We consider the sparse linear regression model $\mathbf{y} = X β+\mathbf{w}$, where $X \in \mathbb{R}^{n \times d}$ is the design, $β\in \mathbb{R}^{d}$ is a $k$-sparse secret, and $\mathbf{w} \sim N(0, I_n)$ is the noise. Given input $X$ and $\mathbf{y}$, the goal is to estimate $β$. In this setting, the Lasso estimate achieves prediction error $O(k \log d / γn)$, where $γ$ is the restricted eigenvalue (RE) constant of $X$ with respect to $\mathrm{support}(β)$. In this paper, we introduce a new $\textit{semirandom}$ family of designs -- which we call $\textit{partially-rotated}$ designs -- for which the RE constant with respect to the secret is bounded away from zero even when a subset of the design columns are arbitrarily correlated among themselves. As an example of such a design, suppose we start with some arbitrary $X$, and then apply a random rotation to the columns of $X$ indexed by $\mathrm{support}(β)$. Let $λ_{\min}$ be the smallest eigenvalue of $\frac{1}{n} X_{\mathrm{support}(β)}^\top X_{\mathrm{support}(β)}$, where $X_{\mathrm{support}(β)}$ is the restriction of $X$ to the columns indexed by $\mathrm{support}(β)$. In this setting, our results imply that Lasso achieves prediction error $O(k \log d / λ_{\min} n)$ with high probability. This prediction error bound is independent of the arbitrary columns of $X$ not indexed by $\mathrm{support}(β)$, and is as good as if all of these columns were perfectly well-conditioned. Technically, our proof reduces to showing that matrices with a certain deterministic property -- which we call $\textit{restricted normalized orthogonality}$ (RNO) -- lead to RE constants that are independent of a subset of the matrix columns. This property is similar but incomparable with the restricted orthogonality condition of [CT05].

The present article studies the minimization of convex, L-smooth functions defined on a separable real Hilbert space. We analyze regularized stochastic gradient descent (reg-SGD), a variant of stochastic gradient descent that uses a Tikhonov regularization with time-dependent, vanishing regularization parameter. We prove strong convergence of reg-SGD to the minimum-norm solution of the original problem without additional boundedness assumptions. Moreover, we quantify the rate of convergence and optimize the interplay between step-sizes and regularization decay. Our analysis reveals how vanishing Tikhonov regularization controls the flow of SGD and yields stable learning dynamics, offering new insights into the design of iterative algorithms for convex problems, including those that arise in ill-posed inverse problems. We validate our theoretical findings through numerical experiments on image reconstruction and ODE-based inverse problems.

This paper addresses the challenge of efficient principal component analysis (PCA) in high-dimensional spaces by analyzing a compressively sampled variant of Oja's algorithm with adaptive sensing. Traditional PCA methods incur substantial computational costs that scale poorly with data dimensionality, whereas subspace tracking algorithms like Oja's offer more efficient alternatives but typically require full-dimensional observations. We analyze a variant where, at each iteration, only two compressed measurements are taken: one in the direction of the current estimate and one in a random orthogonal direction. We prove that this adaptive sensing approach achieves global convergence in the presence of noise when tracking the leading eigenvector of a datastream with eigengap $Δ=λ_1-λ_2$. Our theoretical analysis demonstrates that the algorithm experiences two phases: (1) a warmup phase requiring $O(λ_1λ_2d^2/Δ^2)$ iterations to achieve a constant-level alignment with the true eigenvector, followed by (2) a local convergence phase where the sine alignment error decays at a rate of $O(λ_1λ_2d^2/Δ^2 t)$ for iterations $t$. The guarantee aligns with existing minimax lower bounds with an added factor of $d$ due to the compressive sampling. This work provides the first convergence guarantees in adaptive sensing for subspace tracking with noise. Our proof technique is also considerably simpler than those in prior works. The results have important implications for applications where acquiring full-dimensional samples is challenging or costly.

Averaging, or smoothing, is a fundamental approach to obtain stable, de-noised estimates from noisy observations. In certain scenarios, observations made along trajectories of random dynamical systems are of particular interest. One popular smoothing technique for such a scenario is exponential moving averaging (EMA), which assigns observations a weight that decreases exponentially in their age, thus giving younger observations a larger weight. However, EMA fails to enjoy strong stochastic convergence properties, which stems from the fact that the weight assigned to the youngest observation is constant over time, preventing the noise in the averaged quantity from decreasing to zero. In this work, we consider an adaptation to EMA, which we call $p$-EMA, where the weights assigned to the last observations decrease to zero at a subharmonic rate. We provide stochastic convergence guarantees for this kind of averaging under mild assumptions on the autocorrelations of the underlying random dynamical system. We further discuss the implications of our results for a recently introduced adaptive step size control for Stochastic Gradient Descent (SGD), which uses $p$-EMA for averaging noisy observations.

We study classification problems using binary estimators where the decision boundary is described by horizon functions and where the data distribution satisfies a geometric margin condition. We establish upper and lower bounds for the minimax learning rate over broad function classes with bounded Kolmogorov entropy in Lebesgue norms. A key novelty of our work is the derivation of lower bounds on the worst-case learning rates under a geometric margin condition -- a setting that is almost universally satisfied in practice but remains theoretically challenging. Moreover, our results deal with the noiseless setting, where lower bounds are particularly hard to establish. We apply our general results to classification problems with decision boundaries belonging to several function classes: for Barron-regular functions, and for Hölder-continuous functions with strong margins, we identify optimal rates close to the fast learning rates of $\mathcal{O}(n^{-1})$ for $n \in \mathbb{N}$ samples. Also for merely convex decision boundaries, in a strong margin case optimal rates near $\mathcal{O}(n^{-1/2})$ can be achieved.

Healthcare data frequently contain a substantial proportion of missing values, necessitating effective time series imputation to support downstream disease diagnosis tasks. However, existing imputation methods focus on discrete data points and are unable to effectively model sparse data, resulting in particularly poor performance for imputing substantial missing values. In this paper, we propose a novel approach, ImputeINR, for time series imputation by employing implicit neural representations (INR) to learn continuous functions for time series. ImputeINR leverages the merits of INR in that the continuous functions are not coupled to sampling frequency and have infinite sampling frequency, allowing ImputeINR to generate fine-grained imputations even on extremely sparse observed values. Extensive experiments conducted on eight datasets with five ratios of masked values show the superior imputation performance of ImputeINR, especially for high missing ratios in time series data. Furthermore, we validate that applying ImputeINR to impute missing values in healthcare data enhances the performance of downstream disease diagnosis tasks. Codes are available.

Peter Palensky, Simon H. Tindemans Electrical Sustainable Energy Delft University of T echnology Delft, Netherlands { P .Palensky, S.H.Tindemans}@tudelft.nl Abstract --As the number of rooftop photovoltaic (PV) installations increases, aggregators and system operators are required to monitor and analyze these systems, raising the challenge of integration and management of large, spatially distributed time-series data that are both high-dimensional and affected by missing values. In this work, a probabilistic entity embedding-based clustering framework is proposed to address these problems. Applied to a multi-year residential PV dataset, it produces concise, uncertainty-aware cluster profiles that outperform a physics-based baseline in representativeness and robustness, and support reliable missing-value imputation. A systematic hyperparameter study further offers practical guidance for balancing model performance and robustness. I NTRODUCTION Modern energy systems are undergoing a rapid transformation, increasingly driven by decentralized generation sources, especially rooftop photovoltaic (PV) systems installed across residential and commercial properties.

Long-term time series forecasting (LTSF) offers broad utility in practical settings like energy consumption and weather prediction. Accurately predicting long-term changes, however, is demanding due to the intricate temporal patterns and inherent multi-scale variations within time series. This work confronts key issues in LTSF, including the suboptimal use of multi-granularity information, the neglect of channel-specific attributes, and the unique nature of trend and seasonal components, by introducing a proficient MLP-based forecasting framework. Our method adeptly disentangles complex temporal dynamics using clear, concurrent predictions across various scales. These multi-scale forecasts are then skillfully integrated through a system that dynamically assigns importance to information from different granularities, sensitive to individual channel characteristics. To manage the specific features of temporal patterns, a two-pronged structure is utilized to model trend and seasonal elements independently. Experimental results on eight LTSF benchmarks demonstrate that MDMixer improves average MAE performance by 4.64% compared to the recent state-of-the-art MLP-based method (TimeMixer), while achieving an effective balance between training efficiency and model interpretability.

Continual learning (CL) aims to learn a sequence of tasks without forgetting prior knowledge, but gradient updates for a new task often overwrite the weights learned earlier, causing catastrophic forgetting (CF). We propose Prototype-Augmented Hypernetworks (PAH), a framework where a single hypernetwork, conditioned on learnable task prototypes, dynamically generates task-specific classifier heads on demand. To mitigate forgetting, PAH combines cross-entropy with dual distillation losses, one to align logits and another to align prototypes, ensuring stable feature representations across tasks. Evaluations on Split-CIFAR100 and TinyImageNet demonstrate that PAH achieves state-of-the-art performance, reaching 74.5 % and 63.7 % accuracy with only 1.7 % and 4.4 % forgetting, respectively, surpassing prior methods without storing samples or heads.

Understanding the decision-making process of machine learning models provides valuable insights into the task, the data, and the reasons behind a model's failures. In this work, we propose a method that performs inherently interpretable predictions through the instance-wise sparsification of input images. To align the sparsification with human perception, we learn the masking in the space of semantically meaningful pixel regions rather than on pixel-level. Additionally, we introduce an explicit way to dynamically determine the required level of sparsity for each instance. We show empirically on semi-synthetic and natural image datasets that our inherently interpretable classifier produces more meaningful, human-understandable predictions than state-of-the-art benchmarks.