arXiv subjects

Kobi Cohen

Publications and source records attributed to Kobi Cohen.

At least 19 recordsLinked to original sources

arXiv2026

HiMu: Hierarchical Multimodal Frame Selection for Long Video Question Answering

Long-form video question answering requires reasoning over extended temporal contexts, making frame selection a critical bottleneck for multi-modal large language models (MLLMs) bound by finite context windows. Within the controlled frame-budget regime that governs practical deployment, prior selectors score frames against a single global query embedding; as a result, compositional multimodal questions that involve temporal ordering or cross-modal cues such as ``what happens on screen right after the narrator mentions the reaction?'' are flattened into a representation that loses sub-event ordering and modality bindings. We introduce \textbf{HiMu}, a training-free framework for compositional multimodal frame selection. A single text-only LLM call decomposes the query into a hierarchical logic tree whose leaves are atomic predicates, each routed to a lightweight expert spanning vision (CLIP, open-vocabulary detection, OCR) and audio (speech recognition and non-speech sound matching). Expert signals are normalized, smoothed to align across modalities, and composed bottom-up through fuzzy-logic operators that enforce temporal sequencing and adjacency, yielding a continuous per-frame satisfaction curve. Under the standard 16-frame budget on Video-MME, LongVideoBench, and HERBench-Lite, HiMu achieves state-of-the-art accuracy among frame selection methods and improves over uniform sampling across seven diverse MLLMs as a drop-in module, matching the accuracy of uniform sampling at $4\times$ its frame budget, without retraining and without multiple iterative MLLM calls during selection.

cs.CV↗

arXiv2026

DeepOFW: Deep Learning-Driven OFDM-Flexible Waveform Modulation for Peak-to-Average Power Ratio Reduction

Peak-to-average power ratio (PAPR) remains a major limitation of multicarrier modulation schemes such as orthogonal frequency-division multiplexing (OFDM), reducing power amplifier efficiency and limiting practical transmit power. In this work, we propose DeepOFW, a deep learning-driven OFDM-flexible waveform modulation framework that enables data-driven waveform design while preserving the low-complexity hardware structure of conventional transceivers. The proposed architecture is fully differentiable, allowing end-to-end optimization of waveform generation and receiver processing under practical physical constraints. Unlike neural transceiver approaches that require deep learning inference at both ends of the link, DeepOFW confines the learning stage to an offline or centralized unit, enabling deployment on standard transmitter and receiver hardware without additional computational overhead. The framework jointly optimizes waveform representations and detection parameters while explicitly incorporating PAPR constraints during training. Extensive simulations over 3GPP multipath channels demonstrate that the learned waveforms significantly reduce PAPR compared with classical OFDM while simultaneously improving bit error rate (BER) performance relative to state-of-the-art transmission schemes. These results highlight the potential of data-driven waveform design to enhance multicarrier communication systems while maintaining hardware-efficient implementations. An open-source implementation of the proposed framework is released to facilitate reproducible research and practical adoption.

cs.IT↗

arXiv2025

Distributed Learning in Markovian Restless Bandits over Interference Graphs for Stable Spectrum Sharing

We study distributed learning for spectrum access and sharing among multiple cognitive communication entities, such as cells, subnetworks, or cognitive radio users (collectively referred to as cells), in communication-constrained wireless networks modeled by interference graphs. Our goal is to achieve a globally stable and interference-aware channel allocation. Stability is defined through a generalized Gale-Shapley multi-to-one matching, a well-established solution concept in wireless resource allocation. We consider wireless networks where L cells share S orthogonal channels and cannot simultaneously use the same channel as their neighbors. Each channel evolves as an unknown restless Markov process with cell-dependent rewards, making this the first work to establish global Gale-Shapley stability for channel allocation in a stochastic, temporally varying restless environment. To address this challenge, we develop SMILE (Stable Multi-matching with Interference-aware LEarning), a communication-efficient distributed learning algorithm that integrates restless bandit learning with graph-constrained coordination. SMILE enables cells to distributedly balance exploration of unknown channels with exploitation of learned information. We prove that SMILE converges to the optimal stable allocation and achieves logarithmic regret relative to a genie with full knowledge of expected utilities. Simulations validate the theoretical guarantees and demonstrate SMILE's robustness, scalability, and efficiency across diverse spectrum-sharing scenarios.

cs.LG↗

arXiv2025

HERBench: A Benchmark for Multi-Evidence Integration in Video Question Answering

Video Large Language Models (Video-LLMs) are improving rapidly, yet current Video Question Answering (VideoQA) benchmarks often admit single-cue shortcuts, under-testing reasoning that must integrate evidence across time. We introduce HERBench, a benchmark designed to make multi-evidence integration unavoidable: each question requires at least three non-overlapping cues drawn from distinct video segments. HERBench contains 26,806 five-way multiple-choice questions across 12 compositional tasks. To make evidential demand measurable, we introduce the Minimum Required Frame-Set (MRFS), the smallest number of frames a model must fuse to answer correctly, and show that HERBench imposes higher evidential demand than prior benchmarks. Evaluating 13 state-of-the-art Video-LLMs yields only 31-42% accuracy, only modestly above the 20\% random-guess baseline. We disentangle this failure into two critical bottlenecks: (1) a retrieval deficit, where frame selectors overlook key evidence, and (2) a fusion deficit, where models fail to integrate information even when all necessary evidence is provided. HERBench thus provides a principled benchmark for studying robust multi-evidence video understanding.

cs.CV↗

arXiv2025

Control-Optimized Deep Reinforcement Learning for Artificially Intelligent Autonomous Systems

Deep reinforcement learning (DRL) has become a powerful tool for complex decision-making in machine learning and AI. However, traditional methods often assume perfect action execution, overlooking the uncertainties and deviations between an agent's selected actions and the actual system response. In real-world applications, such as robotics, mechatronics, and communication networks, execution mismatches arising from system dynamics, hardware constraints, and latency can significantly degrade performance. This work advances AI by developing a novel control-optimized DRL framework that explicitly models and compensates for action execution mismatches, a challenge largely overlooked in existing methods. Our approach establishes a structured two-stage process: determining the desired action and selecting the appropriate control signal to ensure proper execution. It trains the agent while accounting for action mismatches and controller corrections. By incorporating these factors into the training process, the AI agent optimizes the desired action with respect to both the actual control signal and the intended outcome, explicitly considering execution errors. This approach enhances robustness, ensuring that decision-making remains effective under real-world uncertainties. Our approach offers a substantial advancement for engineering practice by bridging the gap between idealized learning and real-world implementation. It equips intelligent agents operating in engineering environments with the ability to anticipate and adjust for actuation errors and system disturbances during training. We evaluate the framework in five widely used open-source mechanical simulation environments we restructured and developed to reflect real-world operating conditions, showcasing its robustness against uncertainties and offering a highly practical and efficient solution for control-oriented applications.

cs.RO↗

arXiv2025

Searching for a Hidden Markov Anomaly over Multiple Processes

We address the problem of detecting an anomalous process among a large number of processes. At each time t, normal processes are in state zero (normal state), while the abnormal process may be in either state zero (normal state) or state one (abnormal state), with the states being hidden. The transition between states for the abnormal process is governed by a Markov chain over time. At each time step, observations can be drawn from a selected subset of processes. Each probed process generates an observation depending on its hidden state, either a typical distribution under state zero or an abnormal distribution under state one. The objective is to design a sequential search strategy that minimizes the expected detection time, subject to an error probability constraint. In contrast to prior works that assume i.i.d. observations, we address a new setting where anomalies evolve according to a hidden Markov model. To this end, we propose a novel algorithm, dubbed Anomaly Detection under Hidden Markov model (ADHM), which dynamically adapts the probing strategy based on accumulated statistical evidence and predictive belief updates over hidden states. ADHM effectively leverages temporal correlations to focus sensing resources on the most informative processes. The algorithm is supported by an asymptotic theoretical foundation, grounded in an oracle analysis that characterizes the fundamental limits of detection under the assumption of a known distribution of the hidden states. In addition, the algorithm demonstrates strong empirical performance, consistently outperforming existing methods in extensive simulations.

eess.SP↗

arXiv2025

Tracking the Evolution of Near-Field Photonic Qubits into High-Dimensional Qudits via State Tomography

Quantum nanophotonics offers essential tools and technologies for controlling quantum states, while maintaining a miniature form factor and high scalability. For example, nanophotonic platforms can transfer information from the traditional degrees of freedom (DoFs), such as spin angular momentum (SAM) and orbital angular momentum (OAM), to the DoFs of the nanophotonic platform - and back, opening new directions for quantum information processing. Recent experiments have utilized the total angular momentum (TAM) of a photon as a unique means to produce entangled qubits in nanophotonic platforms. Yet, the process of transferring the information between the free-space DoFs and the TAM was never investigated, and its implications are still unknown. Here, we reveal the evolution of quantum information in heralded single photons as they couple into and out of the near-field of a nanophotonic system. Through quantum state tomography, we discover that the TAM qubit in the near-field becomes a free-space qudit entangled in the photonic SAM and OAM. The extracted density matrix and Wigner function in free-space indicate state preparation fidelity above 97%. The concepts described here bring new concepts and methodologies in developing high-dimensional quantum circuitry on a chip.

quant-ph↗

arXiv2025

PAUSE: Low-Latency and Privacy-Aware Active User Selection for Federated Learning

Federated learning (FL) enables multiple edge devices to collaboratively train a machine learning model without the need to share potentially private data. Federated learning proceeds through iterative exchanges of model updates, which pose two key challenges: First, the accumulation of privacy leakage over time, and second, communication latency. These two limitations are typically addressed separately: The former via perturbed updates to enhance privacy and the latter using user selection to mitigate latency - both at the expense of accuracy. In this work, we propose a method that jointly addresses the accumulation of privacy leakage and communication latency via active user selection, aiming to improve the trade-off among privacy, latency, and model performance. To achieve this, we construct a reward function that accounts for these three objectives. Building on this reward, we propose a multi-armed bandit (MAB)-based algorithm, termed Privacy-aware Active User SElection (PAUSE) which dynamically selects a subset of users each round while ensuring bounded overall privacy leakage. We establish a theoretical analysis, systematically showing that the reward growth rate of PAUSE follows that of the best-known rate in MAB literature. To address the complexity overhead of active user selection, we propose a simulated annealing-based relaxation of PAUSE and analyze its ability to approximate the reward-maximizing policy under reduced complexity. We numerically validate the privacy leakage, associated improved latency, and accuracy gains of our methods for the federated training in various scenarios.

cs.LG↗

arXiv2024

Asymptotically Optimal Search for a Change Point Anomaly under a Composite Hypothesis Model

We address the problem of searching for a change point in an anomalous process among a finite set of M processes. Specifically, we address a composite hypothesis model in which each process generates measurements following a common distribution with an unknown parameter (vector). This parameter belongs to either a normal or abnormal space depending on the current state of the process. Before the change point, all processes, including the anomalous one, are in a normal state; after the change point, the anomalous process transitions to an abnormal state. Our goal is to design a sequential search strategy that minimizes the Bayes risk by balancing sample complexity and detection accuracy. We propose a deterministic search algorithm with the following notable properties. First, we analytically demonstrate that when the distributions of both normal and abnormal processes are unknown, the algorithm is asymptotically optimal in minimizing the Bayes risk as the error probability approaches zero. In the second setting, where the parameter under the null hypothesis is known, the algorithm achieves asymptotic optimality with improved detection time based on the true normal state. Simulation results are presented to validate the theoretical findings.

stat.ML↗

arXiv2024

Plasmonic Twistronics: Discovery of Plasmonic Skyrmion Bags

The study of van der Waals heterostructures with an interlayer twist, known as "twistronics", has been instrumental in advancing contemporary condensed matter research. Most importantly, it has underpinned the emergence of a multitude of strongly-correlated phases, many of which derive from the topology of the physical system. Here, we explore the application of the twistronics paradigm in plasmonic systems with nontrivial topology, by creating a moiré skyrmion superlattice using two superimposed plasmonic skyrmion lattices, twisted at a "magic" angle. The complex electric field distribution of the moiré skyrmion superlattice is measured using time-resolved vector microscopy, revealing that each super-cell possesses very large topological invariants and harbors a "skyrmion bag", the size of which is controllable by the twist angle and center of rotation. Our work shows how twistronics leads to a diversity of topological features in optical fields, providing a new route to locally manipulate electromagnetic field distributions, which is crucial for future structured light-matter interaction.

physics.optics↗

arXiv2024

A Stable Polygamy Approach to Spectrum Access with Channel Reuse

We introduce a new and broader formulation of the stable marriage problem (SMP), called the stable polygamy problem (SPP), where multiple individuals from a larger group $L$ of $|L|$ individuals can be matched with a single individual from a smaller group $S$ of $|S|$ individuals. Each individual $\ell \in L$ possesses a social constraint set $C_{\ell}$ that contains a subset of $L$ with whom they cannot coexist harmoniously. We define a generalized concept of stability based on the preference and constraints of the individuals. We explore two common settings: common utility, where the utility of a match is the same for individuals from both sets, and preference ranking, where each individual has a preference ranking for every other individual in the opposite set. Our analysis is conducted using a novel graph-theoretical framework. The classic SMP has been investigated in recent years for spectrum access to match cells or users to channels, where only one-to-one matching is allowed. By contrast, the new SPP formulation allows us to solve more general models with channel reuse, where multiple users may access the same channel simultaneously. Interestingly, we show that classic algorithms, such as propose and reject (P&R), and Hungarian method are no longer efficient in the polygamy setting. We develop efficient algorithms to solve the SPP in polynomial time, tailored for implementations in spectrum access with channel reuse. We analytically show that our algorithm always solves the SPP with common utility. While the SPP with preference ranking cannot be solved by any algorithm in all cases, we prove that our algorithm effectively solves it in specific graph structures representing strong and weak interference regimes. Simulation results demonstrate the efficiency of our algorithms across various spectrum access scenarios.

eess.SY↗

arXiv2024

RRO: A Regularized Routing Optimization Algorithm for Enhanced Throughput and Low Latency with Efficient Complexity

In the rapidly evolving landscape of wireless networks, achieving enhanced throughput with low latency for data transmission is crucial for future communication systems. While low complexity OSPF-type solutions have shown effectiveness in lightly-loaded networks, they often falter in the face of increasing congestion. Recent approaches have suggested utilizing backpressure and deep learning techniques for route optimization. However, these approaches face challenges due to their high implementation and computational complexity, surpassing the capabilities of networks with limited hardware devices. A key challenge is developing algorithms that improve throughput and reduce latency while keeping complexity levels compatible with OSPF. In this collaborative research between Ben-Gurion University and Ceragon Networks Ltd., we address this challenge by developing a novel approach, dubbed Regularized Routing Optimization (RRO). The RRO algorithm offers both distributed and centralized implementations with low complexity, making it suitable for integration into 5G and beyond technologies, where no significant changes to the existing protocols are needed. It increases throughput while ensuring latency remains sufficiently low through regularized optimization. We analyze the computational complexity of RRO and prove that it converges with a level of complexity comparable to OSPF. Extensive simulation results across diverse network topologies demonstrate that RRO significantly outperforms existing methods.

cs.NI↗

arXiv2024

SINR-Aware Deep Reinforcement Learning for Distributed Dynamic Channel Allocation in Cognitive Interference Networks

We consider the problem of dynamic channel allocation (DCA) in cognitive communication networks with the goal of maximizing a global signal-to-interference-plus-noise ratio (SINR) measure under a specified target quality of service (QoS)-SINR for each network. The shared bandwidth is partitioned into K channels with frequency separation. In contrast to the majority of existing studies that assume perfect orthogonality or a one- to-one user-channel allocation mapping, this paper focuses on real-world systems experiencing inter-carrier interference (ICI) and channel reuse by multiple large-scale networks. This realistic scenario significantly increases the problem dimension, rendering existing algorithms inefficient. We propose a novel multi-agent reinforcement learning (RL) framework for distributed DCA, named Channel Allocation RL To Overlapped Networks (CARLTON). The CARLTON framework is based on the Centralized Training with Decentralized Execution (CTDE) paradigm, utilizing the DeepMellow value-based RL algorithm. To ensure robust performance in the interference-laden environment we address, CARLTON employs a low-dimensional representation of observations, generating a QoS-type measure while maximizing a global SINR measure and ensuring the target QoS-SINR for each network. Our results demonstrate exceptional performance and robust generalization, showcasing superior efficiency compared to alternative state-of-the-art methods, while achieving a marginally diminished performance relative to a fully centralized approach.

eess.SP↗

arXiv2023

Sparse Training for Federated Learning with Regularized Error Correction

Federated Learning (FL) has attracted much interest due to the significant advantages it brings to training deep neural network (DNN) models. However, since communications and computation resources are limited, training DNN models in FL systems face challenges such as elevated computational and communication costs in complex tasks. Sparse training schemes gain increasing attention in order to scale down the dimensionality of each client (i.e., node) transmission. Specifically, sparsification with error correction methods is a promising technique, where only important updates are sent to the parameter server (PS) and the rest are accumulated locally. While error correction methods have shown to achieve a significant sparsification level of the client-to-PS message without harming convergence, pushing sparsity further remains unresolved due to the staleness effect. In this paper, we propose a novel algorithm, dubbed Federated Learning with Accumulated Regularized Embeddings (FLARE), to overcome this challenge. FLARE presents a novel sparse training approach via accumulated pulling of the updated models with regularization on the embeddings in the FL process, providing a powerful solution to the staleness effect, and pushing sparsity to an exceptional level. The performance of FLARE is validated through extensive experiments on diverse and complex models, achieving a remarkable sparsity level (10 times and more beyond the current state-of-the-art) along with significantly improved accuracy. Additionally, an open-source software package has been developed for the benefit of researchers and developers in related fields.

cs.LG↗

arXiv2023

A Masked Pruning Approach for Dimensionality Reduction in Communication-Efficient Federated Learning Systems

Federated Learning (FL) represents a growing machine learning (ML) paradigm designed for training models across numerous nodes that retain local datasets, all without directly exchanging the underlying private data with the parameter server (PS). Its increasing popularity is attributed to notable advantages in terms of training deep neural network (DNN) models under privacy aspects and efficient utilization of communication resources. Unfortunately, DNNs suffer from high computational and communication costs, as well as memory consumption in intricate tasks. These factors restrict the applicability of FL algorithms in communication-constrained systems with limited hardware resources. In this paper, we develop a novel algorithm that overcomes these limitations by synergistically combining a pruning-based method with the FL process, resulting in low-dimensional representations of the model with minimal communication cost, dubbed Masked Pruning over FL (MPFL). The algorithm operates by initially distributing weights to the nodes through the PS. Subsequently, each node locally trains its model and computes pruning masks. These low-dimensional masks are then transmitted back to the PS, which generates a consensus pruning mask, broadcasted back to the nodes. This iterative process enhances the robustness and stability of the masked pruning model. The generated mask is used to train the FL model, achieving significant bandwidth savings. We present an extensive experimental study demonstrating the superior performance of MPFL compared to existing methods. Additionally, we have developed an open-source software package for the benefit of researchers and developers in related fields.

cs.LG↗

arXiv2023

Deep Multi-Agent Reinforcement Learning for Decentralized Active Hypothesis Testing

We consider a decentralized formulation of the active hypothesis testing (AHT) problem, where multiple agents gather noisy observations from the environment with the purpose of identifying the correct hypothesis. At each time step, agents have the option to select a sampling action. These different actions result in observations drawn from various distributions, each associated with a specific hypothesis. The agents collaborate to accomplish the task, where message exchanges between agents are allowed over a rate-limited communications channel. The objective is to devise a multi-agent policy that minimizes the Bayes risk. This risk comprises both the cost of sampling and the joint terminal cost incurred by the agents upon making a hypothesis declaration. Deriving optimal structured policies for AHT problems is generally mathematically intractable, even in the context of a single agent. As a result, recent efforts have turned to deep learning methodologies to address these problems, which have exhibited significant success in single-agent learning scenarios. In this paper, we tackle the multi-agent AHT formulation by introducing a novel algorithm rooted in the framework of deep multi-agent reinforcement learning. This algorithm, named Multi-Agent Reinforcement Learning for AHT (MARLA), operates at each time step by having each agent map its state to an action (sampling rule or stopping rule) using a trained deep neural network with the goal of minimizing the Bayes risk. We present a comprehensive set of experimental results that effectively showcase the agents' ability to learn collaborative strategies and enhance performance using MARLA. Furthermore, we demonstrate the superiority of MARLA over single-agent learning approaches. Finally, we provide an open-source implementation of the MARLA framework, for the benefit of researchers and developers in related domains.

stat.ML↗

arXiv2023

A Genetic Algorithm-Based Approach to Power Allocation in Rate-Splitting Multiple Access Systems

We consider the problem of power allocation in Rate-Splitting Multiple Access (RSMA) systems, where messages are split into common and private messages. The common and private streams are jointly transmitted to allow efficient use of the bandwidth, and decoded by Successive Interference Cancellation (SIC) at the receiver. However, the power allocation between streams significantly affects the overall performance. In this letter, we address this problem. We develop a novel algorithm, dubbed Power Allocation in RSMA systems using Genetic Algorithm (PARGA), to allocate the power between streams in RSMA systems in order to maximize the user sum-rate. Simulation results demonstrate the high efficiency of PARGA compared to existing methods.

eess.SY↗

arXiv2023

Multi-Flow Transmission in Wireless Interference Networks: A Convergent Graph Learning Approach

We consider the problem of of multi-flow transmission in wireless networks, where data signals from different flows can interfere with each other due to mutual interference between links along their routes, resulting in reduced link capacities. The objective is to develop a multi-flow transmission strategy that routes flows across the wireless interference network to maximize the network utility. However, obtaining an optimal solution is computationally expensive due to the large state and action spaces involved. To tackle this challenge, we introduce a novel algorithm called Dual-stage Interference-Aware Multi-flow Optimization of Network Data-signals (DIAMOND). The design of DIAMOND allows for a hybrid centralized-distributed implementation, which is a characteristic of 5G and beyond technologies with centralized unit deployments. A centralized stage computes the multi-flow transmission strategy using a novel design of graph neural network (GNN) reinforcement learning (RL) routing agent. Then, a distributed stage improves the performance based on a novel design of distributed learning updates. We provide a theoretical analysis of DIAMOND and prove that it converges to the optimal multi-flow transmission strategy as time increases. We also present extensive simulation results over various network topologies (random deployment, NSFNET, GEANT2), demonstrating the superior performance of DIAMOND compared to existing methods.

cs.LG↗