Papers

Gaussian processes are widely used for accurate emulation of unknown surfaces in sequential design of expensive simulation experiments. Integrated mean squared error (IMSE) is an effective acquisition function for sequential designs based on Gaussian processes. However, existing approaches struggle with its implementation because the required integrals often lack closed-form expressions for most kernel functions. We propose a novel and computationally efficient Hilbert space Gaussian process approximation for the IMSE acquisition function, where a truncated eigenbasis representation of the integral enables closed-form evaluation. We establish sharp global non-asymptotic bounds for both the approximation error of isotropic kernels and the resulting error in the acquisition function. In a series of numerical experiments with $\gamma$-stabilizing, the proposed method achieves substantially lower prediction error and reduced computation time compared to existing benchmarks. These results demonstrate that the proposed Hilbert space Gaussian process framework provides an accurate and computationally efficient approach for Gaussian process based sequential design.

We introduce and study the problem of calibrating conditional risk, which involves estimating the expected loss of a prediction model conditional on input features. We analyze this problem in both classification and regression settings and show that it is fundamentally equivalent to a standard regression task. For classification settings, we further establish a connection between conditional risk calibration and individual/conditional probability calibration, and develop theoretical insights for the performance metric. This reveals that while conditional risk calibration is related to existing uncertainty quantification problems, it remains a distinct and standalone machine learning problem. Empirically, we validate our theoretical findings and demonstrate the practical implications of conditional risk calibration in the learning to defer (L2D) framework. Our systematic experiments provide both qualitative and quantitative assessments, offering guidance for future research in uncertainty-aware decision-making.

Forecasting the life-cycle trajectory of a newly launched product is important for launch planning, resource allocation, and early risk assessment. This task is especially difficult in the pre-launch and early post-launch phases, when product-specific outcome history is limited or unavailable, creating a cold-start problem. In these phases, firms must make decisions before demand patterns become reliably observable, while early signals are often sparse, noisy, and unstable We propose the Conditional Diffusion Life-cycle Forecaster (CDLF), a conditional generative framework for forecasting new-product life-cycle trajectories under cold start. CDLF combines three sources of information: static descriptors, reference trajectories from similar products, and newly arriving observations when available. Here, static descriptors refer to structured pre-launch characteristics of the product, such as category, price tier, brand or organization identity, scale, and access conditions. This structure allows the model to condition forecasts on relevant product context and to update them adaptively over time without retraining, yielding flexible multi-modal predictive distributions under extreme data scarcity. The method satisfies consistency with a horizon-uniform distributional error bound for recursive generation. Across studies on Intel microprocessor stock keeping unit (SKU) life cycles and the platform-mediated adoption of open large language model repositories, CDLF delivers more accurate point forecasts and higher-quality probabilistic forecasts than classical diffusion models, Bayesian updating approaches, and other state-of-the-art machine-learning baselines.

Ditlevsen and Ditlevsen [Nature Communications, 2023] (DD23 hereafter) propose a statistical framework to estimate the timing of a potential collapse of the Atlantic Meridional Overturning Circulation (AMOC) based on extrapolating information from observed sea-surface temperature (SST) variability. By fitting a stochastic one-dimensional fold-bifurcation model to an SST-based fingerprint of the AMOC using Maximum Likelihood Estimation (MLE), they conclude that a collapse is most likely to occur in the middle of the 21st century, with a reported 95% confidence interval covering the time span from 2037 to 2109. Given the profound implications of such a claim for both climate and society, it is essential to thoroughly test the robustness of this result, to critically assess the underlying assumptions and uncertainties, and to estimate the extent to which the reported confidence interval reflects the true limits of current knowledge. Here we examine the sensitivity of DD23's results and argue that four types of uncertainty are insufficiently explored in their analysis: (i) structural uncertainty associated with the assumed low-order bifurcation model, (ii) statistical uncertainty in their model fit, (iii) uncertainty in the representativeness of SST-based fingerprints as proxies for the high-dimensional AMOC dynamics, and (iv) uncertainty in the underlying data, arising from non-stationary observational coverage and dataset preprocessing. Using synthetic experiments and a systematic analysis of alternative fingerprints and observational products, we show that the tipping times estimated by DD23 are highly sensitive to the uncertainties listed above, and extend several millennia into the future when these uncertainties are thoroughly propagated.

The zero-inflated logistic regression model accommodates binary responses with excess zeros, which often arise from a latent mixture of susceptible and insusceptible subpopulations or asymmetric misclassification of the response. The model has two components: regression for the binary response and a latent binary indicator for the zero-inflation state. In applied settings, it is common to use the same design matrix for both components if there is no prior knowledge. However, this shared-design specification lacks guaranteed identifiability of the regression parameters, as established in prior works. This paper investigates the theoretical properties of the zero-inflated logistic regression model under the shared-design setting and computational methods for applications. First, to motivate the use of the zero-inflated model, we prove that ignoring the zero-inflation mechanism can lead to a sign flip in the pseudo-true coefficient value relative to the true value. We then establish sufficient conditions for the existence of the maximum likelihood estimate. As a main result, we establish that the model under the shared-design setting is identifiable up to exchange symmetry of the parameters for two components and that the expected log-likelihood has a unique maximizer on the resulting quotient space. The posterior bimodality is examined using a Pólya-Gamma Gibbs sampler with replica exchange. Finally, we propose a simple relabeling rule to select a single ordered parameter pair, and evaluate its performance through simulation studies and an application to self-reported diabetes data.

We consider the problem of sampling from a probability distribution $\pi$. It is well known that this can be written as an optimisation problem over the space of probability distributions in which we aim to minimise the Kullback–Leibler divergence from $\pi$. We consider the effect of replacing $\pi$ with a sequence of moving targets $(\pi_t)_{t\ge0}$ defined via geometric tempering on the Wasserstein and Fisher–Rao gradient flows. We show that convergence occurs exponentially in continuous time, providing novel bounds in both cases. We also consider popular time discretisations and explore their convergence properties. We show that in the Fisher–Rao case, replacing the target distribution with a geometric mixture of initial and target distribution never leads to a convergence speed up both in continuous time and in discrete time. Finally, we explore the gradient flow structure of tempered dynamics and derive novel adaptive tempering schedules.

Survival analysis is a widely used statistical framework for modeling time-to-event data under censoring. Classical methods, such as the Cox proportional hazards (Cox PH) model, offer a semiparametric approach to estimating the effects of covariates on the hazard function. Despite its importance, survival analysis has been largely unexplored in online settings, particularly within the bandit framework, where decisions must be made sequentially to optimize treatments as new data arrive over time. In this work, we take an initial step toward integrating survival analysis into a purely online learning setting under the Cox PH model, addressing key challenges including staggered entry, delayed feedback, and right censoring. We adapt three canonical bandit algorithms to balance exploration and exploitation, with theoretical guarantees of sublinear regret bounds. Extensive simulations and semi-real experiments using SEER cancer data demonstrate that our approach enables rapid and effective learning of near-optimal treatment policies.

Football fans frequently exhibit pronounced emotional and physiological reactions during high-stakes matches. However, the temporal dynamics of this football fever are rarely modeled as a latent process. Using intensive longitudinal data from Arminia Bielefeld supporters who wore smartwatches during the 2025 German Football Association (DFB) Cup final, we investigate how football fever unfolds. The devices recorded heart rate, stress level, and related indicators in short intervals, allowing us to construct a latent variable for football fever and model its dynamics. We specify a time-dependent structural equation model with latent growth components and autoregressive effects to capture both overall trends and short-term carry-over effects in fans' physiological responses. Results are aggregated across multiple imputations of missing measurements. Model fit is evaluated using adjustments for the high data dimensionality. The results show that football fever follows a V-shaped trajectory: high at kick-off, followed by a steady decline until the renewed arousal in the second half, with substantial between-fan heterogeneity in both baseline level and temporal dynamics. Our findings demonstrate that football fever can be adequately represented as a latent variable using structural equation modeling and reflected by wearable technology data. This highlights the importance of accounting for temporal dependence when studying dynamic emotional phenomena, e. g., in sports spectatorship.

The analysis of neural representation has become an integral part of research aiming to better understand the inner workings of neural networks. While there are many different approaches to investigate neural representations, an important line of research has focused on doing so through the lens of intrinsic dimensions (IDs). Although this perspective has provided valuable insights and stimulated substantial follow-up research, important limitations of this approach have remained largely unaddressed. In this paper, we highlight a crucial discrepancy between theory and practice of IDs in neural representations, theoretically and empirically showing that common ID estimators are, in fact, not tracking the true underlying ID of the representation. We contrast this negative result with an investigation of the underlying factors that may drive commonly reported ID-related results on neural representation in the literature. Building on these insights, we offer a new perspective on ID estimation in neural representations.

Weighted networks encode not only the presence of interactions but also their strength. Existing methods for weighted network community detection often rely on Poisson models, which can be restrictive for overdispersed data and make efficient posterior computation difficult when covariates are incorporated. We propose Bayesian stochastic block models based on the zero-inflated negative binomial distribution: ZINB-SBM without covariates and CZINB-SBM with pairwise covariates. The proposed models accommodate overdispersion, naturally account for missing interactions through zero inflation, and admit efficient Gibbs sampling. In CZINB-SBM, Pólya-Gamma data augmentation enables posterior inference for regression coefficients with uncertainty quantification. We further employ a dynamic mixture of finite mixtures, which allows the number of communities to be inferred from the data and can lead to more accurate clustering. Simulation studies show that ZINB-SBM is more robust than a zero-inflated Poisson SBM for highly overdispersed networks. Real data analysis demonstrates interpretable block specific covariate effects and substantially improved missing link prediction compared with a Poisson regression-based Bayesian SBM.

This invited paper proposes and discusses several Bayesian attempts at nonparametric and semiparametric density estimation. The main categories of these ideas are as follows: 1) Build a nonparametric prior around a given parametric model. We look at cases where the nonparametric part of the construction is a Dirichlet process or relatives thereof. (2) Express the density as an additive expansion of orthogonal basis functions, and place priors on the coefficients. Here attention is given to a certain robust Hermite expansion around the normal distribution. Multiplicative expansions are also considered. (3) Express the unknown density as locally being of a certain parametric form, then construct suitable local likelihood functions to express information content, and place local priors on the local parameters.

Depth is widely viewed as a central contributor to the success of deep neural networks, whereas standard neural network approximation theory typically provides guarantees only for the final output and leaves the role of intermediate layers largely unclear. We address this gap by developing a quantitative framework in which depth admits a precise scale-dependent interpretation. Specifically, we design a single shared mixed-activation architecture of fixed width $2dN+d+2$ and any prescribed finite depth such that each intermediate readout $\Phi_\ell$ is itself an approximant to the target function $f$. For $f\in L^p([0,1]^d)$ with $p\in [1,\infty)$, the approximation error of $\Phi_\ell$ is controlled by $(2d+1)$ times the $L^p$ modulus of continuity at the geometric scale $N^{-\ell}$ for all $\ell$. The estimate reduces to the geometric rate $(2d+1)N^{-\ell}$ if $f$ is $1$-Lipschitz. Our network design is inspired by multigrade deep learning, where depth serves as a progressive refinement mechanism. For every prescribed terminal depth, the construction yields a finite nested family of prefix readouts whose earlier correction terms remain embedded in later readouts. Thus the approximation may be truncated within the prescribed depth range once the desired certified accuracy is reached.

Consider betting against a sequence of $[0,1]$-valued observations using bets that are fair if the data have a conditional mean $m_0 \in (0,1)$. Cover's universal portfolio algorithm achieves a worst-case regret of $O(\ln n)$ compared to the best constant bet in hindsight, and this logarithmic order is unimprovable against adversarial data. We construct an explicit full-support Robbins mixture that combines insights from Robbins and Cover. It retains $O(\ln n)$ regret on every path while also eventually satisfying $R_n\le(1+o(1))\ln\ln V_n$ on every path with bounded mixture wealth and $V_n\to\infty$, where $V_n$ is an intrinsic variance process. It also attains the optimal asymptotic wealth growth rate under iid alternatives. Under the conditional-mean claim and almost-sure divergence of $V_n$, the sharper regret bound holds almost surely, and its leading constant $1$ is unimprovable. We also give explicit finite-time, path-wise regret bounds. A simple hedging principle shows how splitting initial wealth between two strategies achieves a best-of-both-worlds guarantee by retaining the smaller of their regrets and the larger of their growth rates. We contrast our results to those for unbounded data in Agrawal and Ramdas (2026). Their worst-case regret is unbounded. However, hedging yields almost-sure iterated-logarithmic regret together with optimal growth. Finally, our wealth process witnesses a sharp game-theoretic upper law of iterated logarithm, analogous to Shafer and Vovk (2005).

Multi-task learning is effective for related applications, but its performance can deteriorate when the target sample size is small. Transfer learning can borrow strength from related studies; yet, many existing methods rely on restrictive bounded-difference assumptions between the source and target models. We propose SMART, a spectral transfer method for multi-task linear regression that instead assumes spectral similarity: the target left and right singular subspaces lie within the corresponding source subspaces and are sparsely aligned with the source singular bases. Such an assumption is natural when studies share latent structures and enables transfer beyond the bounded-difference settings. SMART estimates the target coefficient matrix through structured regularization that incorporates spectral information from a source study. Importantly, it requires only a fitted source model rather than the raw source data, making it useful when data sharing is limited. Although the optimization problem is nonconvex, we develop a practical ADMM-based algorithm. We establish general, non-asymptotic error bounds and a minimax lower bound in the noiseless-source regime. Under additional regularity conditions, these results yield near-minimax Frobenius error rates up to logarithmic factors. Simulations confirm improved estimation accuracy and robustness to negative transfer, and analysis of multi-modal single-cell data demonstrates better predictive performance. The Python implementation of SMART, along with the code to reproduce all experiments in this paper, is publicly available at https://github.com/boxinz17/smart.

Bilevel optimization and bilevel minimax optimization have recently emerged as unifying frameworks for a range of machine-learning tasks, including hyperparameter optimization and reinforcement learning. The existing literature focuses on empirical efficiency and convergence guarantees, leaving a critical theoretical gap in understanding how well these algorithms generalize. To bridge this gap, we provide the first systematic generalization analysis for first-order gradient-based bilevel minimax solvers with lower-level minimax problems. Specifically, by leveraging algorithmic stability arguments, we derive fine-grained generalization bounds for three representative algorithms, including single-timescale stochastic gradient descent-ascent, and two variants of two-timescale stochastic gradient descent-ascent. Our results reveal a precise trade-off among algorithmic stability, generalization gaps, and practical settings. Furthermore, extensive empirical evaluations corroborate our theoretical insights on realistic optimization tasks with bilevel minimax structures.

Sparse additive models have attracted much attention in high-dimensional data analysis due to their flexible representation and strong interpretability. However, most existing models are limited to single-level learning under the mean-squared error criterion, whose empirical performance can degrade significantly in the presence of complex noise, such as non-Gaussian perturbations, outliers, noisy labels, and imbalanced categories. The sample reweighting strategy is widely used to reduce the model's sensitivity to atypical data; however, it typically requires prespecifying the weighting functions and manually selecting additional hyperparameters. To address this issue, we propose a new meta additive model (MAM) based on the bilevel optimization framework, which learns data-driven weighting of individual losses by parameterizing the weighting function via an MLP trained on meta data. MAM is capable of a variety of learning tasks, including variable selection, robust regression estimation, and imbalanced classification. Theoretically, MAM provides guarantees on convergence in computation, algorithmic generalization, and variable selection consistency under mild conditions. Empirically, MAM outperforms several state-of-the-art additive models on both synthetic and real-world data under various data corruptions.

Graph-based techniques and spectral graph theory have enriched the field of machine learning with a variety of critical advances. A central object in the analysis is the graph Laplacian L, which encodes the structure of the graph. We consider the problem of learning over this Laplacian in a distributed streaming setting, where new edges of the graph are observed in real time by a network of workers. In this setting, it is hard to learn quickly or approximately while keeping a distributed representation of L. To address this challenge, we present a novel algorithm, GSQUEAK, which efficiently sparsifies the Laplacian by maintaining a small subset of effective resistances. We show that our algorithm produces sparsifiers with strong spectral approximation guarantees, all while processing edges in a single pass and in a distributed fashion.

Large-scale kernel ridge regression (KRR) is limited by the need to store a large kernel matrix K_t. To avoid storing the entire matrix K_t, Nystrom methods subsample a subset of columns of the kernel matrix, and efficiently find an approximate KRR solution on the reconstructed matrix. The chosen subsampling distribution in turn affects the statistical and computational tradeoffs. For KRR problems, recent works show that a sampling distribution proportional to the ridge leverage scores (RLSs) provides strong reconstruction guarantees for the approximation. While exact RLSs are as difficult to compute as a KRR solution, we may be able to approximate them well enough. In this paper, we study KRR problems in a sequential setting and introduce the INK-ESTIMATE algorithm, that incrementally computes the RLSs estimates. INK-ESTIMATE maintains a small sketch of K_t, that at each step is used to compute an intermediate estimate of the RLSs. First, our sketch update does not require access to previously seen columns, and therefore a single pass over the kernel matrix is sufficient. Second, the algorithm requires a fixed, small space budget to run dependent only on the effective dimension of the kernel matrix. Finally, our sketch provides strong approximation guarantees on the distance between the true kernel matrix and its approximation, and on the statistical risk of the approximate KRR solution at any time, because all our guarantees hold at any intermediate step.

A popular approach to apprenticeship learning (AL) is to formulate it as an inverse reinforcement learning (IRL) problem. The MaxEnt-IRL algorithm successfully integrates the maximum entropy principle into IRL and unlike its predecessors, it resolves the ambiguity arising from the fact that a possibly large number of policies could match the expert's behavior. In this paper, we study an AL setting in which in addition to the expert's trajectories, a number of unsupervised trajectories is available. We introduce MESSI, a novel algorithm that combines MaxEnt-IRL with principles coming from semi-supervised learning. In particular, MESSI integrates the unsupervised data into the MaxEnt-IRL framework using a pairwise penalty on trajectories. Empirical results in a highway driving and grid-world problems indicate that MESSI is able to take advantage of the unsupervised trajectories and improve the performance of MaxEnt-IRL.

Inference for time series of networks often relies on accurate vertex correspondence between network realizations at different times. In practice, such vertex alignments can be misspecified or unknown. We study the impact of errorful vertex alignment on changepoint localization for dynamic networks through two illustrative models, with the key distinction being whether changepoint information is contained in marginal or joint distributions of the time-varying latent positions. We compare localization techniques ranging from the simple network statistic of average degree to the modern procedure of Euclidean mirrors. Our results demonstrate that understanding the interplay of marginal and joint information in the observed network time series is key to robust network inference.

Infectious disease dynamics operate across multiple biological scales, with within-host viral dynamics being a key driver of between-host transmission. However, while models that explicitly link these scales exist, none have been developed with statistical inference as a primary goal. In this paper we propose a multiscale model that jointly captures heterogeneous individual-level viral load trajectories and stochastic household transmission, and develop efficient inference methods to fit it to data. Since full joint inference is computationally difficult, we employ a cut approach that passes information from the within-host to the between-host model but not vice versa. This enables the data on viral loads to inform the transmission parameters such as the infection times and symptom onset thresholds. We evaluate the framework on simulated household outbreak data, assessing parameter recovery, computational efficiency, and the effect of viral load sampling frequency on inference quality. Parameter recovery is unbiased when the sampling frequency of the viral loads is high enough. When sampling is sparse, some bias is introduced, but incorporating external viral load data can mitigate this.

We study the deterministic global optimization of trained Gaussian process posterior mean functions over hyperrectangular domains. Although the posterior mean function has a compact closed-form representation, its global optimization is challenging because it remains nonlinear and nonconvex. Existing exact deterministic approaches become increasingly difficult to scale as the number of training data points grows, leading to approximation-based methods that improve tractability by optimizing a modified (inexact) objective. In this work, we propose PALM-Mean, a piecewise-analytic lower-bounding framework embedded in reduced-space spatial branch-and-bound. At each node, kernel terms that are locally important are replaced by a sign-aware piecewise-linear relaxation in an appropriate scalar distance variable, while the remaining terms are bounded analytically in closed form. We show this hybrid approach yields a valid lower bound for the posterior mean, while limiting the size of the branch-and-bound subproblems. We establish validity of the node lower bounds and $\varepsilon$-global convergence of the resulting algorithm. Computational results on synthetic benchmarks and real-world application problems show that PALM-Mean improves scalability relative to representative general-purpose deterministic global solvers, particularly as the number of training data points increases.

We study the community detection problem in the non-uniform hypergraph stochastic block model (HSBM), where hyperedges of varying sizes coexist. This setting captures higher-order and multi-view interactions and raises a fundamental question: can multiple uniform hypergraph layers below the detection threshold be combined to enable weak recovery? We answer this question by establishing a Kesten–Stigum-type bound for weak recovery in a general class of non-uniform HSBMs with $r$ blocks, generated according to multiple symmetric probability tensors. In the case $r=2$, we show that weak recovery is possible whenever the sum of the signal-to-noise ratios across all uniform hypergraph layers exceeds one, thereby confirming the positive part of a conjecture in (Chodrow et al., 2023). Moreover, we provide a polynomial-time spectral algorithm that achieves this threshold via an optimally weighted non-backtracking operator. For the unweighted non-backtracking matrix, our spectral method attains a different algorithmic threshold, also conjectured in (Chodrow et al., 2023). Our approach develops a spectral theory for weighted non-backtracking operators on non-uniform hypergraphs, including a precise characterization of outlier eigenvalues and eigenvector overlaps. We introduce a novel Ihara–Bass formula tailored to weighted non-uniform hypergraphs, which yields an efficient low-dimensional representation and leads to a provable spectral reconstruction algorithm. Taken together, these results provide a principled and computationally efficient approach to clustering in non-uniform hypergraphs, and highlight the role of optimal weighting in aggregating heterogeneous higher-order interactions.

Estimating average treatment effects from observational data is challenging under practical violations of the positivity assumption. Targeted Maximum Likelihood Estimators (TMLEs) are widely used because of their double robustness and efficiency, but they can remain sensitive to such violations. We conduct extensive simulation studies to examine how targeting strategies and truncation levels affect TMLE performance under varying degrees of outcome regression misspecification and practical positivity stress. We show that loss-weighted targeting can induce substantial systematic bias relative to clever-covariate-scaled targeting, while insufficient truncation for clever-covariate-scaled targeting leads to inflated variance and unstable estimation. We further find that fixed truncation rules of the form c/(sqrt(n) log n), especially with c = 5 or c = 6, provide robust practical defaults in many settings, although the optimal choice varies with sample size. Motivated by the limitations of standard Lepski selection, we propose a Lepski-type adaptive truncation procedure with a brake mechanism that improves stability in data-adaptive tuning. We also compare variance estimators and find that targeted bootstrap variance estimation provides a stable alternative across truncation levels.

We propose Annealed Langevin Monte Carlo for Flow ODE Sampling (ALMC-ODE), a method for generating samples from unnormalized target distributions, with a particular emphasis on multimodal densities that are challenging for standard Markov chain Monte Carlo methods. ALMC-ODE is based on a probability-flow ordinary differential equation (ODE) derived from stochastic interpolants, which continuously transports a standard Gaussian reference distribution at $t = 0$ to the target distribution $\rho$ at $t = 1$. The key innovation lies in an annealed Langevin Markov chain that evolves through a sequence of intermediate distributions bridging the reference and the target. The resulting importance-weighted particles, reweighted via a Jarzynski-based scheme, yield a low-variance estimator of the velocity field governing the ODE. On the theoretical side, we establish a Jarzynski-type reweighting identity for general time-inhomogeneous transition kernels, characterize the optimal backward kernel that minimizes the variance of the importance weights, and prove an $\mathcal{O}(1/n)$ mean squared error bound for the resulting velocity-field estimator. Numerical experiments on challenging benchmarks, including Gaussian mixture models and a 64-dimensional Allen–Cahn field system, demonstrate that ALMC-ODE significantly outperforms both direct Monte Carlo ODE approaches and Hamiltonian Monte Carlo when applied to highly multimodal target distributions.

We present a general nonparametric approach for testing whether a statistical parameter defined through conditional distributions is constant across the conditioning variables. Such hypotheses arise naturally in problems such as assessing treatment effect heterogeneity, conditional associational effects, and conditional mean dependence. Our framework studies function-valued parameters obtained by evaluating a smooth statistical functional on conditional probability distributions. We establish an explicit connection between our test and procedures based on studying the norm of the function-valued parameter. Unlike many existing norm-based tests, which exhibit poor asymptotic behavior under the null, the proposed test statistic admits a tractable limiting null distribution. We illustrate the applicability of the proposed test through several examples, assess its operating characteristics in simulation studies, and apply it to data from a breast cancer trial to identify predictive biomarkers for response to adjuvant chemotherapy.

We consider Decision-Focused Federated Learning (DFFL), a predict-then-optimize setting in which multiple clients collaboratively train predictive models for downstream linear optimization problems without exchanging raw data. Besides the data heterogeneity typical of standard federated learning, clients may also have different objective functions and feasible regions. Building on the SPO+ surrogate loss, we derive heterogeneity bounds that separate objective shift, measured through cost-vector distances, from feasible-set shift, measured through support-function and shape-distance terms. We show that, for general compact feasible sets, small objective perturbations can still induce nonvanishing decision-focused loss discrepancies, while strongly convex feasible regions yield sharper stability-based bounds. We then lift these pointwise bounds to a local-versus-federated excess-risk comparison, showing that federation is beneficial when the statistical advantage of pooling exceeds a client-specific heterogeneity penalty. Computational experiments on polyhedral and strongly convex problems confirm that federation is substantially more robust under strongly convex feasible regions. Finally, we evaluate a simple validation-based interpolation between local and federated DFFL models. This interpolation mitigates the theoretical tradeoff and reduces aggregate regret and worst-client harm in both synthetic experiments and a PJM energy-pricing case study.

In many statistical applications, particularly in clinical studies, hypotheses may carry different levels of importance, motivating the use of weighted multiple testing procedures (wMTPs) to control the familywise error rate (FWER). Among these approaches, two weighted Holm procedures are commonly used: the weighted Holm procedure (WHP), which is based on ordered weighted $p$-values, and the weighted alternative Holm procedure (WAP), which relies on ordered raw $p$-values. This paper provides a systematic comparison of these two procedures, along with practical recommendations for their use. We first examine their corresponding closed testing procedures (CTPs) and show that WHP is uniformly more powerful than WAP. We further investigate their structural properties, demonstrating that WAP, while consonant, lacks monotonicity. To facilitate communication with non-statisticians, we introduce graphical representations of both procedures using a common initial graph and distinct updating strategies. In addition, we derive adjusted $p$-values and adjusted weighted $p$-values for both methods. Finally, we establish an optimality result: WHP cannot be improved by enlarging any of its critical values without violating FWER control, whereas WAP is optimal only under specific conditions. Simulation studies support these theoretical findings and highlight the superior FWER control and average power of WHP.

Network meta-analysis of diagnostic test accuracy (NMA-DTA) is a relatively new field, involving combining evidence across studies to evaluate and compare the accuracy of different tests for a given condition. However, the methods proposed to date cannot always capture complex aspects of the data. In fact, many commonly used diagnostic tests are continuous biomarkers, whose accuracy is evaluated at multiple thresholds within a study. Using current NMA-DTA methods we are feasibly able to include in our analysis only a few thresholds per study, discarding this way a big amount of data which could have provided us with useful information. We introduce an approach that can efficiently encompass all available data. This is a hierarchical model that incorporates multinomial likelihoods for studies reporting results across multiple thresholds and a parametric structure for the relationship between the probability of testing positive and threshold within each disease class. This approach enables us to obtain accuracy estimates of tests across the whole range of observed thresholds, while it retains all the useful properties of standard NMA-DTA methods. We explore different variations of this model based on different covariance structures, the inclusion of study-level random effects, and the addition of a further hierarchical structure on the test-level variance components. This framework is applied to data from two systematic reviews, allowing the inclusion of a larger number of tests (compared to alternative approaches) and estimation of sensitivity and specificity at different thresholds with increased precision.

Learning about causal effects in target populations and their subsets may be facilitated by combining information from multiple sources. One major class of study designs that combine information involves appending an index study with data from an external comparator, which may facilitate head-to-head comparisons of treatments initially studied in different populations. We delineate external comparator analyses under two distinct, but related, identification strategies. The first strategy relies on exchangeability (transportability) of potential outcome means, which uses information only on the treatments that are to be compared. The second strategy relies on transportability in effect measure, requiring additional use of information on a third treatment common to the populations that have been combined. In a time-fixed setting with a point treatment and non-failure time outcome, we examine identification and estimation under a basic setup where information from an index trial is combined with a second, and external to the index trial, data source. We propose estimators for identifying observed data functionals, with a particular focus on semiparametric efficient augmented weighting estimators that incorporate models for the probability of trial participation, the probability of treatment, and conditional outcome means. We derive the asymptotic properties of these augmented weighting estimators – including robustness to model misspecification and slower rates of convergence for some nuisance function models – and use simulation to compare their finite sample performance to estimators based only on outcome modeling or weighting. Last, we provide a practical demonstration of the proposed methods by combining the ACCEPT and PHOENIX 1 randomized trials to evaluate the effect of various biologic agents on plaque psoriasis, a chronic inflammatory disorder.