arXiv subjects

Zvika Ben-Haim

Publications and source records attributed to Zvika Ben-Haim.

14 recordsLinked to original sources

arXiv2025

Leveraging LLM Inconsistency to Boost Pass@k Performance

Large language models (LLMs) achieve impressive abilities in numerous domains, but exhibit inconsistent performance in response to minor input changes. Rather than view this as a drawback, in this paper we introduce a novel method for leveraging models' inconsistency to boost Pass@k performance. Specifically, we present a "Variator" agent that generates k variants of a given task and submits one candidate solution for each one. Our variant generation approach is applicable to a wide range of domains as it is task agnostic and compatible with free-form inputs. We demonstrate the efficacy of our agent theoretically using a probabilistic model of the inconsistency effect, and show empirically that it outperforms the baseline on the APPS dataset. Furthermore, we establish that inconsistency persists even in frontier reasoning models across coding and cybersecurity domains, suggesting our method is likely to remain relevant for future model generations.

cs.LG↗

arXiv2021

Flood forecasting with machine learning models in an operational framework

The operational flood forecasting system by Google was developed to provide accurate real-time flood warnings to agencies and the public, with a focus on riverine floods in large, gauged rivers. It became operational in 2018 and has since expanded geographically. This forecasting system consists of four subsystems: data validation, stage forecasting, inundation modeling, and alert distribution. Machine learning is used for two of the subsystems. Stage forecasting is modeled with the Long Short-Term Memory (LSTM) networks and the Linear models. Flood inundation is computed with the Thresholding and the Manifold models, where the former computes inundation extent and the latter computes both inundation extent and depth. The Manifold model, presented here for the first time, provides a machine-learning alternative to hydraulic modeling of flood inundation. When evaluated on historical data, all models achieve sufficiently high-performance metrics for operational use. The LSTM showed higher skills than the Linear model, while the Thresholding and Manifold models achieved similar performance metrics for modeling inundation extent. During the 2021 monsoon season, the flood warning system was operational in India and Bangladesh, covering flood-prone regions around rivers with a total area of 287,000 km2, home to more than 350M people. More than 100M flood alerts were sent to affected populations, to relevant authorities, and to emergency organizations. Current and future work on the system includes extending coverage to additional flood-prone locations, as well as improving modeling capabilities and accuracy.

cs.LG↗

arXiv2021

Physics-Aware Downsampling with Deep Learning for Scalable Flood Modeling

Background: Floods are the most common natural disaster in the world, affecting the lives of hundreds of millions. Flood forecasting is therefore a vitally important endeavor, typically achieved using physical water flow simulations, which rely on accurate terrain elevation maps. However, such simulations, based on solving partial differential equations, are computationally prohibitive on a large scale. This scalability issue is commonly alleviated using a coarse grid representation of the elevation map, though this representation may distort crucial terrain details, leading to significant inaccuracies in the simulation. Contributions: We train a deep neural network to perform physics-informed downsampling of the terrain map: we optimize the coarse grid representation of the terrain maps, so that the flood prediction will match the fine grid solution. For the learning process to succeed, we configure a dataset specifically for this task. We demonstrate that with this method, it is possible to achieve a significant reduction in computational cost, while maintaining an accurate solution. A reference implementation accompanies the paper as well as documentation and code for dataset reproduction.

cs.LG↗

arXiv2019

Inundation Modeling in Data Scarce Regions

Flood forecasts are crucial for effective individual and governmental protective action. The vast majority of flood-related casualties occur in developing countries, where providing spatially accurate forecasts is a challenge due to scarcity of data and lack of funding. This paper describes an operational system providing flood extent forecast maps covering several flood-prone regions in India, with the goal of being sufficiently scalable and cost-efficient to facilitate the establishment of effective flood forecasting systems globally.

cs.LG↗

arXiv2013

Minimum Variance Estimation of a Sparse Vector within the Linear Gaussian Model: An RKHS Approach

We consider minimum variance estimation within the sparse linear Gaussian model (SLGM). A sparse vector is to be estimated from a linearly transformed version embedded in Gaussian noise. Our analysis is based on the theory of reproducing kernel Hilbert spaces (RKHS). After a characterization of the RKHS associated with the SLGM, we derive novel lower bounds on the minimum variance achievable by estimators with a prescribed bias function. This includes the important case of unbiased estimation. The variance bounds are obtained via an orthogonal projection of the prescribed mean function onto a subspace of the RKHS associated with the SLGM. Furthermore, we specialize our bounds to compressed sensing measurement matrices and express them in terms of the restricted isometry and coherence parameters. For the special case of the SLGM given by the sparse signal in noise model (SSNM), we derive closed-form expressions of the minimum achievable variance (Barankin bound) and the corresponding locally minimum variance estimator. We also analyze the effects of exact and approximate sparsity information and show that the minimum achievable variance for exact sparsity is not a limiting case of that for approximate sparsity. Finally, we compare our bounds with the variance of three well-known estimators, namely, the maximum-likelihood estimator, the hard-thresholding estimator, and compressive reconstruction using the orthogonal matching pursuit.

cs.IT↗

arXiv2010

A Lower Bound on the Estimator Variance for the Sparse Linear Model

We study the performance of estimators of a sparse nonrandom vector based on an observation which is linearly transformed and corrupted by additive white Gaussian noise. Using the reproducing kernel Hilbert space framework, we derive a new lower bound on the estimator variance for a given differentiable bias function (including the unbiased case) and an almost arbitrary transformation matrix (including the underdetermined case considered in compressed sensing theory). For the special case of a sparse vector corrupted by white Gaussian noise-i.e., without a linear transformation-and unbiased estimation, our lower bound improves on previously proposed bounds.

math.ST↗

arXiv2010

Performance Bounds and Design Criteria for Estimating Finite Rate of Innovation Signals

In this paper, we consider the problem of estimating finite rate of innovation (FRI) signals from noisy measurements, and specifically analyze the interaction between FRI techniques and the underlying sampling methods. We first obtain a fundamental limit on the estimation accuracy attainable regardless of the sampling method. Next, we provide a bound on the performance achievable using any specific sampling approach. Essential differences between the noisy and noise-free cases arise from this analysis. In particular, we identify settings in which noise-free recovery techniques deteriorate substantially under slight noise levels, thus quantifying the numerical instability inherent in such methods. This instability, which is only present in some families of FRI signals, is shown to be related to a specific type of structure, which can be characterized by viewing the signal model as a union of subspaces. Finally, we develop a methodology for choosing the optimal sampling kernels based on a generalization of the Karhunen--Loève transform. The results are illustrated for several types of time-delay estimation problems.

cs.IT↗

arXiv2010

Near-Oracle Performance of Greedy Block-Sparse Estimation Techniques from Noisy Measurements

This paper examines the ability of greedy algorithms to estimate a block sparse parameter vector from noisy measurements. In particular, block sparse versions of the orthogonal matching pursuit and thresholding algorithms are analyzed under both adversarial and Gaussian noise models. In the adversarial setting, it is shown that estimation accuracy comes within a constant factor of the noise power. Under Gaussian noise, the Cramer-Rao bound is derived, and it is shown that the greedy techniques come close to this bound at high SNR. The guarantees are numerically compared with the actual performance of block and non-block algorithms, highlighting the advantages of block sparse techniques.

cs.IT↗

arXiv2010

Unbiased Estimation of a Sparse Vector in White Gaussian Noise

We consider unbiased estimation of a sparse nonrandom vector corrupted by additive white Gaussian noise. We show that while there are infinitely many unbiased estimators for this problem, none of them has uniformly minimum variance. Therefore, we focus on locally minimum variance unbiased (LMVU) estimators. We derive simple closed-form lower and upper bounds on the variance of LMVU estimators or, equivalently, on the Barankin bound (BB). Our bounds allow an estimation of the threshold region separating the low-SNR and high-SNR regimes, and they indicate the asymptotic behavior of the BB at high SNR. We also develop numerical lower and upper bounds which are tighter than the closed-form bounds and thus characterize the BB more accurately. Numerical studies compare our characterization of the BB with established biased estimation schemes, and demonstrate that while unbiased estimators perform poorly at low SNR, they may perform better than biased estimators at high SNR. An interesting conclusion of our analysis is that the high-SNR behavior of the BB depends solely on the value of the smallest nonzero component of the sparse vector, and that this type of dependence is also exhibited by the performance of certain practical estimators.

math.ST↗

arXiv2010

On Unbiased Estimation of Sparse Vectors Corrupted by Gaussian Noise

We consider the estimation of a sparse parameter vector from measurements corrupted by white Gaussian noise. Our focus is on unbiased estimation as a setting under which the difficulty of the problem can be quantified analytically. We show that there are infinitely many unbiased estimators but none of them has uniformly minimum mean-squared error. We then provide lower and upper bounds on the Barankin bound, which describes the performance achievable by unbiased estimators. These bounds are used to predict the threshold region of practical estimators.

cs.IT↗

arXiv2009

Coherence-Based Performance Guarantees for Estimating a Sparse Vector Under Random Noise

We consider the problem of estimating a deterministic sparse vector x from underdetermined measurements Ax+w, where w represents white Gaussian noise and A is a given deterministic dictionary. We analyze the performance of three sparse estimation algorithms: basis pursuit denoising (BPDN), orthogonal matching pursuit (OMP), and thresholding. These algorithms are shown to achieve near-oracle performance with high probability, assuming that x is sufficiently sparse. Our results are non-asymptotic and are based only on the coherence of A, so that they are applicable to arbitrary dictionaries. Differences in the precise conditions required for the performance guarantees of each algorithm are manifested in the observed performance at high and low signal-to-noise ratios. This provides insight on the advantages and drawbacks of convex relaxation techniques such as BPDN as opposed to greedy approaches such as OMP and thresholding.

math.ST↗

arXiv2009

The Cramer-Rao Bound for Sparse Estimation

The goal of this paper is to characterize the best achievable performance for the problem of estimating an unknown parameter having a sparse representation. Specifically, we consider the setting in which a sparsely representable deterministic parameter vector is to be estimated from measurements corrupted by Gaussian noise, and derive a lower bound on the mean-squared error (MSE) achievable in this setting. To this end, an appropriate definition of bias in the sparse setting is developed, and the constrained Cramer-Rao bound (CRB) is obtained. This bound is shown to equal the CRB of an estimator with knowledge of the support set, for almost all feasible parameter values. Consequently, in the unbiased case, our bound is identical to the MSE of the oracle estimator. Combined with the fact that the CRB is achieved at high signal-to-noise ratios by the maximum likelihood technique, our result provides a new interpretation for the common practice of using the oracle estimator as a gold standard against which practical approaches are compared.

math.ST↗

arXiv2009

A Lower Bound on the Bayesian MSE Based on the Optimal Bias Function

A lower bound on the minimum mean-squared error (MSE) in a Bayesian estimation problem is proposed in this paper. This bound utilizes a well-known connection to the deterministic estimation setting. Using the prior distribution, the bias function which minimizes the Cramer-Rao bound can be determined, resulting in a lower bound on the Bayesian MSE. The bound is developed for the general case of a vector parameter with an arbitrary probability distribution, and is shown to be asymptotically tight in both the high and low signal-to-noise ratio regimes. A numerical study demonstrates several cases in which the proposed technique is both simpler to compute and tighter than alternative methods.

cs.IT↗

arXiv2007

Blind Minimax Estimation

We consider the linear regression problem of estimating an unknown, deterministic parameter vector based on measurements corrupted by colored Gaussian noise. We present and analyze blind minimax estimators (BMEs), which consist of a bounded parameter set minimax estimator, whose parameter set is itself estimated from measurements. Thus, one does not require any prior assumption or knowledge, and the proposed estimator can be applied to any linear regression problem. We demonstrate analytically that the BMEs strictly dominate the least-squares estimator, i.e., they achieve lower mean-squared error for any value of the parameter vector. Both Stein's estimator and its positive-part correction can be derived within the blind minimax framework. Furthermore, our approach can be readily extended to a wider class of estimation problems than Stein's estimator, which is defined only for white noise and non-transformed measurements. We show through simulations that the BMEs generally outperform previous extensions of Stein's technique.

math.ST↗