Archives AI News

Understanding and Improving the Shampoo Optimizer via Kullback-Leibler Minimization

arXiv:2509.03378v1 Announce Type: new Abstract: As an adaptive method, Shampoo employs a structured second-moment estimation, and its effectiveness has attracted growing attention. Prior work has primarily analyzed its estimation scheme through the Frobenius norm. Motivated by the natural connection between the second moment and a covariance matrix, we propose studying Shampoo's estimation as covariance estimation through the lens of Kullback-Leibler (KL) minimization. This alternative perspective reveals a previously hidden limitation, motivating improvements to Shampoo's design. Building on this insight, we develop a practical estimation scheme, termed KL-Shampoo, that eliminates Shampoo's reliance on Adam for stabilization, thereby removing the additional memory overhead introduced by Adam. Preliminary results show that KL-Shampoo improves Shampoo's performance, enabling it to stabilize without Adam and even outperform its Adam-stabilized variant, SOAP, in neural network pretraining.

Off-Policy Learning in Large Action Spaces: Optimization Matters More Than Estimation

arXiv:2509.03456v1 Announce Type: new Abstract: Off-policy evaluation (OPE) and off-policy learning (OPL) are foundational for decision-making in offline contextual bandits. Recent advances in OPL primarily optimize OPE estimators with improved statistical properties, assuming that better estimators inherently yield superior policies. Although theoretically justified, we argue this estimator-centric approach neglects a critical practical obstacle: challenging optimization landscapes. In this paper, we provide theoretical insights and extensive empirical evidence showing that current OPL methods encounter severe optimization issues, particularly as action spaces become large. We demonstrate that simpler weighted log-likelihood objectives enjoy substantially better optimization properties and still recover competitive, often superior, learned policies. Our findings emphasize the necessity of explicitly addressing optimization considerations in the development of OPL algorithms for large action spaces.

Non-Linear Counterfactual Aggregate Optimization

arXiv:2509.03438v1 Announce Type: new Abstract: We consider the problem of directly optimizing a non-linear function of an outcome, where this outcome itself is the sum of many small contributions. The non-linearity of the function means that the problem is not equivalent to the maximization of the expectation of the individual contribution. By leveraging the concentration properties of the sum of individual outcomes, we derive a scalable descent algorithm that directly optimizes for our stated objective. This allows for instance to maximize the probability of successful A/B test, for which it can be wiser to target a success criterion, such as exceeding a given uplift, rather than chasing the highest expected payoff.

When a Reinforcement Learning Agent Encounters Unknown Unknowns

arXiv:2505.13188v2 Announce Type: replace-cross Abstract: An AI agent might surprisingly find she has reached an unknown state which she has never been aware of -- an unknown unknown. We mathematically ground this scenario in reinforcement learning: an agent, after taking an action calculated from value functions $Q$ and $V$ defined on the {it {aware domain}}, reaches a state out of the domain. To enable the agent to handle this scenario, we propose an {it episodic Markov decision {process} with growing awareness} (EMDP-GA) model, taking a new {it noninformative value expansion} (NIVE) approach to expand value functions to newly aware areas: when an agent arrives at an unknown unknown, value functions $Q$ and $V$ whereon are initialised by noninformative beliefs -- the averaged values on the aware domain. This design is out of respect for the complete absence of knowledge in the newly discovered state. The upper confidence bound momentum Q-learning is then adapted to the growing awareness for training the EMDP-GA model. We prove that (1) the regret of our approach is asymptotically consistent with the state of the art (SOTA) without exposure to unknown unknowns in an extremely uncertain environment, and (2) our computational complexity and space complexity are comparable with the SOTA -- these collectively suggest that though an unknown unknown is surprising, it will be asymptotically properly discovered with decent speed and an affordable cost.

Calibration Prediction Interval for Non-parametric Regression and Neural Networks

arXiv:2509.02735v1 Announce Type: cross Abstract: Accurate conditional prediction in the regression setting plays an important role in many real-world problems. Typically, a point prediction often falls short since no attempt is made to quantify the prediction accuracy. Classically, under the normality and linearity assumptions, the Prediction Interval (PI) for the response variable can be determined routinely based on the $t$ distribution. Unfortunately, these two assumptions are rarely met in practice. To fully avoid these two conditions, we develop a so-called calibration PI (cPI) which leverages estimations by Deep Neural Networks (DNN) or kernel methods. Moreover, the cPI can be easily adjusted to capture the estimation variability within the prediction procedure, which is a crucial error source often ignored in practice. Under regular assumptions, we verify that our cPI has an asymptotically valid coverage rate. We also demonstrate that cPI based on the kernel method ensures a coverage rate with a high probability when the sample size is large. Besides, with several conditions, the cPI based on DNN works even with finite samples. A comprehensive simulation study supports the usefulness of cPI, and the convincing performance of cPI with a short sample is confirmed with two empirical datasets.

Statistical Test for Saliency Maps of Graph Neural Networks via Selective Inference

arXiv:2505.16893v2 Announce Type: replace Abstract: Graph Neural Networks (GNNs) have gained prominence for their ability to process graph-structured data across various domains. However, interpreting GNN decisions remains a significant challenge, leading to the adoption of saliency maps for identifying salient subgraphs composed of influential nodes and edges. Despite their utility, the reliability of GNN saliency maps has been questioned, particularly in terms of their robustness to input noise. In this study, we propose a statistical testing framework to rigorously evaluate the significance of saliency maps. Our main contribution lies in addressing the inflation of the Type I error rate caused by double-dipping of data, leveraging the framework of Selective Inference. Our method provides statistically valid $p$-values while controlling the Type I error rate, ensuring that identified salient subgraphs contain meaningful information rather than random artifacts. The method is applicable to a variety of saliency methods with piecewise linearity (e.g., Class Activation Mapping). We validate our method on synthetic and real-world datasets, demonstrating its capability in assessing the reliability of GNN interpretations.

Fast kernel methods: Sobolev, physics-informed, and additive models

arXiv:2509.02649v1 Announce Type: new Abstract: Kernel methods are powerful tools in statistical learning, but their cubic complexity in the sample size n limits their use on large-scale datasets. In this work, we introduce a scalable framework for kernel regression with O(n log n) complexity, fully leveraging GPU acceleration. The approach is based on a Fourier representation of kernels combined with non-uniform fast Fourier transforms (NUFFT), enabling exact, fast, and memory-efficient computations. We instantiate our framework in three settings: Sobolev kernel regression, physics-informed regression, and additive models. When known, the proposed estimators are shown to achieve minimax convergence rates, consistent with classical kernel theory. Empirical results demonstrate that our methods can process up to tens of billions of samples within minutes, providing both statistical accuracy and computational scalability. These contributions establish a flexible approach, paving the way for the routine application of kernel methods in large-scale learning tasks.

Principled model selection for stochastic dynamics

arXiv:2501.10339v3 Announce Type: replace-cross Abstract: Complex dynamical systems, from macromolecules to ecosystems, are often modeled by stochastic differential equations. To learn such models from data, a common approach involves sparse selection among a large function library. However, we show that overfitting arises not just from individual model complexity, but also from the combinatorial growth of possible models. To address this, we introduce Parsimonious Stochastic Inference (PASTIS), a principled method combining likelihood-estimation statistics with extreme value theory to suppress superfluous parameters. PASTIS outperforms existing methods and reliably identifies minimal models, even with low sampling rates or measurement error. It extends to stochastic partial differential equations, and applies to ecological networks and reaction-diffusion dynamics.

The case for and against fixed step-size: Stochastic approximation algorithms in optimization and machine learning

arXiv:2309.02944v3 Announce Type: replace-cross Abstract: Theory and application of stochastic approximation (SA) have become increasingly relevant due in part to applications in optimization and reinforcement learning. This paper takes a new look at SA with constant step-size $alpha>0$, defined by the recursion, $$theta_{n+1} = theta_{n}+ alpha f(theta_n,Phi_{n+1})$$ in which $theta_ninmathbb{R}^d$ and ${Phi_{n}}$ is a Markov chain. The goal is to approximately solve root finding problem $bar{f}(theta^*)=0$, where $bar{f}(theta)=mathbb{E}[f(theta,Phi)]$ and $Phi$ has the steady-state distribution of ${Phi_{n}}$. The following conclusions are obtained under an ergodicity assumption on the Markov chain, compatible assumptions on $f$, and for $alpha>0$ sufficiently small: $textbf{1.}$ The pair process ${(theta_n,Phi_n)}$ is geometrically ergodic in a topological sense. $textbf{2.}$ For every $1le ple 4$, there is a constant $b_p$ such that $limsup_{ntoinfty}mathbb{E}[|theta_n-theta^*|^p]le b_p alpha^{p/2}$ for each initial condition. $textbf{3.}$ The Polyak-Ruppert-style averaged estimates $theta^{text{PR}}_n=n^{-1}sum_{k=1}^{n}theta_k$ converge to a limit $theta^{text{PR}}_infty$ almost surely and in mean square, which satisfies $theta^{text{PR}}_infty=theta^*+alpha bar{Upsilon}^*+O(alpha^2)$ for an identified non-random $bar{Upsilon}^*inmathbb{R}^d$. Moreover, the covariance is approximately optimal: The limiting covariance matrix of $theta{text PR}_n$ is approximately minimal in a matricial sense. The two main take-aways for practitioners are application-dependent. It is argued that, in applications to optimization, constant gain algorithms may be preferable even when the objective has multiple local minima; while a vanishing gain algorithm is preferable in applications to reinforcement learning due to the presence of bias.

Towards Digital Twins for Optimal Radioembolization

arXiv:2509.02607v1 Announce Type: cross Abstract: Radioembolization is a localized liver cancer treatment that delivers radioactive microspheres (30 micron) to tumors via a catheter inserted in the hepatic arterial tree. The goal is to maximize therapeutic efficacy while minimizing damage to healthy liver tissue. However, optimization is challenging due to complex hepatic artery anatomy, variable blood flow, and uncertainty in microsphere transport. The creation of dynamic, patient-specific digital twins may provide a transformative solution to these challenges. This work outlines a framework for a liver radioembolization digital twin using high-fidelity computational fluid dynamics (CFD) and/or recent physics-informed machine learning approaches. The CFD approach involves microsphere transport calculations in the hepatic arterial tree with individual patient data, which enables personalized treatment planning. Although accurate, traditional CFD is computationally expensive and limits clinical applicability. To accelerate simulations, physics-informed neural networks (PINNs) and their generative extensions play an increasingly important role. PINNs integrate governing equations, such as the Navier-Stokes equations, directly into the neural network training process, enabling mesh-free, data-efficient approximation of blood flow and microsphere transport. Physics-informed generative adversarial networks (PI-GANs), diffusion models (PI-DMs), and transformer-based architectures further enable uncertainty-aware, temporally resolved predictions with reduced computational cost. These AI surrogates not only maintain physical fidelity but also support rapid sampling of diverse flow scenarios, facilitating real-time decision support. Together, CFD and physics-informed AI methods form the foundation of dynamic, patient-specific digital twin to optimize radioembolization planning and ultimately improve clinical outcomes.