captured by the continuous latent factors. To estimate the model, a moment-based estimation method is developed. The proposed method is able to deal with both discontinuity and high dimensionality. Most importantly, the asymptotic properties of the resulting estimators are rigorously established. Extensive simulation studies are presented to demonstrate the proposed methodology. A real dataset about product descriptions is analysed for illustration.

Computed tomography (CT) has been a powerful diagnostic tool since its emergence in the 1970s. Using CT data, three-dimensional (3D) structures of human internal organs and tissues, such as blood vessels, can be reconstructed using professional software. This 3D reconstruction is crucial for surgical operations and can serve as a vivid medical teaching example. However, traditional 3D reconstruction heavily relies on manual operations, which are time-consuming, subjective, and require substantial experience. To address this problem, we develop a novel semiparametric Gaussian mixture model tailored for the 3D reconstruction of blood vessels. This model extends the classical Gaussian mixture model by enabling nonparametric variations in the component-wise parameters of interest according to voxel positions. We develop a kernel-based expectation-maximization algorithm for estimating the model parameters, accompanied by a supporting asymptotic theory. Furthermore, we propose a novel regression method for optimal bandwidth selection. Compared to the conventional cross-validation-based (CV) method, the regression method outperforms the CV method in terms of computational and statistical efficiency. In application, this methodology facilitates the fully automated reconstruction of 3D blood vessel structures with remarkable accuracy.

Spatial autoregressive (SAR) models are important tools for studying network effects. However, with an increasing emphasis on data privacy, data providers often implement privacy protection measures that make classical SAR models inapplicable. In this study, we introduce a privacy-protected SAR model with noise-added response and covariates to meet privacy-protection requirements. However, in this scenario, the traditional quasi-maximum likelihood estimator becomes infeasible because the likelihood function cannot be directly formulated. To address this issue, we first consider an explicit expression for the likelihood function with only noise-added responses. Then, we develop techniques to correct the biases for derivatives introduced by noise. Correspondingly, a Newton-Raphson-type algorithm is proposed to obtain the estimator, leading to a corrected likelihood estimator. To further enhance computational efficiency, we introduce a corrected least squares estimator based on the idea of bias correction. These two estimation methods ensure both data security and the attainment of statistically valid estimators. Theoretical analysis of both estimators is carefully conducted, statistical inference methods and model extensions are discussed. The finite sample performances of different methods are demonstrated through extensive simulations and the analysis of a real dataset.

This paper presents a selective review of statistical computation methods for massive data analysis. A huge amount of statistical methods for massive data computation have been rapidly developed in the past decades. In this work, we focus on three categories of statistical computation methods: (1) distributed computing, (2) subsampling methods, and (3) minibatch gradient techniques. The first class of literature is about distributed computing and focuses on the situation, where the dataset size is too huge to be comfortably handled by one single computer. In this case, a distributed computation system with multiple computers has to be utilized. The second class of literature is about subsampling methods and concerns about the situation, where the sample size of dataset is small enough to be placed on one single computer but too large to be easily processed by its memory as a whole. The last class of literature studies those minibatch gradient related optimization techniques, which have been extensively used for optimizing various deep learning models.

This article introduces CluBear, a Python-based open-source package for interactive massive data analysis. The key feature of CluBear is that it enables users to conduct convenient and interactive statistical analysis of massive data with only a traditional single-computer system. Thus, CluBear provides a cost-effective solution when mining large-scale datasets. In addition, the CluBear package integrates many commonly used statistical and graphical tools, which are useful for most commonly encountered data analysis tasks.

In this paper, we studied a buffered mini-batch gradient descent (BMGD) algorithm for training complex model on massive datasets. The algorithm studied here is designed for fast training on a GPU-CPU system, which contains two steps: the buffering step and the computation step. In the buffering step, a large batch of data (i.e., a buffer) are loaded from the hard drive to the graphical memory of GPU. In the computation step, a standard mini-batch gradient descent(MGD) algorithm is applied to the buffered data. Compared to traditional MGD algorithm, the proposed BMGD algorithm can be more efficient for two reasons.First, the BMGD algorithm uses the buffered data for multiple rounds of gradient update, which reduces the expensive communication cost from the hard drive to GPU memory. Second, the buffering step can be executed in parallel so that the GPU does not have to stay idle when loading new data. We first investigate the theoretical properties of BMGD algorithms under a linear regression setting.The analysis is then extended to the Polyak-Lojasiewicz loss function class. The theoretical claims about the BMGD algorithm are numerically verified by simulation studies. The practical usefulness of the proposed method is demonstrated by three image-related real data analysis.

Testing judicial impartiality is a problem of fundamental importance in empirical legal studies, for which standard regression methods have been popularly used to estimate the extralegal factor effects. However, those methods cannot handle control variables with ultrahigh dimensionality, such as found in judgment documents recorded in text format. To solve this problem, we develop a novel mixture conditional regression (MCR) approach, assuming that the whole sample can be classified into a number of latent classes. Within each latent class, a standard linear regression model can be used to model the relationship between the response and a key feature vector, which is assumed to be of a fixed dimension. Meanwhile, ultrahigh dimensional control variables are then used to determine the latent class membership, where a Na\"ive Bayes type model is used to describe the relationship. Hence, the dimension of control variables is allowed to be arbitrarily high. A novel expectation-maximization algorithm is developed for model estimation. Therefore, we are able to estimate the interested key parameters as efficiently as if the true class membership were known in advance. Simulation studies are presented to demonstrate the proposed MCR method. A real dataset of Chinese burglary offenses is analyzed for illustration purpose.

The utilization of multi-layer network structures now enables the explanation of complex systems in nature from multiple perspectives. Multi-layer academic networks capture diverse relationships among academic entities, facilitating the study of academic development and the prediction of future directions. However, there are currently few academic network datasets that simultaneously consider multi-layer academic networks; often, they only include a single layer. In this study, we provide a large-scale multi-layer academic network dataset, namely, LMANStat, which includes collaboration, co-institution, citation, co-citation, journal citation, author citation, author-paper and keyword co-occurrence networks. Furthermore, each layer of the multi-layer academic network is dynamic. Additionally, we expand the attributes of nodes, such as authors' research interests, productivity, region and institution. Supported by this dataset, it is possible to study the development and evolution of statistical disciplines from multiple perspectives. This dataset also provides fertile ground for studying complex systems with multi-layer structures.

Distributed computing is critically important for modern statistical analysis. Herein, we develop a distributed quasi-Newton (DQN) framework with excellent statistical, computation, and communication efficiency. In the DQN method, no Hessian matrix inversion or communication is needed. This considerably reduces the computation and communication complexity of the proposed method. Notably, related existing methods only analyze numerical convergence and require a diverging number of iterations to converge. However, we investigate the statistical properties of the DQN method and theoretically demonstrate that the resulting estimator is statistically efficient over a small number of iterations under mild conditions. Extensive numerical analyses demonstrate the finite sample performance., Comment: 56 pages, 3 figures

Large-scale networks are commonly encountered in practice (e.g., Facebook and Twitter) by researchers. In order to study the network interaction between different nodes of large-scale networks, the spatial autoregressive (SAR) model has been popularly employed. Despite its popularity, the estimation of a SAR model on large-scale networks remains very challenging. On the one hand, due to policy limitations or high collection costs, it is often impossible for independent researchers to observe or collect all network information. On the other hand, even if the entire network is accessible, estimating the SAR model using the quasi-maximum likelihood estimator (QMLE) could be computationally infeasible due to its high computational cost. To address these challenges, we propose here a subnetwork estimation method based on QMLE for the SAR model. By using appropriate sampling methods, a subnetwork, consisting of a much-reduced number of nodes, can be constructed. Subsequently, the standard QMLE can be computed by treating the sampled subnetwork as if it were the entire network. This leads to a significant reduction in information collection and model computation costs, which increases the practical feasibility of the effort. Theoretically, we show that the subnetwork-based QMLE is consistent and asymptotically normal under appropriate regularity conditions. Extensive simulation studies, based on both simulated and real network structures, are presented.

We study here a fixed mini-batch gradient decent (FMGD) algorithm to solve optimization problems with massive datasets. In FMGD, the whole sample is split into multiple non-overlapping partitions. Once the partitions are formed, they are then fixed throughout the rest of the algorithm. For convenience, we refer to the fixed partitions as fixed mini-batches. Then for each computation iteration, the gradients are sequentially calculated on each fixed mini-batch. Because the size of fixed mini-batches is typically much smaller than the whole sample size, it can be easily computed. This leads to much reduced computation cost for each computational iteration. It makes FMGD computationally efficient and practically more feasible. To demonstrate the theoretical properties of FMGD, we start with a linear regression model with a constant learning rate. We study its numerical convergence and statistical efficiency properties. We find that sufficiently small learning rates are necessarily required for both numerical convergence and statistical efficiency. Nevertheless, an extremely small learning rate might lead to painfully slow numerical convergence. To solve the problem, a diminishing learning rate scheduling strategy can be used. This leads to the FMGD estimator with faster numerical convergence and better statistical efficiency. Finally, the FMGD algorithms with random shuffling and a general loss function are also studied.

Labeling mistakes are frequently encountered in real-world applications. If not treated well, the labeling mistakes can deteriorate the classification performances of a model seriously. To address this issue, we propose an improved Naive Bayes method for text classification. It is analytically simple and free of subjective judgements on the correct and incorrect labels. By specifying the generating mechanism of incorrect labels, we optimize the corresponding log-likelihood function iteratively by using an EM algorithm. Our simulation and experiment results show that the improved Naive Bayes method greatly improves the performances of the Naive Bayes method with mislabeled data.

Bagging is a useful method for large-scale statistical analysis, especially when the computing resources are very limited. We study here the asymptotic properties of bagging estimators for $M$-estimation problems but with massive datasets. We theoretically prove that the resulting estimator is consistent and asymptotically normal under appropriate conditions. The results show that the bagging estimator can achieve the optimal statistical efficiency, provided that the bagging subsample size and the number of subsamples are sufficiently large. Moreover, we derive a variance estimator for valid asymptotic inference. All theoretical findings are further verified by extensive simulation studies. Finally, we apply the bagging method to the US Airline Dataset to demonstrate its practical usefulness.

Modern statistical analysis often encounters high dimensional models but with limited sample sizes. This makes the target data based statistical estimation very difficult. Then how to borrow information from another large sized source data for more accurate target model estimation becomes an interesting problem. This leads to the useful idea of transfer learning. Various estimation methods in this regard have been developed recently. In this work, we study transfer learning from a different perspective. Specifically, we consider here the problem of testing for transfer learning sufficiency. By transfer learning sufficiency (denoted as the null hypothesis), we mean that, with the help of the source data, the useful information contained in the feature vectors of the target data can be sufficiently extracted for predicting the interested target response. Therefore, the rejection of the null hypothesis implies that information useful for prediction remains in the feature vectors of the target data and thus calls for further exploration. To this end, we develop a novel testing procedure and a centralized and standardized test statistic, whose asymptotic null distribution is analytically derived. Simulation studies are presented to demonstrate the finite sample performance of the proposed method. A deep learning related real data example is presented for illustration purpose.

The rapid emergence of massive datasets in various fields poses a serious challenge to traditional statistical methods. Meanwhile, it provides opportunities for researchers to develop novel algorithms. Inspired by the idea of divide-and-conquer, various distributed frameworks for statistical estimation and inference have been proposed. They were developed to deal with large-scale statistical optimization problems. This paper aims to provide a comprehensive review for related literature. It includes parametric models, nonparametric models, and other frequently used models. Their key ideas and theoretical properties are summarized. The trade-off between communication cost and estimate precision together with other concerns are discussed.

Modern statistical analysis often encounters datasets with large sizes. For these datasets, conventional estimation methods can hardly be used immediately because practitioners often suffer from limited computational resources. In most cases, they do not have powerful computational resources (e.g., Hadoop or Spark). How to practically analyze large datasets with limited computational resources then becomes a problem of great importance. To solve this problem, we propose here a novel subsampling-based method with jackknifing. The key idea is to treat the whole sample data as if they were the population. Then, multiple subsamples with greatly reduced sizes are obtained by the method of simple random sampling with replacement. It is remarkable that we do not recommend sampling methods without replacement because this would incur a significant cost for data processing on the hard drive. Such cost does not exist if the data are processed in memory. Because subsampled data have relatively small sizes, they can be comfortably read into computer memory as a whole and then processed easily. Based on subsampled datasets, jackknife-debiased estimators can be obtained for the target parameter. The resulting estimators are statistically consistent, with an extremely small bias. Finally, the jackknife-debiased estimators from different subsamples are averaged together to form the final estimator. We theoretically show that the final estimator is consistent and asymptotically normal. Its asymptotic statistical efficiency can be as good as that of the whole sample estimator under very mild conditions. The proposed method is simple enough to be easily implemented on most practical computer systems and thus should have very wide applicability.

Large-scale rare events data are commonly encountered in practice. To tackle the massive rare events data, we propose a novel distributed estimation method for logistic regression in a distributed system. For a distributed framework, we face the following two challenges. The first challenge is how to distribute the data. In this regard, two different distribution strategies (i.e., the RANDOM strategy and the COPY strategy) are investigated. The second challenge is how to select an appropriate type of objective function so that the best asymptotic efficiency can be achieved. Then, the under-sampled (US) and inverse probability weighted (IPW) types of objective functions are considered. Our results suggest that the COPY strategy together with the IPW objective function is the best solution for distributed logistic regression with rare events. The finite sample performance of the distributed methods is demonstrated by simulation studies and a real-world Sweden Traffic Sign dataset.

The bootstrap is a widely used procedure for statistical inference because of its simplicity and attractive statistical properties. However, the vanilla version of bootstrap is no longer feasible computationally for many modern massive datasets due to the need to repeatedly resample the entire data. Therefore, several improvements to the bootstrap method have been made in recent years, which assess the quality of estimators by subsampling the full dataset before resampling the subsamples. Naturally, the performance of these modern subsampling methods is influenced by tuning parameters such as the size of subsamples, the number of subsamples, and the number of resamples per subsample. In this paper, we develop a novel hyperparameter selection methodology for selecting these tuning parameters. Formulated as an optimization problem to find the optimal value of some measure of accuracy of an estimator subject to computational cost, our framework provides closed-form solutions for the optimal hyperparameter values for subsampled bootstrap, subsampled double bootstrap and bag of little bootstraps, at no or little extra time cost. Using the mean square errors as a proxy of the accuracy measure, we apply our methodology to study, compare and improve the performance of these modern versions of bootstrap developed for massive data through simulation study. The results are promising.

In this paper, we propose a dictionary screening method for embedding compression in text classification tasks. The key purpose of this method is to evaluate the importance of each keyword in the dictionary. To this end, we first train a pre-specified recurrent neural network-based model using a full dictionary. This leads to a benchmark model, which we then use to obtain the predicted class probabilities for each sample in a dataset. Next, to evaluate the impact of each keyword in affecting the predicted class probabilities, we develop a novel method for assessing the importance of each keyword in a dictionary. Consequently, each keyword can be screened, and only the most important keywords are reserved. With these screened keywords, a new dictionary with a considerably reduced size can be constructed. Accordingly, the original text sequence can be substantially compressed. The proposed method leads to significant reductions in terms of parameters, average text sequence, and dictionary size. Meanwhile, the prediction power remains very competitive compared to the benchmark model. Extensive numerical studies are presented to demonstrate the empirical performance of the proposed method.

The rapid growth of online network platforms generates large-scale network data and it poses great challenges for statistical analysis using the spatial autoregression (SAR) model. In this work, we develop a novel distributed estimation and statistical inference framework for the SAR model on a distributed system. We first propose a distributed network least squares approximation (DNLSA) method. This enables us to obtain a one-step estimator by taking a weighted average of local estimators on each worker. Afterwards, a refined two-step estimation is designed to further reduce the estimation bias. For statistical inference, we utilize a random projection method to reduce the expensive communication cost. Theoretically, we show the consistency and asymptotic normality of both the one-step and two-step estimators. In addition, we provide theoretical guarantee of the distributed statistical inference procedure. The theoretical findings and computational advantages are validated by several numerical simulations implemented on the Spark system. Lastly, an experiment on the Yelp dataset further illustrates the usefulness of the proposed methodology.

We study a fully decentralized federated learning algorithm, which is a novel gradient descent algorithm executed on a communication-based network. For convenience, we refer to it as a network gradient descent (NGD) method. In the NGD method, only statistics (e.g., parameter estimates) need to be communicated, minimizing the risk of privacy. Meanwhile, different clients communicate with each other directly according to a carefully designed network structure without a central master. This greatly enhances the reliability of the entire algorithm. Those nice properties inspire us to carefully study the NGD method both theoretically and numerically. Theoretically, we start with a classical linear regression model. We find that both the learning rate and the network structure play significant roles in determining the NGD estimator's statistical efficiency. The resulting NGD estimator can be statistically as efficient as the global estimator, if the learning rate is sufficiently small and the network structure is well balanced, even if the data are distributed heterogeneously. Those interesting findings are then extended to general models and loss functions. Extensive numerical studies are presented to corroborate our theoretical findings. Classical deep learning models are also presented for illustration purpose.

Momentum methods have been shown to accelerate the convergence of the standard gradient descent algorithm in practice and theory. In particular, the minibatch-based gradient descent methods with momentum (MGDM) are widely used to solve large-scale optimization problems with massive datasets. Despite the success of the MGDM methods in practice, their theoretical properties are still underexplored. To this end, we investigate the theoretical properties of MGDM methods based on the linear regression models. We first study the numerical convergence properties of the MGDM algorithm and further provide the theoretically optimal tuning parameters specification to achieve faster convergence rate. In addition, we explore the relationship between the statistical properties of the resulting MGDM estimator and the tuning parameters. Based on these theoretical findings, we give the conditions for the resulting estimator to achieve the optimal statistical efficiency. Finally, extensive numerical experiments are conducted to verify our theoretical results., Comment: 45 pages, 5 figures

The emergence of massive data in recent years brings challenges to automatic statistical inference. This is particularly true if the data are too numerous to be read into memory as a whole. Accordingly, new sampling techniques are needed to sample data from a hard drive. In this paper, we propose a sequential addressing subsampling (SAS) method, that can sample data directly from the hard drive. The newly proposed SAS method is time saving in terms of addressing cost compared to that of the random addressing subsampling (RAS) method. Estimators (e.g., the sample mean) based on the SAS subsamples are constructed, and their properties are studied. We conduct a series of simulation studies to verify the finite sample performance of the proposed SAS estimators. The time cost is also compared between the SAS and RAS methods. An analysis of the airline data is presented for illustration purpose.

This article introduces subbagging (subsample aggregating) estimation approaches for big data analysis with memory constraints of computers. Specifically, for the whole dataset with size $N$, $m_N$ subsamples are randomly drawn, and each subsample with a subsample size $k_N\ll N$ to meet the memory constraint is sampled uniformly without replacement. Aggregating the estimators of $m_N$ subsamples can lead to subbagging estimation. To analyze the theoretical properties of the subbagging estimator, we adapt the incomplete $U$-statistics theory with an infinite order kernel to allow overlapping drawn subsamples in the sampling procedure. Utilizing this novel theoretical framework, we demonstrate that via a proper hyperparameter selection of $k_N$ and $m_N$, the subbagging estimator can achieve $\sqrt{N}$-consistency and asymptotic normality under the condition $(k_Nm_N)/N\to \alpha \in (0,\infty]$. Compared to the full sample estimator, we theoretically show that the $\sqrt{N}$-consistent subbagging estimator has an inflation rate of $1/\alpha$ in its asymptotic variance. Simulation experiments are presented to demonstrate the finite sample performances. An American airline dataset is analyzed to illustrate that the subbagging estimate is numerically close to the full sample estimate, and can be computationally fast under the memory constraint.

Modern statistical analyses often encounter datasets with massive sizes and heavy-tailed distributions. For datasets with massive sizes, traditional estimation methods can hardly be used to estimate the extreme value index directly. To address the issue, we propose here a subsampling-based method. Specifically, multiple subsamples are drawn from the whole dataset by using the technique of simple random subsampling with replacement. Based on each subsample, an approximate maximum likelihood estimator can be computed. The resulting estimators are then averaged to form a more accurate one. Under appropriate regularity conditions, we show theoretically that the proposed estimator is consistent and asymptotically normal. With the help of the estimated extreme value index, we can estimate high-level quantiles and tail probabilities of a heavy-tailed random variable consistently. Extensive simulation experiments are provided to demonstrate the promising performance of our method. A real data analysis is also presented for illustration purpose.

Massive data analysis becomes increasingly prevalent, subsampling methods like BLB (Bag of Little Bootstraps) serves as powerful tools for assessing the quality of estimators for massive data. However, the performance of the subsampling methods are highly influenced by the selection of tuning parameters ( e.g., the subset size, number of resamples per subset ). In this article we develop a hyperparameter selection methodology, which can be used to select tuning parameters for subsampling methods. Specifically, by a careful theoretical analysis, we find an analytically simple and elegant relationship between the asymptotic efficiency of various subsampling estimators and their hyperparameters. This leads to an optimal choice of the hyperparameters. More specifically, for an arbitrarily specified hyperparameter set, we can improve it to be a new set of hyperparameters with no extra CPU time cost, but the resulting estimator's statistical efficiency can be much improved. Both simulation studies and real data analysis demonstrate the superior advantage of our method.

In deep learning tasks, the learning rate determines the update step size in each iteration, which plays a critical role in gradient-based optimization. However, the determination of the appropriate learning rate in practice typically replies on subjective judgement. In this work, we propose a novel optimization method based on local quadratic approximation (LQA). In each update step, given the gradient direction, we locally approximate the loss function by a standard quadratic function of the learning rate. Then, we propose an approximation step to obtain a nearly optimal learning rate in a computationally efficient way. The proposed LQA method has three important features. First, the learning rate is automatically determined in each update step. Second, it is dynamically adjusted according to the current loss function value and the parameter estimates. Third, with the gradient direction fixed, the proposed method leads to nearly the greatest reduction in terms of the loss function. Extensive experiments have been conducted to prove the strengths of the proposed LQA method., Comment: 10 pages, 5 figures

Distributed systems have been widely used in practice to accomplish data analysis tasks of huge scales. In this work, we target on the estimation problem of generalized linear models on a distributed system with nonrandomly distributed data. We develop a Pseudo-Newton-Raphson algorithm for efficient estimation. In this algorithm, we first obtain a pilot estimator based on a small random sample collected from different Workers. Then conduct one-step updating based on the computed derivatives of log-likelihood functions in each Worker at the pilot estimator. The final one-step estimator is proved to be statistically efficient as the global estimator, even with nonrandomly distributed data. In addition, it is computationally efficient, in terms of both communication cost and storage usage. Based on the one-step estimator, we also develop a likelihood ratio test for hypothesis testing. The theoretical properties of the one-step estimator and the corresponding likelihood ratio test are investigated. The finite sample performances are assessed through simulations. Finally, an American Airline dataset is analyzed on a Spark cluster for illustration purpose.

In this work, we develop a distributed least squares approximation (DLSA) method that is able to solve a large family of regression problems (e.g., linear regression, logistic regression, and Cox's model) on a distributed system. By approximating the local objective function using a local quadratic form, we are able to obtain a combined estimator by taking a weighted average of local estimators. The resulting estimator is proved to be statistically as efficient as the global estimator. Moreover, it requires only one round of communication. We further conduct a shrinkage estimation based on the DLSA estimation using an adaptive Lasso approach. The solution can be easily obtained by using the LARS algorithm on the master node. It is theoretically shown that the resulting estimator possesses the oracle property and is selection consistent by using a newly designed distributed Bayesian information criterion (DBIC). The finite sample performance and computational efficiency are further illustrated by an extensive numerical study and an airline dataset. The airline dataset is 52 GB in size. The entire methodology has been implemented in Python for a {\it de-facto} standard Spark system. The proposed DLSA algorithm on the Spark system takes 26 minutes to obtain a logistic regression estimator, which is more efficient and memory friendly than conventional methods.