Collecting gold-standard phenotype data via manual extraction is typically labor-intensive and slow, whereas automated computational phenotypes (ACPs) offer a systematic and much faster alternative. However, simply replacing the gold-standard with ACPs, without acknowledging their differences, could lead to biased results and misleading conclusions. Motivated by the complexity of incorporating ACPs while maintaining the validity of downstream analyses, in this paper, we consider a semi-supervised learning setting that consists of both labeled data (with gold-standard) and unlabeled data (without gold-standard), under the covariate shift framework. We develop doubly robust and semiparametrically efficient estimators that leverage ACPs for general target parameters in the unlabeled and combined populations. In addition, we carefully analyze the efficiency gains achieved by incorporating ACPs, comparing scenarios with and without their inclusion. Notably, we identify that ACPs for the unlabeled data, instead of for the labeled data, drive the enhanced efficiency gains. To validate our theoretical findings, we conduct comprehensive synthetic experiments and apply our method to multiple real-world datasets, confirming the practical advantages of our approach.

A primary challenge facing modern scientific research is the limited availability of gold-standard data, which can be costly, labor-intensive, or invasive to obtain. With the rapid development of machine learning (ML), scientists can now employ ML algorithms to predict gold-standard outcomes using variables that are easier to obtain. However, these predicted outcomes are often used directly in subsequent statistical analyses, ignoring imprecision and heterogeneity introduced by the prediction procedure. This will likely result in false positive findings and invalid scientific conclusions. In this work, we introduce PoSt-Prediction Adaptive inference (PSPA) that allows valid and powerful inference based on ML-predicted data. Its “assumption-lean” property guarantees reliable statistical inference without assumptions on the ML prediction. Its “data-adaptive” feature guarantees an efficiency gain over existing methods, regardless of the accuracy of ML prediction. We demonstrate the statistical superiority and broad applicability of our method through simulations and real-data applications.

The presence of distribution shifts poses a significant challenge for deploying modern machine learning models in real-world applications. This work focuses on the target shift problem in a regression setting (Zhang et al., 2013; Nguyen et al., 2016). More specifically, the target variable $y$ (also known as the response variable), which is continuous, has different marginal distributions in the training source and testing domain, while the conditional distribution of features $\boldsymbol{x}$ given $y$ remains the same. While most literature focuses on classification tasks with finite target space, the regression problem has an *infinite dimensional* target space, which makes many of the existing methods inapplicable. In this work, we show that the continuous target shift problem can be addressed by estimating the importance weight function from an ill-posed integral equation. We propose a nonparametric regularized approach named *ReTaSA* to solve the ill-posed integral equation and provide theoretical justification for the estimated importance weight function. The effectiveness of the proposed method has been demonstrated with extensive numerical studies on synthetic and real-world datasets.

High speed train (HST) communications with multi-user multiple-input multiple-output (MU-MIMO) techniques have shown great potential in system performance improvements. However, the challenges caused by higher pilot overhead and the belated channel state information (CSI) feedback in such high mobility scenarios still need to be further addressed. To this end, fully-decoupled radio access network (FD-RAN) with novel location-based feedback-free transmission scheme and cooperative transmission/reception in separated downlink/uplink networks is regarded as a promising solution. In this paper, we study the uplink precoding design in FD-RAN for HST communications. To capture the inherent relation between location and precoding, the line-of-sight (LoS) channel is derived from location and then fed into a proposed precoding design neural network (PDNN) which learns to jointly optimize the precoding scheme for all the multi-antenna mobile relays (MRs) in uplink MU-MIMO. We adopt a custom loss function to optimize the spectrum efficiency (SE). Moreover, a joint signal reception method is given, also based solely on the LoS channel derived from location, so as to avoid frequent pilot transmission. Simulation results show the advantages of FD-RAN against other architectures and demonstrate that our proposed PDNN achieves better performance than traditional precoding scheme in high mobility scenarios.

With the ever-increasing number of connected vehicles in the fifth-generation mobile communication networks (5G) and beyond 5G (B5G), ensuring the reliability and high-speed demand of cellular vehicle-to-everything (C-V2X) communication in scenarios where vehicles are moving at high speeds poses a significant challenge. Recently, multiconnectivity technology has become a promising network access paradigm for improving network performance and reliability for C-V2X in the 5G and B5G era. To this end, this article proposes an analytical framework for the performance of downlink in multiconnectivity C-V2X networks. Specifically, by modeling the vehicles and base stations (BSs) as 1-D Poisson point processes, we first derive and analyze the joint distance distribution of multiconnectivity. Then through leveraging the tools of stochastic geometry, the coverage probability and spectral efficiency are obtained based on the previous results for general multiconnectivity cases in C-V2X. Additionally, we evaluate the effect of the path-loss exponent and the density of downlink BS on system performance indicators. We demonstrate through extensive Monte Carlo simulations that multiconnectivity technology can effectively enhance network performance in C-V2X. Our findings have important implications for the research and application of multiconnectivity C-V2X in the 5G and B5G era.

The increasingly severe energy consumption caused by exploding wireless demands attracts considerable research. Remarkably, base station (BS) sleeping is a promising technique to enable the green network. A disruptive and original fully-decoupled radio access network (FD-RAN) architecture aiming at the next-generation mobile communication networks is developed, which removes the obstacles to achieving BS sleeping, i.e. deficient cooperation between BSs, coupled data-control transmission and coupled uplink-downlink transmission. In this paper, we investigate the joint user association and uplink BS sleeping considering power control in the FD-RAN with the superiority of fully decoupled architectures. Specifically, we propose an energy consumption model for the uplink FD-RAN and tackle the mixed-integer second-order cone problem to minimize the whole network energy consumption by leveraging the many-to-many swap matching theory. Extensive simulation results validate a higher energy efficiency of the uplink FD-RAN compared to the traditional cellular network and cell-free networks and demonstrate the effectiveness of our proposed algorithm.

We study the domain adaptation problem with label shift in this work. Under the label shift context, the marginal distribution of the label varies across the training and testing datasets, while the conditional distribution of features given the label is the same. Traditional label shift adaptation methods either suffer from large estimation errors or require cumbersome post-prediction calibrations. To address these issues, we first propose a moment-matching framework for adapting the label shift based on the geometry of the influence function. Under such a framework, we propose a novel method named $\underline{\mathrm{E}}$fficient $\underline{\mathrm{L}}$abel $\underline{\mathrm{S}}$hift $\underline{\mathrm{A}}$daptation (ELSA), in which the adaptation weights can be estimated by solving linear systems. Theoretically, the ELSA estimator is $\sqrt{n}$-consistent ($n$ is the sample size of the source data) and asymptotically normal. Empirically, we show that ELSA can achieve state-of-the-art estimation performances without post-prediction calibrations, thus, gaining computational efficiency.

We consider a fully-decoupled radio access network (FD-RAN), where base stations (BSs) are physically decoupled into control BSs, uplink BSs and downlink BSs, and multi-connectivity becomes the default user equipment (UE) association mode. Specifically, we study the inter-frequency multi-connectivity in downlink of FD-RAN and present a deep reinforcement learning based online multi-connectivity mobility management scheme. We formulate a UE dynamic multiple access problem and transform it into a handover decision problem, then apply the double deep Q-network (DDQN) algorithm to make real time mobility management decisions. Simulation results show that the proposed scheme outperforms benchmarks in terms of handover frequency and quality of service, while ensuring real-time performance.

Conducting valid statistical analyses is challenging in the presence of missing-not-at-random (MNAR) data, where the missingness mechanism is dependent on the missing values themselves even conditioned on the observed data. Here, we consider a MNAR model that generalizes several prior popular MNAR models in two ways: first, it is less restrictive in terms of statistical independence assumptions imposed on the underlying joint data distribution, and second, it allows for all variables in the observed sample to have missing values. This MNAR model corresponds to a so-called criss-cross structure considered in the literature on graphical models of missing data that prevents nonparametric identification of the entire missing data model. Nonetheless, part of the complete-data distribution remains nonparametrically identifiable. By exploiting this fact and considering a rich class of exponential family distributions, we establish sufficient conditions for identification of the complete-data distribution as well as the entire missingness mechanism. We then propose methods for testing the independence restrictions encoded in such models using odds ratio as our parameter of interest. We adopt two semiparametric approaches for estimating the odds ratio parameter and establish the corresponding asymptotic theories: one involves maximizing a conditional likelihood with order statistics and the other uses estimating equations. The utility of our methods is illustrated via simulation studies.

With the development of the Internet of Everything (IoE), future 6G networks will face the challenge of massive terminal access. However, deploying substantial high-cost, full-function base stations will undoubtedly further increase the cost of mobile network deployment, making it difficult for mobile operators to afford it. In this paper, we tackle the problem of low-cost network deployment for fully-decoupled radio access network (FD-RAN) with personalized service for large-scale terminals. We first propose a techno-economic cost model (TECM) for FD-RAN deployment based on the techno-economic approach. Then, we further formulate a cost-minimization problem for decoupled network deployment. Based on the independence brought by uplink and downlink decoupling in FD-RANs, we decompose the original problem into separate subproblems for uplink and downlink network deployment. In the following, we propose a branch and cut based network deployment (BCND) algorithm to solve two decoupled deployment subproblems, respectively. Finally, simulation results show that FD-RANs have significant cost advantages when facing differentiated service demands, and the main factors affecting network cost are power consumption and rental costs.

To enable flexible base stations (BS) association and dynamic resource management for personalized user equipment (UE) download service provision in the next-generation mobile communication network (6G), in this paper, we investigate the downlink (DL) transmission scenario in an origin fully-decoupled radio access network (FD-RAN) architecture. Considering the unique fully-decoupled UL/DL access feature, we propose an efficient two-stage DL channel estimation method in the FD-RAN. We formulate a novel multi-connectivity and dynamic resource cooperation problem with joint multiple-BS and multiple-UE association and coordinated beamforming, aiming at maximizing the weighted sum achievable rate in DL FD-RAN. By leveraging the many-to-many swap-matching theory and fractional relaxation approach, we solve the dynamic UE scheduling problem with multiple-BS and multiple-UE association and the coordinated beamforming problem, respectively. Extensive simulation results based on standard 3GPP 36.873 urban micro channel demonstrate that the proposed framework can improve the average spectral efficiency by 34.9% as compared to the traditional maximum ratio transmission beamforming method.

To cope with the even more urgent spectrum and energy efficiency challenge for trillion-level terminal access and data uploading in the next generation mobile communication network (6G), in this paper, we investigate the uplink transmission in an original fully-decoupled radio access networks (FD-RAN) architecture. Specifically, we propose a resilient uplink base station cooperative reception framework in FD-RAN, which is a large-scale fading based two-tier signal combination approach for the uplink transmission, including the localized signal combination at the base station and centralized signal combination at the edge cloud, respectively. Then, we formulate a weighted sum-rate maximization problem for the uplink transmission optimization, and decompose it into two subproblems. A spectrum-efficiency maximized virtual service cluster selection (SEMVS) algorithm is designed by leveraging the channel statistical information for solving subproblem one, and a fractional programming based power control (FPPC) algorithm is introduced for the power optimization of subproblem two. Compared to the typical RAN architectures with corresponding access and power control methods, simulation results demonstrate the significant performance improvements of uplink FD-RAN with the proposed solution.

Recent advancements of Vehicle-to-Everything (V2X) communication combined with artificial intelligence (AI) technologies have shown enormous potentials for improving traffic management efficiency and intelligence. To provide innovative and effective data-driven traffic management solution for the coming automated vehicle era, we present a vehicle-road collaboration-enabled nonsignalized intersection management architecture in this paper. First, by dividing the intersection zone into the central section (CS) and the waiting section (WS), a vehicle regulation scheme involved with communication and computation planes is developed for V2X-enabled nonsignalized intersection management. Specifically, in order to guarantee vehicle safety, the definition of no overlapping occupation time in CS and the fastest crossing time point (FCTP) algorithm are employed for vehicle collision avoidance. Second, considering the relative coordination between adjacent intersections, a multiagent-based deep reinforcement learning scheduling (MA-DRLS) algorithm is proposed to realize cooperative multiple intersection management. Through information exchange with different intersection agents, each agent can obtain an optimal scheduling strategy using independent deep reinforcement learning (DRL) network. The features of fixed Q-targets and experience replay are leveraged to improve the reliability of neural network during the training process. Finally, simulation performances in terms of intersection throughput and vehicle waiting time have been provided to validate the effectiveness and demonstrate the superiority of the proposed nonsignalized intersection management solution.

Recent advancement of connected vehicles technologies combined with machine learning (MA) methods has shown great potential for the improvement of efficiency of Intelligent Transportation System. In this work, considering the spatio-temporal correlations under vehicle distribution on urban road network, neural network based deep learning solution is adopted to obtain vehicle driving characteristics and predict future traffic conditions. First, to address the huge challenge brought by complex traffic environment, we present a fine-grained regional-level forecast structure for the prediction of traffic flow at each road. After that, a residual network based deep learning traffic prediction algorithm called DST-RGTP is proposed for the performance enhancement of vehicle regulation in the entire traffic system. Finally, we use the real traffic data of Beijing and open-source road network data on Openstreetmap to test the proposed method. Simulation results verify the accuracy of prediction approach DST-RGTP, which can help to improve the urban traffic management efficiency.