arXiv subjects

Ronen Talmon

Publications and source records attributed to Ronen Talmon.

At least 19 recordsLinked to original sources

arXiv2026

Spatial Power Estimation via Riemannian Covariance Matching

We propose a new method for spatial power spectrum estimation in array processing that leverages the Riemannian geometry of Hermitian positive definite (HPD) matrices. We show that conventional approaches minimize variants of the Euclidean distance between the sample covariance matrix and a model covariance matrix, without considering the fact that covariance matrices lie on the Riemannian manifold of HPD matrices. By exploiting this manifold, we present a Riemannian-aware covariance matching algorithm, termed SERCOM, using the Jensen-Bregman LogDet (JBLD) divergence, which, unlike other Riemannian distances, can be evaluated efficiently without eigen-decomposition. We theoretically compare the JBLD divergence to other Euclidean- and Riemannian-based distances, demonstrating robustness to spectral distortions. Experimental results demonstrate that SERCOM consistently outperforms existing methods in direction-of-arrival (DOA) and power estimation, particularly in challenging scenarios with low SNR, limited number of snapshots, and correlated sources.

eess.SP↗

arXiv2026

Complex Interpolation of Matrices with an application to Multi-Manifold Learning

Given two symmetric positive-definite matrices $A, B \in \mathbb{R}^{n \times n}$, we study the spectral properties of the interpolation $A^{1-x} B^x$ for $0 \leq x \leq 1$. The presence of `common structures' in $A$ and $B$, eigenvectors pointing in a similar direction, can be investigated using this interpolation perspective. Generically, exact log-linearity of the operator norm $\|A^{1-x} B^x\|$ is equivalent to the existence of a shared eigenvector in the original matrices; stability bounds show that approximate log-linearity forces principal singular vectors to align with leading eigenvectors of both matrices. These results give rise to and provide theoretical justification for a multi-manifold learning framework that identifies common and distinct latent structures in multiview data.

cs.LG↗

arXiv2026

Geometric Mode-Selection Scores for Delay-Coordinates Dynamic Mode Decomposition

Delay-coordinates dynamic mode decomposition (DC-DMD) is widely used to extract coherent spatiotemporal modes from high-dimensional time series. A central challenge is distinguishing dynamically meaningful modes from spurious modes induced by noise and order overestimation. We frame this as a mode-selection scoring problem: each mode receives a score that ranks it as true or spurious; any hard selection (threshold or clustering) is a downstream choice. We show that mode selection in DC-DMD is fundamentally a problem of subspace geometry. True modes are characterized by concentration within a low-dimensional signal subspace, whereas spurious modes tend to retain non-negligible components outside any moderate overestimate of that subspace. This geometric distinction defines true and spurious modes and motivates fully data-driven robust scoring criteria. The framework yields two complementary scores. The first uses a data-driven proxy of the signal subspace to compute a residual. The second comes from a new operator-theoretic analysis of delay embedding: using a block-companion formulation, we show that all modes exhibit a Kronecker-Vandermonde (KV) structure, with true modes distinguished by the degree of conformity to it. This deviation is governed by the geometric residual. Our analysis further explains the empirical behavior of magnitude- and norm-based heuristics and clarifies when and why they fail under delay coordinates. Numerical experiments, evaluated by precision-recall AUC of true-vs-spurious ranking, show that the proposed scores outperform the tested baselines across most of the small-spatial-dimension regime.

eess.SP↗

arXiv2025

Unsupervised Machine Learning for Experimental Detection of Quantum-Many-Body Phase Transitions

Quantum many-body (QMB) systems are generally computationally hard: the computing resources necessary to simulate them exactly can often exceed the existing computation resources by orders of magnitude. For this reason, Richard Feynman proposed the concept of a quantum simulator: quantum systems engineered to obey a prescribed evolution equation and repeating the experiment multiple times. Experimentally, however, as we explain below, the vast majority of observables describing the system are inaccessible. Thus, while Feynman's idea addresses the problem of simulating quantum dynamics, it leaves unsolved the equally fundamental problem of inferring the underlying physics from the limited observables accessible in experiments. Indeed, many complex phenomena associated with QMB systems remain elusive. Perhaps, the most important example is identifying phase transitions in QMB systems when no simple order-parameter exists, which poses major challenges to this day. Complicating the problem further is the fact that, in most cases, it is impossible to learn from numerical simulations, as the underlying systems are often too large to be computable, and small QMB can show strong finite size effects, masking the presence of the transition. Here, we present an unsupervised machine learning approach to study QMB experiments, specifically aimed at detecting phase transitions and crossovers directly from raw experimental measurements. We demonstrate our methodology on systems undergoing Many-Body Localization cross-over and Mott-to-Superfluid phase-transition, showing that it reveals collective phenomena from the very partial experimental data and without any model-specific prior knowledge of the system. This approach offers a general and scalable route for data-driven discovery of emergent phenomena in complex quantum many-body systems.

quant-ph↗

arXiv2025

Sparsity-Driven Entanglement Detection in High-Dimensional Quantum States

The characterization of high-dimensional quantum entanglement is crucial for advanced quantum computing and quantum information algorithms. Traditional methods require extensive data acquisition and suffer from limited visibility due to experimental noise. Here, we introduce a sparsity-driven framework to enhance the detection and certification of high-dimensional entanglement in spatially entangled photon pairs. By applying $\ell_1$-regularized reconstruction to sample covariance matrices obtained from measurements on photons produced via spontaneous parametric down-conversion (SPDC) measurements, we enhance the visibility of the correlation signal while suppressing noise. We demonstrate, using a position-momentum Einstein-Podolsky-Rosen (EPR) entanglement criterion, that this approach enables certification of an entanglement dimensionality that cannot be achieved without regularization. Our method is scalable, simple to use and compatible with existing quantum-optics platforms, thus paves the way for efficient, real-time analysis of high-dimensional quantum states.

quant-ph↗

arXiv2025

Enhancing VICReg: Random-Walk Pairing for Improved Generalization and Better Global Semantics Capturing

In this paper, we argue that viewing VICReg-a popular self-supervised learning (SSL) method--through the lens of spectral embedding reveals a potential source of sub-optimality: it may struggle to generalize robustly to unseen data due to overreliance on the training data. This observation invites a closer look at how well this method achieves its goal of producing meaningful representations of images outside of the training set as well. Here, we investigate this issue and introduce SAG-VICReg (Stable and Generalizable VICReg), a method that builds on VICReg by incorporating new training techniques. These enhancements improve the model's ability to capture global semantics within the data and strengthen the generalization capabilities. Experiments demonstrate that SAG-VICReg effectively addresses the generalization challenge while matching or surpassing diverse state-of-the-art SSL baselines. Notably, our method exhibits superior performance on metrics designed to evaluate global semantic understanding, while simultaneously maintaining competitive results on local evaluation metrics. Furthermore, we propose a new standalone evaluation metric for embeddings that complements the standard evaluation methods and accounts for the global data structure without requiring labels--a key issue when tagged data is scarce or not available.

cs.CV↗

arXiv2025

Learning Shared Representations from Unpaired Data

Learning shared representations is a primary area of multimodal representation learning. The current approaches to achieve a shared embedding space rely heavily on paired samples from each modality, which are significantly harder to obtain than unpaired ones. In this work, we demonstrate that shared representations can be learned almost exclusively from unpaired data. Our arguments are grounded in the spectral embeddings of the random walk matrices constructed independently from each unimodal representation. Empirical results in computer vision and natural language processing domains support its potential, revealing the effectiveness of unpaired data in capturing meaningful cross-modal relations, demonstrating high capabilities in retrieval tasks, generation, arithmetics, zero-shot, and cross-domain classification. This work, to the best of our knowledge, is the first to demonstrate these capabilities almost exclusively from unpaired samples, giving rise to a cross-modal embedding that could be viewed as universal, i.e., independent of the specific modalities of the data. Our project page: https://shaham-lab.github.io/SUE_page.

cs.CV↗

arXiv2025

It Takes a Graph to Know a Graph: Rewiring for Homophily with a Reference Graph

Graph Neural Networks (GNNs) excel at analyzing graph-structured data but struggle on heterophilic graphs, where connected nodes often belong to different classes. While this challenge is commonly addressed with specialized GNN architectures, graph rewiring remains an underexplored strategy in this context. We provide theoretical foundations linking edge homophily, GNN embedding smoothness, and node classification performance, motivating the need to enhance homophily. Building on this insight, we introduce a rewiring framework that increases graph homophily using a reference graph, with theoretical guarantees on the homophily of the rewired graph. To broaden applicability, we propose a label-driven diffusion approach for constructing a homophilic reference graph from node features and training labels. Through extensive simulations, we analyze how the homophily of both the original and reference graphs influences the rewired graph homophily and downstream GNN performance. We evaluate our method on 11 real-world heterophilic datasets and show that it outperforms existing rewiring techniques and specialized GNNs for heterophilic graphs, achieving improved node classification accuracy while remaining efficient and scalable to large graphs.

cs.LG↗

arXiv2025

Finsler Multi-Dimensional Scaling: Manifold Learning for Asymmetric Dimensionality Reduction and Embedding

Dimensionality reduction is a fundamental task that aims to simplify complex data by reducing its feature dimensionality while preserving essential patterns, with core applications in data analysis and visualisation. To preserve the underlying data structure, multi-dimensional scaling (MDS) methods focus on preserving pairwise dissimilarities, such as distances. They optimise the embedding to have pairwise distances as close as possible to the data dissimilarities. However, the current standard is limited to embedding data in Riemannian manifolds. Motivated by the lack of asymmetry in the Riemannian metric of the embedding space, this paper extends the MDS problem to a natural asymmetric generalisation of Riemannian manifolds called Finsler manifolds. Inspired by Euclidean space, we define a canonical Finsler space for embedding asymmetric data. Due to its simplicity with respect to geodesics, data representation in this space is both intuitive and simple to analyse. We demonstrate that our generalisation benefits from the same theoretical convergence guarantees. We reveal the effectiveness of our Finsler embedding across various types of non-symmetric data, highlighting its value in applications such as data visualisation, dimensionality reduction, directed graph embedding, and link prediction.

cs.CV↗

arXiv2025

Structure-Aware Matrix Pencil Method

We address the problem of detecting the number of complex exponentials and estimating their parameters from a noisy signal using the Matrix Pencil (MP) method. We introduce the MP modes and present their informative spectral structure. We show theoretically that these modes can be divided into signal and noise modes, where the signal modes exhibit a perturbed Vandermonde structure. Leveraging this structure, we propose a new MP algorithm, termed the SAMP algorithm, which has two novel components: (i) a robust, theoretically grounded model-order detection method, and (ii) an efficient estimation of the signal amplitudes. We show empirically that SAMP significantly outperforms the standard MP method in challenging conditions, with closely-spaced frequencies and low Signal-to-Noise Ratio (SNR) values, approaching the Cramer-Rao lower bound (CRB) for a broad SNR range. Additionally, compared with prevalent information-based criteria, we show that SAMP achieves comparable or better results with lower computational time, and lower sensitivity to noise model mismatch.

eess.SP↗

arXiv2025

Joint Hierarchical Representation Learning of Samples and Features via Informed Tree-Wasserstein Distance

High-dimensional data often exhibit hierarchical structures in both modes: samples and features. Yet, most existing approaches for hierarchical representation learning consider only one mode at a time. In this work, we propose an unsupervised method for jointly learning hierarchical representations of samples and features via Tree-Wasserstein Distance (TWD). Our method alternates between the two data modes. It first constructs a tree for one mode, then computes a TWD for the other mode based on that tree, and finally uses the resulting TWD to build the second mode's tree. By repeatedly alternating through these steps, the method gradually refines both trees and the corresponding TWDs, capturing meaningful hierarchical representations of the data. We provide a theoretical analysis showing that our method converges. We show that our method can be integrated into hyperbolic graph convolutional networks as a pre-processing technique, improving performance in link prediction and node classification tasks. In addition, our method outperforms baselines in sparse approximation and unsupervised Wasserstein distance learning tasks on word-document and single-cell RNA-sequencing datasets.

cs.LG↗

arXiv2024

Tree-Wasserstein Distance for High Dimensional Data with a Latent Feature Hierarchy

Finding meaningful distances between high-dimensional data samples is an important scientific task. To this end, we propose a new tree-Wasserstein distance (TWD) for high-dimensional data with two key aspects. First, our TWD is specifically designed for data with a latent feature hierarchy, i.e., the features lie in a hierarchical space, in contrast to the usual focus on embedding samples in hyperbolic space. Second, while the conventional use of TWD is to speed up the computation of the Wasserstein distance, we use its inherent tree as a means to learn the latent feature hierarchy. The key idea of our method is to embed the features into a multi-scale hyperbolic space using diffusion geometry and then present a new tree decoding method by establishing analogies between the hyperbolic embedding and trees. We show that our TWD computed based on data observations provably recovers the TWD defined with the latent feature hierarchy and that its computation is efficient and scalable. We showcase the usefulness of the proposed TWD in applications to word-document and single-cell RNA-sequencing datasets, demonstrating its advantages over existing TWDs and methods based on pre-trained models.

cs.LG↗

arXiv2024

Domain Adaptation for DoA Estimation in Multipath Channels with Interferences

We consider the problem of estimating the direction-of-arrival (DoA) of a desired source located in a known region of interest in the presence of interfering sources and multipath. We propose an approach that precedes the DoA estimation and relies on generating a set of reference steering vectors. The steering vectors' generative model is a free space model, which is beneficial for many DoA estimation algorithms. The set of reference steering vectors is then used to compute a function that maps the received signals from the adverse environment to a reference domain free from interfering sources and multipath. We show theoretically and empirically that the proposed map, which is analogous to domain adaption, improves DoA estimation by mitigating interference and multipath effects. Specifically, we demonstrate a substantial improvement in accuracy when the proposed approach is applied before three commonly used beamformers: the delay-and-sum (DS), the minimum variance distortionless response (MVDR), and the Multiple Signal Classification (MUSIC).

eess.SP↗

arXiv2024

On Learning what to Learn: heterogeneous observations of dynamics and establishing (possibly causal) relations among them

Before we attempt to learn a function between two (sets of) observables of a physical process, we must first decide what the inputs and what the outputs of the desired function are going to be. Here we demonstrate two distinct, data-driven ways of initially deciding ``the right quantities'' to relate through such a function, and then proceed to learn it. This is accomplished by processing multiple simultaneous heterogeneous data streams (ensembles of time series) from observations of a physical system: multiple observation processes of the system. We thus determine (a) what subsets of observables are common between the observation processes (and therefore observable from each other, relatable through a function); and (b) what information is unrelated to these common observables, and therefore particular to each observation process, and not contributing to the desired function. Any data-driven function approximation technique can subsequently be used to learn the input-output relation, from k-nearest neighbors and Geometric Harmonics to Gaussian Processes and Neural Networks. Two particular ``twists'' of the approach are discussed. The first has to do with the identifiability of particular quantities of interest from the measurements. We now construct mappings from a single set of observations of one process to entire level sets of measurements of the process, consistent with this single set. The second attempts to relate our framework to a form of causality: if one of the observation processes measures ``now'', while the second observation process measures ``in the future'', the function to be learned among what is common across observation processes constitutes a dynamical model for the system evolution.

cs.LG↗

arXiv2024

Equivariant Machine Learning on Graphs with Nonlinear Spectral Filters

Equivariant machine learning is an approach for designing deep learning models that respect the symmetries of the problem, with the aim of reducing model complexity and improving generalization. In this paper, we focus on an extension of shift equivariance, which is the basis of convolution networks on images, to general graphs. Unlike images, graphs do not have a natural notion of domain translation. Therefore, we consider the graph functional shifts as the symmetry group: the unitary operators that commute with the graph shift operator. Notably, such symmetries operate in the signal space rather than directly in the spatial space. We remark that each linear filter layer of a standard spectral graph neural network (GNN) commutes with graph functional shifts, but the activation function breaks this symmetry. Instead, we propose nonlinear spectral filters (NLSFs) that are fully equivariant to graph functional shifts and show that they have universal approximation properties. The proposed NLSFs are based on a new form of spectral domain that is transferable between graphs. We demonstrate the superior performance of NLSFs over existing spectral GNNs in node and graph classification benchmarks.

cs.LG↗

arXiv2024

Landmark Alternating Diffusion

Alternating Diffusion (AD) is a commonly applied diffusion-based sensor fusion algorithm. While it has been successfully applied to various problems, its computational burden remains a limitation. Inspired by the landmark diffusion idea considered in the Robust and Scalable Embedding via Landmark Diffusion (ROSELAND), we propose a variation of AD, called Landmark AD (LAD), which captures the essence of AD while offering superior computational efficiency. We provide a series of theoretical analyses of LAD under the manifold setup and apply it to the automatic sleep stage annotation problem with two electroencephalogram channels to demonstrate its application.

cs.LG↗

arXiv2024

Riemannian Covariance Fitting for Direction-of-Arrival Estimation

Covariance fitting (CF) is a comprehensive approach for direction of arrival (DoA) estimation, consolidating many common solutions. Standard practice is to use Euclidean criteria for CF, disregarding the intrinsic Hermitian positive-definite (HPD) geometry of the spatial covariance matrices. We assert that this oversight leads to inherent limitations. In this paper, as a remedy, we present a comprehensive study of the use of various Riemannian metrics of HPD matrices in CF. We focus on the advantages of the Affine-Invariant (AI) and the Log-Euclidean (LE) Riemannian metrics. Consequently, we propose a new practical beamformer based on the LE metric and derive analytically its spatial characteristics, such as the beamwidth and sidelobe attenuation, under noisy conditions. Comparing these features to classical beamformers shows significant advantage. In addition, we demonstrate, both theoretically and experimentally, the LE beamformer's robustness in scenarios with small sample sizes and in the presence of noise, interference, and multipath channels.

eess.SP↗

arXiv2024

The Expected Loss of Preconditioned Langevin Dynamics Reveals the Hessian Rank

Langevin dynamics (LD) is widely used for sampling from distributions and for optimization. In this work, we derive a closed-form expression for the expected loss of preconditioned LD near stationary points of the objective function. We use the fact that at the vicinity of such points, LD reduces to an Ornstein-Uhlenbeck process, which is amenable to convenient mathematical treatment. Our analysis reveals that when the preconditioning matrix satisfies a particular relation with respect to the noise covariance, LD's expected loss becomes proportional to the rank of the objective's Hessian. We illustrate the applicability of this result in the context of neural networks, where the Hessian rank has been shown to capture the complexity of the predictor function but is usually computationally hard to probe. Finally, we use our analysis to compare SGD-like and Adam-like preconditioners and identify the regimes under which each of them leads to a lower expected loss.

cs.LG↗