搜索结果：sure

In this paper we study the range of possible almost sure dimensions of random measures arising from a natural model of random Moran measures. Specifically, we consider the Assouad-like ``large'' $Φ$-dimensions of these measures. These dimensions can be tuned to consider a specific range of depths in scale and so provide refined local geometric information. The quasi-Assouad dimension is a well-known and important example of a ``large'' $Φ$-dimension. We determine the range of possible almost sure $Φ$-dimensions for random measures generated by the model and supported on any given random Moran set. We do this for both the case when the probability weights depend on the scaling factors and the case when they do not. In the later situation, we show that usually there is a ``gap'' between the dimension of the set and that of the smallest attainable upper dimension and largest attainable lower dimension. As a consequence of our results, we also determine the a.s. dimensions of the underlying random Moran set.

Robotic tasks involving contact interactions pose significant challenges for trajectory optimization due to discontinuous dynamics. Conventional formulations typically assume deterministic contact events, which limit robustness and adaptability in real-world settings. In this work, we propose SURE, a robust trajectory optimization framework that explicitly accounts for contact timing uncertainty. By allowing multiple trajectories to branch from possible pre-impact states and later rejoin a shared trajectory, SURE achieves both robustness and computational efficiency within a unified optimization framework. We evaluate SURE on two representative tasks with unknown impact times. In a cart-pole balancing task involving uncertain wall location, SURE achieves an average improvement of 21.6% in success rate when branch switching is enabled during control. In an egg-catching experiment using a robotic manipulator, SURE improves the success rate by 40%. These results demonstrate that SURE substantially enhances robustness compared to conventional nominal formulations.

We derive the asymptotic risk function of regularized empirical risk minimization (ERM) estimators tuned by $n$-fold cross-validation (CV). The out-of-sample prediction loss of such estimators converges in distribution to the squared-error loss (risk function) of shrinkage estimators in the normal means model, tuned by Stein's unbiased risk estimate (SURE). This risk function provides a more fine-grained picture of predictive performance than uniform bounds on worst-case regret, which are common in learning theory: it quantifies how risk varies with the true parameter. As key intermediate steps, we show that (i) $n$-fold CV converges uniformly to SURE, and (ii) while SURE typically has multiple local minima, its global minimum is generically well separated. Well-separation ensures that uniform convergence of CV to SURE translates into convergence of the tuning parameter chosen by CV to that chosen by SURE.

We consider turn-based stochastic two-player games with a combination of a parity condition that must hold surely, that is in all possible outcomes, and of a parity condition that must hold almost-surely, that is with probability 1. The problem of deciding the existence of a winning strategy in such games is central in the framework of synthesis beyond worst-case where a hard requirement that must hold surely is combined with a softer requirement. Recent works showed that the problem is coNP-complete, and infinite-memory strategies are necessary in general, even in one-player games (i.e., Markov decision processes). However, memoryless strategies are sufficient for the opponent player. Despite these comprehensive results, the known algorithmic solution enumerates all memoryless strategies of the opponent, which is exponential in all cases, and does not construct a winning strategy when one exists. We present a recursive algorithm, based on a characterisation of the winning region, that gives a deeper insight into the problem. In particular, we show how to construct a winning strategy to achieve the combination of sure and almost-sure parity, and we derive new complexity and memory

We provide certificates for almost sure reachability of continuous-time stochastic systems governed by stochastic differential equations (SDEs). We first show that a standard Euler-Maruyama discretization may fail to preserve almost sure reachability property of the system using a double-well Langevin system. This observation motivates us to develop certificates for almost sure reachability directly on the continuous-time system. We introduce a pair of certificates, a drift function and a variant function, and prove necessity and sufficiency for almost sure reachability of an open bounded target set. Using these certificates, for linear SDEs, we give a characterization of almost sure reachability in terms of the spectral structure of the system matrices. For polynomial SDEs, we fix a polynomial template for the drift function and choose the variant function template as an exponential function composed with a polynomial. This allows us to translate the conditions in the certificates into sum-of-squares (SOS) constraints. We then propose an alternating scheme to resolve bilinearities. We illustrate the approach on the double-well Langevin example, showing that continuous-time SOS cer

Almost sure reachability refers to the property of a stochastic system whereby, from any initial condition, the system state reaches a given target set with probability one. In this paper, we study the problem of certifying almost sure reachability in discrete-time stochastic systems using drift and variant conditions. While these conditions are both necessary and sufficient in theory, computational approaches often rely on restricting the search to fixed templates, such as polynomial or quadratic functions. We show that this restriction compromises completeness: there exists a polynomial system for which a given target set is almost surely reachable but admits no polynomial certificate, and a linear system for which a neighborhood of the origin is almost surely reachable but admits no quadratic certificate. We then provide a complete characterization of reachability certificates for linear systems with additive noise. Our analysis yields conditions on the system matrices under which valid certificates exist, and shows how the structure and dimension of the system determine the need for non-quadratic templates. Our results generalize the classical random walk behavior to a broader

The stochastic three points (STP) algorithm is a derivative-free optimization technique designed for unconstrained optimization problems in $\mathbb{R}^d$. In this paper, we analyze this algorithm for three classes of functions: smooth functions that may lack convexity, smooth convex functions, and smooth functions that are strongly convex. Our work provides the first almost sure convergence results of the STP algorithm, alongside some convergence results in expectation. For the class of smooth functions, we establish that the best gradient iterate of the STP algorithm converges almost surely to zero at a rate of $o(1/{T^{\frac{1}{2}-ε}})$ for any $ε\in (0,\frac{1}{2})$, where $T$ is the number of iterations. Furthermore, within the same class of functions, we establish both almost sure convergence and convergence in expectation of the final gradient iterate towards zero. For the class of smooth convex functions, we establish that $f(θ^T)$ converges to $\inf_{θ\in \mathbb{R}^d} f(θ)$ almost surely at a rate of $o(1/{T^{1-ε}})$ for any $ε\in (0,1)$, and in expectation at a rate of $O(\frac{d}{T})$ where $d$ is the dimension of the space. Finally, for the class of smooth functions th

Large language models (LLMs) have made significant advancements in various natural language processing tasks, including question answering (QA) tasks. While incorporating new information with the retrieval of relevant passages is a promising way to improve QA with LLMs, the existing methods often require additional fine-tuning which becomes infeasible with recent LLMs. Augmenting retrieved passages via prompting has the potential to address this limitation, but this direction has been limitedly explored. To this end, we design a simple yet effective framework to enhance open-domain QA (ODQA) with LLMs, based on the summarized retrieval (SuRe). SuRe helps LLMs predict more accurate answers for a given question, which are well-supported by the summarized retrieval that could be viewed as an explicit rationale extracted from the retrieved passages. Specifically, SuRe first constructs summaries of the retrieved passages for each of the multiple answer candidates. Then, SuRe confirms the most plausible answer from the candidate set by evaluating the validity and ranking of the generated summaries. Experimental results on diverse ODQA benchmarks demonstrate the superiority of SuRe, with

Combining the approaches made in works with Galeotti and Passmann, we define and study a notion of "almost sure" realizability with parameter-free ordinal Turing machines (OTMs). In particular, we show that, in contrast to the classical case, almost sure realizability differs from plain realizability, while closure under intuitionistic predicate logic and realizability of Kripke-Platek set theory continue to hold.

In this paper, we revisit techniques for uncertainty estimation within deep neural networks and consolidate a suite of techniques to enhance their reliability. Our investigation reveals that an integrated application of diverse techniques--spanning model regularization, classifier and optimization--substantially improves the accuracy of uncertainty predictions in image classification tasks. The synergistic effect of these techniques culminates in our novel SURE approach. We rigorously evaluate SURE against the benchmark of failure prediction, a critical testbed for uncertainty estimation efficacy. Our results showcase a consistently better performance than models that individually deploy each technique, across various datasets and model architectures. When applied to real-world challenges, such as data corruption, label noise, and long-tailed class distribution, SURE exhibits remarkable robustness, delivering results that are superior or on par with current state-of-the-art specialized methods. Particularly on Animal-10N and Food-101N for learning with noisy labels, SURE achieves state-of-the-art performance without any task-specific adjustments. This work not only sets a new bench

The vast majority of convergence rates analysis for stochastic gradient methods in the literature focus on convergence in expectation, whereas trajectory-wise almost sure convergence is clearly important to ensure that any instantiation of the stochastic algorithms would converge with probability one. Here we provide a unified almost sure convergence rates analysis for stochastic gradient descent (SGD), stochastic heavy-ball (SHB), and stochastic Nesterov's accelerated gradient (SNAG) methods. We show, for the first time, that the almost sure convergence rates obtained for these stochastic gradient methods on strongly convex functions, are arbitrarily close to their optimal convergence rates possible. For non-convex objective functions, we not only show that a weighted average of the squared gradient norms converges to zero almost surely, but also the last iterates of the algorithms. We further provide last-iterate almost sure convergence rates analysis for stochastic gradient methods on weakly convex smooth functions, in contrast with most existing results in the literature that only provide convergence in expectation for a weighted average of the iterates.

Stein's formula states that a random variable of the form $z^\top f(z) - \text{div} f(z)$ is mean-zero for functions $f$ with integrable gradient. Here, $\text{div} f$ is the divergence of the function $f$ and $z$ is a standard normal vector. This paper aims to propose a Second Order Stein formula to characterize the variance of such random variables for all functions $f(z)$ with square integrable gradient, and to demonstrate the usefulness of this formula in various applications. In the Gaussian sequence model, a consequence of Stein's formula is Stein's Unbiased Risk Estimate (SURE), an unbiased estimate of the mean squared risk for almost any estimator $\hatμ$ of the unknown mean. A first application of the Second Order Stein formula is an Unbiased Risk Estimate for SURE itself (SURE for SURE): an unbiased estimate {providing} information about the squared distance between SURE and the squared estimation error of $\hatμ$. SURE for SURE has a simple form as a function of the data and is applicable to all $\hatμ$ with square integrable gradient, e.g. the Lasso and the Elastic Net. In addition to SURE for SURE, the following applications are developed: (1) Upper bounds on the risk

Nearly all estimators in statistical prediction come with an associated tuning parameter, in one way or another. Common practice, given data, is to choose the tuning parameter value that minimizes a constructed estimate of the prediction error of the estimator; we focus on Stein's unbiased risk estimator, or SURE (Stein, 1981; Efron, 1986) which forms an unbiased estimate of the prediction error by augmenting the observed training error with an estimate of the degrees of freedom of the estimator. Parameter tuning via SURE minimization has been advocated by many authors, in a wide variety of problem settings, and in general, it is natural to ask: what is the prediction error of the SURE-tuned estimator? An obvious strategy would be simply use the apparent error estimate as reported by SURE, i.e., the value of the SURE criterion at its minimum, to estimate the prediction error of the SURE-tuned estimator. But this is no longer unbiased; in fact, we would expect that the minimum of the SURE criterion is systematically biased downwards for the true prediction error. In this paper, we formally describe and study this bias.

The goal of this work is to prove a new sure upper bound in a setting that can be thought of as a simplified function field analogue. This result is comparable to a recent result of the author concerning almost sure upper bound of random multiplicative functions. Having a simpler quantity allows us to make the proof more accessible.

We study the almost sure convergence of the normalized columns in an infinite product of nonnegative matrices, and the almost sure rank one property of its limit points. Given a probability on the set of $2\times2$ nonnegative matrices, with finite support $\mathcal A=\{A(0),\dots,A(s-1)\}$, and assuming that at least one of the $A(k)$ is not diagonal, the normalized columns of the product matrix $P_n=A(ω_1)\dots A(ω_n)$ converge almost surely (for the product probability) with an exponential rate of convergence if and only if the Lyapunov exponents are almost surely distinct. If this condition is satisfied, given a nonnegative column vector $V$ the column vector $\frac{P_nV}{\Vert P_nV\Vert}$ also converges almost surely with an exponential rate of convergence. On the other hand if we assume only that at least one of the $A(k)$ do not have the form $\begin{pmatrix}a&0\\0&d\end{pmatrix}$, $ad e0$, nor the form $\begin{pmatrix}0&b\\d&0\end{pmatrix}$, $bc e0$, the limit-points of the normalized product matrix $\frac{P_n}{\Vert P_n\Vert}$ have almost surely rank 1 -although the limits of the normalized columns can be distinct- and $\frac{P_nV}{\Vert P_nV\Vert}$ converg

Suppose $(f,\mathcal{X},ν)$ is a measure preserving dynamical system and $φ:\mathcal{X}\to\mathbb{R}$ is an observable with some degree of regularity. We investigate the maximum process $M_n:=\max\{X_1,\ldots,X_n\}$, where $X_i=φ\circ f^i$ is a time series of observations on the system. When $M_n\to\infty$ almost surely, we establish results on the almost sure growth rate, namely the existence (or otherwise) of a sequence $u_n\to\infty$ such that $M_n/u_n\to 1$ almost surely. The observables we consider will be functions of the distance to a distinguished point $\tilde{x}\in \mathcal{X}$. Our results are based on the interplay between shrinking target problem estimates at $\tilde{x}$ and the form of the observable (in particular polynomial or logarithmic) near $\tilde{x}$. We determine where such an almost sure limit exists and give examples where it does not. Our results apply to a wide class of non-uniformly hyperbolic dynamical systems, under mild assumptions on the rate of mixing, and on regularity of the invariant measure.

Stein's unbiased risk estimate (SURE) was proposed by Stein for the independent, identically distributed (iid) Gaussian model in order to derive estimates that dominate least-squares (LS). In recent years, the SURE criterion has been employed in a variety of denoising problems for choosing regularization parameters that minimize an estimate of the mean-squared error (MSE). However, its use has been limited to the iid case which precludes many important applications. In this paper we begin by deriving a SURE counterpart for general, not necessarily iid distributions from the exponential family. This enables extending the SURE design technique to a much broader class of problems. Based on this generalization we suggest a new method for choosing regularization parameters in penalized LS estimators. We then demonstrate its superior performance over the conventional generalized cross validation approach and the discrepancy method in the context of image deblurring and deconvolution. The SURE technique can also be used to design estimates without predefining their structure. However, allowing for too many free parameters impairs the performance of the resulting estimates. To address this

Consider $n$ independent and identically distributed $p$-dimensional Gaussian random vectors with covariance matrix $Σ.$ The problem of estimating $Σ$ when $p$ is much larger than $n$ has received a lot of attention in recent years. Yet little is known about the information criterion for covariance matrix estimation. How to properly define such a criterion and what are the statistical properties? We attempt to answer these questions in the present paper by focusing on the estimation of bandable covariance matrices when $p>n$ but $\log(p)=o(n)$. Motivated by the deep connection between Stein's unbiased risk estimation (SURE) and AIC in regression models, we propose a family of generalized SURE ($\text{SURE}_c$) indexed by $c$ for covariance matrix estimation, where $c$ is some constant. When $c$ is 2, $\text{SURE}_2$ provides an unbiased estimator of the Frobenious risk of the covariance matrix estimator. Furthermore, we show that by minimizing $\text{SURE}_2$ over all possible banding covariance matrix estimators we attain the minimax optimal rate of convergence and the resulting estimator behaves like the covariance matrix estimator obtained by the so-called oracle tuning. On t

In this paper, we develop tools to establish almost sure stability of stochastic switched systems whose switching signal is constrained by an automaton. After having provided the necessary generalizations of existing results in the setting of stochastic graphs, we provide a characterization of almost sure stability in terms of multiple Lyapunov functions. We introduce the concept of lifts, providing formal expansions of stochastic graphs, together with the guarantee of conserving the underlying probability framework. We show how these techniques, firstly introduced in the deterministic setting, provide hierarchical methods in order to compute tight upper bounds for the almost sure decay rate. The theoretical developments are finally illustrated via a numerical example.