arXiv subjects

Wei Pan

Publications and source records attributed to Wei Pan.

At least 73 records · Page 4Linked to original sources

arXiv2025

ResBench: A Comprehensive Framework for Evaluating Database Resilience

Existing database benchmarks primarily focus on performance under ideal running environments. However, in real-world scenarios, databases probably face numerous adverse events. Quantifying the ability to cope with these events from a comprehensive perspective remains an open problem. We provide the definition of database resilience to describe its performance when facing adversity and propose ResBench, a benchmark for evaluating database resilience. This framework achieves automation, standardization, and visualization of the testing process through clear hierarchical decoupling. ResBench simulates adverse events and injects them during normal transaction processing, utilizing a module to gather multiple metrics for the evaluation model. We assess database resilience across eight dimensions: throughput, latency, stability, resistance, recovery, disturbance period, adaptation capability and metric deviation. All the results are presented to users via a user-friendly graphical interface. We demonstrate the execution process and result interpretation of ResBench using two types of adversity datasets.

cs.DB↗

arXiv2025

Synergistic Antenna-Modulator Integration for Monolithic Photonic RF Receiver

Integrated radio-frequency (RF) photonics plays a pivotal role in wireless communications, sensing, and radar due to its large intrinsic bandwidth, remote distribution capability, and compact footprint. However, despite significant advances in photonic integrated circuits (PICs), the practical deployment of these systems remains constrained by the bulky nature of essential RF components (e.g., bulky antennas, amplifiers, and cables), especially in covert, conformal, and space-constrained applications. To overcome these limitations, monolithic electronic-photonic integrated circuits (EPICs), enabling miniaturized and synergistic integration of both RF and photonic components, are gaining notable attention. As a groundbreaking advancement, we demonstrate a novel photonic RF receiver that monolithically integrates a bow-tie antenna and a microring modulator on a thin-film lithium niobate platform. The chip innovatively leverages dual-resonance enhancement mechanism, RF resonance from the antenna and optical resonance from the microring, to significantly boost the RF-to-optical conversion efficiency. A record-high figure of merit (FOM) of 3.88 W-1/2 is achieved within a compact footprint of 2*1.7 mm2. As the first demonstrations, the integrated receiver is deployed in an integrated sensing and communication (ISAC) system, achieving centimeter-level radar ranging accuracy and 3.2 Gbps wireless communication capacity, as well as real-time video transmission function in moving scenarios. This seminal work paves a new way for covert, conformal, and miniaturized frontends in wireless communication and sensing applications, including body area networks, unmanned aerial vehicles, high-speed vacuum maglevs, and electronic warfare systems.

physics.optics↗

arXiv2025

LoKO: Low-Rank Kalman Optimizer for Online Fine-Tuning of Large Models

Training large models with millions or even billions of parameters from scratch incurs substantial computational costs. Parameter Efficient Fine-Tuning (PEFT) methods, particularly Low-Rank Adaptation (LoRA), address this challenge by adapting only a reduced number of parameters to specific tasks with gradient-based optimizers. In this paper, we cast PEFT as an optimal filtering/state estimation problem and present Low-Rank Kalman Optimizer (LoKO) to estimate the optimal trainable parameters in an online manner. We leverage the low-rank decomposition in LoRA to significantly reduce matrix sizes in Kalman iterations and further capitalize on a diagonal approximation of the covariance matrix to effectively decrease computational complexity from quadratic to linear in the number of trainable parameters. Moreover, we discovered that the initialization of the covariance matrix within the Kalman algorithm and the accurate estimation of the observation noise covariance are the keys in this formulation, and we propose robust approaches that work well across a vast range of well-established computer vision and language models. Our results show that LoKO converges with fewer iterations and yields better performance models compared to commonly used optimizers with LoRA in both image classifications and language tasks. Our study opens up the possibility of leveraging the Kalman filter as an effective optimizer for the online fine-tuning of large models.

cs.LG↗

arXiv2025

Analog-to-Digital Converter Based on Voltage-controlled Superconducting Device

The increasing demand for cryogenic electronics in superconducting and quantum computing systems calls for ultra energy efficient data conversion architectures that remain functional at deep cryogenic temperatures.In this work, we present the first design of a voltage-controlled superconducting flash analog-to-digital converter (ADC) based on a novel quantum enhanced Josephson junction field effect transistor (JJFET).Exploiting its strong gate tunability and transistor-like behavior, the JJFET offers a scalable alternative to conventional current controlled superconducting devices while aligning naturally with CMOS style design methodologies.Building on our previously developed Verilog A compact model calibrated to experimental data, we design and simulate a three bit JJFET based flash ADC.The core comparator block is realized through careful bias current selection and augmented with a three terminal nanocryotron to precisely define reference voltages.Cascaded JJFET comparators ensure robust voltage gain, cascadability, and logic level restoration across stages.Simulation results demonstrate accurate quantization behavior with ultra-low power dissipation, underscoring the feasibility of voltage driven superconducting mixed signal circuits.This work establishes a critical step toward unifying superconducting logic and data conversion, paving the way for scalable cryogenic architectures in quantum classical co-processors, low-power AI accelerators, and next generation energy constrained computing platforms.

cond-mat.supr-con↗

arXiv2025

Bayesian Natural Gradient Fine-Tuning of CLIP Models via Kalman Filtering

Vision-language pre-trained models, such as CLIP, have established new benchmarks in multimodal data mining. In such models, few-shot fine-tuning is a major challenge to achieve optimal performance on both in-distribution (ID) and out-of-distribution (OOD) datasets, especially when labeled data is scarce. Most existing fine-tuning approaches rely on first-order gradient-based optimizers, which typically suffer from slow convergence, sensitivity to step-size hyperparameters, and poor generalization in OOD settings. In contrast, second-order methods utilize local curvature information of the loss landscape to adjust the update step size. This is particularly beneficial for CLIP models, whose non-convex loss functions often contain sharp critical points. In such cases, natural gradient direction can offer more substantial and efficient per-iteration updates when fine-tuning with limited data. Natural Gradient Descent (NGD) is obtained by preconditioning the standard gradient with the inverse Fisher Information Matrix (FIM), which is computationally expensive for large models. To address this, we propose a Bayesian approximation of NGD using a Kalman filter for CLIP models. Our method combines the benefits of second-order optimization with Bayesian inference, which enhances generalization while providing uncertainty quantification. Extensive experiments conducted on diverse image classification datasets demonstrate that our algorithm consistently achieves superior--or comparable--ID performance and improved OOD robustness compared to state-of-the-art baselines. To the best of our knowledge, this work represents the first successful application of Kalman filtering to fine-tuning CLIP-based models, which enables more robust and efficient learning in vision-language tasks.

cs.LG↗

arXiv2025

PoseDiff: A Unified Diffusion Model Bridging Robot Pose Estimation and Video-to-Action Control

We present PoseDiff, a conditional diffusion model that unifies robot state estimation and control within a single framework. At its core, PoseDiff maps raw visual observations into structured robot states-such as 3D keypoints or joint angles-from a single RGB image, eliminating the need for multi-stage pipelines or auxiliary modalities. Building upon this foundation, PoseDiff extends naturally to video-to-action inverse dynamics: by conditioning on sparse video keyframes generated by world models, it produces smooth and continuous long-horizon action sequences through an overlap-averaging strategy. This unified design enables scalable and efficient integration of perception and control. On the DREAM dataset, PoseDiff achieves state-of-the-art accuracy and real-time performance for pose estimation. On Libero-Object manipulation tasks, it substantially improves success rates over existing inverse dynamics modules, even under strict offline settings. Together, these results show that PoseDiff provides a scalable, accurate, and efficient bridge between perception, planning, and control in embodied AI. The video visualization results can be found on the project page: https://haozhuo-zhang.github.io/PoseDiff-project-page/.

cs.RO↗

arXiv2025

Spatial-Aware Decision-Making with Ring Attractors in Reinforcement Learning Systems

Ring attractors, mathematical models inspired by neural circuit dynamics, provide a biologically plausible mechanism to improve learning speed and accuracy in Reinforcement Learning (RL). Serving as specialized brain-inspired structures that encode spatial information and uncertainty, ring attractors explicitly encode the action space, facilitate the organization of neural activity, and enable the distribution of spatial representations across the neural network in the context of Deep Reinforcement Learning (DRL). These structures also provide temporal filtering that stabilizes action selection during exploration, for example, by preserving the continuity between rotation angles in robotic control or adjacency between tactical moves in game-like environments. The application of ring attractors in the action selection process involves mapping actions to specific locations on the ring and decoding the selected action based on neural activity. We investigate the application of ring attractors by both building an exogenous model and integrating them as part of DRL agents. Our approach significantly improves state-of-the-art performance on the Atari 100k benchmark, achieving a 53% increase in performance over selected baselines.

cs.LG↗

arXiv2025

Semiconductor Meta-Graphene and Valleytronics

Nano-patterned semiconductor interfaces offer a versatile platform for creating quantum metamaterials and exploring novel electronic phenomena. In this study, we illustrate this concept using artificial graphene--a metamaterial featuring distinctive properties including Dirac and saddle points. We demonstrate that introducing additional nano-patterning can open a Dirac band gap, giving rise to what we term artificial hexagonal boron nitride (AhBN). The calculated valley Chern number of AhBN indicates the presence of topological valley Hall states confined to Dirac-gap domain walls. A key question is whether these one-dimensional edge states are topologically protected against disorder, given their vulnerability to Anderson localization. To this end, we perform band structure and electronic transport simulations under experimentally relevant disorder, including charge puddles and geometric imperfections. Our results reveal the resilience of the domain wall states against typical experimental disorder, particularly while the AhBN band gap remains open. The localization length along the domain wall can reach several microns--several times longer than the bulk electron mean free path--even though the number of bulk transport channels is greater. To enhance the effectiveness of the low-dissipation domain wall channel, we propose ribbon geometries with a large length-to-width ratio. These findings underscore both the potential and challenges of AhBN for low-energy, power-efficient microelectronic applications.

cond-mat.mes-hall↗

arXiv2025

Practical security of local local oscillator continuous-variable quantum key distribution systems with pulse width mismatch

In continuous-variable quantum key distribution (CVQKD) systems, using a local local oscillator (LLO) scheme removes the local oscillator side channel, enhances Bob's detection performance and reduces the excess noise caused by photon leakage, thereby effectively improves the system's security and performance. However, in this scheme, since the signal and the LO are generated by different lasers and the signal propagates through the untrusted quantum channel, their pulse widths may become mismatched. Such mismatches may reduce the precision of detection at Bob, affecting parameter estimation processes and leading to inaccurate calculations of the secret key rate. Moreover, mismatches may introduce potential security loopholes. Thus, this paper investigates the practical security issues of the LLO-CVQKD system when pulse width mismatch occurs between the local oscillator and the signal. We first model the case of pulse width mismatch and analyze its impact on Bob's detection. Then, we simulate the secret key rate under different mismatch levels. Based on the analysis, we find that under such mismatch, the key parameters involved in secret key rate calculation are incorrectly estimated, leading to an overestimation of the system's secret key rate. Therefore, this imperfect mismatch can open a loophole for Eve to perform attacks in practical systems. To close this loophole, we design a scheme at Bob to monitor the pulse widths of both the signal and the local oscillator, and to reshape the waveform of the local oscillator so that the two lights are matched again. This method eliminates the adverse effects caused by pulse width mismatch and effectively resists Eve's attacks which exploit this loophole.

quant-ph↗

arXiv2025

From Grunts to Lexicons: Emergent Language from Cooperative Foraging

Language is a powerful communicative and cognitive tool. It enables humans to express thoughts, share intentions, and reason about complex phenomena. Despite our fluency in using and understanding language, the question of how it arises and evolves over time remains unsolved. A leading hypothesis in linguistics and anthropology posits that language evolved to meet the ecological and social demands of early human cooperation. Language did not arise in isolation, but through shared survival goals. Inspired by this view, we investigate the emergence of language in multi-agent Foraging Games. These environments are designed to reflect the cognitive and ecological constraints believed to have influenced the evolution of communication. Agents operate in a shared grid world with only partial knowledge about other agents and the environment, and must coordinate to complete games like picking up high-value targets or executing temporally ordered actions. Using end-to-end deep reinforcement learning, agents learn both actions and communication strategies from scratch. We find that agents develop communication protocols with hallmark features of natural language: arbitrariness, interchangeability, displacement, cultural transmission, and compositionality. We quantify each property and analyze how different factors, such as population size, social dynamics, and temporal dependencies, shape specific aspects of the emergent language. Our framework serves as a platform for studying how language can evolve from partial observability, temporal reasoning, and cooperative goals in embodied multi-agent settings. We will release all data, code, and models publicly.

cs.AI↗

arXiv2025

Embodied Arena: A Comprehensive, Unified, and Evolving Evaluation Platform for Embodied AI

Embodied AI development significantly lags behind large foundation models due to three critical challenges: (1) lack of systematic understanding of core capabilities needed for Embodied AI, making research lack clear objectives; (2) absence of unified and standardized evaluation systems, rendering cross-benchmark evaluation infeasible; and (3) underdeveloped automated and scalable acquisition methods for embodied data, creating critical bottlenecks for model scaling. To address these obstacles, we present Embodied Arena, a comprehensive, unified, and evolving evaluation platform for Embodied AI. Our platform establishes a systematic embodied capability taxonomy spanning three levels (perception, reasoning, task execution), seven core capabilities, and 25 fine-grained dimensions, enabling unified evaluation with systematic research objectives. We introduce a standardized evaluation system built upon unified infrastructure supporting flexible integration of 22 diverse benchmarks across three domains (2D/3D Embodied Q&A, Navigation, Task Planning) and 30+ advanced models from 20+ worldwide institutes. Additionally, we develop a novel LLM-driven automated generation pipeline ensuring scalable embodied evaluation data with continuous evolution for diversity and comprehensiveness. Embodied Arena publishes three real-time leaderboards (Embodied Q&A, Navigation, Task Planning) with dual perspectives (benchmark view and capability view), providing comprehensive overviews of advanced model capabilities. Especially, we present nine findings summarized from the evaluation results on the leaderboards of Embodied Arena. This helps to establish clear research veins and pinpoint critical research problems, thereby driving forward progress in the field of Embodied AI.

cs.RO↗

arXiv2025

Generative Visual Foresight Meets Task-Agnostic Pose Estimation in Robotic Table-Top Manipulation

Robotic manipulation in unstructured environments requires systems that can generalize across diverse tasks while maintaining robust and reliable performance. We introduce {GVF-TAPE}, a closed-loop framework that combines generative visual foresight with task-agnostic pose estimation to enable scalable robotic manipulation. GVF-TAPE employs a generative video model to predict future RGB-D frames from a single side-view RGB image and a task description, offering visual plans that guide robot actions. A decoupled pose estimation model then extracts end-effector poses from the predicted frames, translating them into executable commands via low-level controllers. By iteratively integrating video foresight and pose estimation in a closed loop, GVF-TAPE achieves real-time, adaptive manipulation across a broad range of tasks. Extensive experiments in both simulation and real-world settings demonstrate that our approach reduces reliance on task-specific action data and generalizes effectively, providing a practical and scalable solution for intelligent robotic systems.

cs.RO↗

arXiv2025

Explosive Jumping with Rigid and Articulated Soft Quadrupeds via Example Guided Reinforcement Learning

Achieving controlled jumping behaviour for a quadruped robot is a challenging task, especially when introducing passive compliance in mechanical design. This study addresses this challenge via imitation-based deep reinforcement learning with a progressive training process. To start, we learn the jumping skill by mimicking a coarse jumping example generated by model-based trajectory optimization. Subsequently, we generalize the learned policy to broader situations, including various distances in both forward and lateral directions, and then pursue robust jumping in unknown ground unevenness. In addition, without tuning the reward much, we learn the jumping policy for a quadruped with parallel elasticity. Results show that using the proposed method, i) the robot learns versatile jumps by learning only from a single demonstration, ii) the robot with parallel compliance reduces the landing error by 11.1%, saves energy cost by 15.2% and reduces the peak torque by 15.8%, compared to the rigid robot without parallel elasticity, iii) the robot can perform jumps of variable distances with robustness against ground unevenness (maximal 4cm height perturbations) using only proprioceptive perception.

cs.RO↗

arXiv2025

Diffusion-Driven High-Dimensional Variable Selection

Variable selection for high-dimensional, highly correlated data has long been a challenging problem, often yielding unstable and unreliable models. We propose a resample-aggregate framework that exploits diffusion models' ability to generate high-fidelity synthetic data. Specifically, we draw multiple pseudo-data sets from a diffusion model fitted to the original data, apply any off-the-shelf selector (e.g., lasso or SCAD), and store the resulting inclusion indicators and coefficients. Aggregating across replicas produces a stable subset of predictors with calibrated stability scores for variable selection. Theoretically, we show that the proposed method is selection consistent under mild assumptions. Because the generative model imports knowledge from large pre-trained weights, the procedure naturally benefits from transfer learning, boosting power when the observed sample is small or noisy. We also extend the framework of aggregating synthetic data to other model selection problems, including graphical model selection, and statistical inference that supports valid confidence intervals and hypothesis tests. Extensive simulations show consistent gains over the lasso, stability selection, and knockoff baselines, especially when predictors are strongly correlated, achieving higher true-positive rates and lower false-discovery proportions. By coupling diffusion-based data augmentation with principled aggregation, our method advances variable selection methodology and broadens the toolkit for interpretable, statistically rigorous analysis in complex scientific applications.

stat.ME↗

arXiv2025

TAR: Teacher-Aligned Representations via Contrastive Learning for Quadrupedal Locomotion

Quadrupedal locomotion via Reinforcement Learning (RL) is commonly addressed using the teacher-student paradigm, where a privileged teacher guides a proprioceptive student policy. However, key challenges such as representation misalignment between privileged teacher and proprioceptive-only student, covariate shift due to behavioral cloning, and lack of deployable adaptation; lead to poor generalization in real-world scenarios. We propose Teacher-Aligned Representations via Contrastive Learning (TAR), a framework that leverages privileged information with self-supervised contrastive learning to bridge this gap. By aligning representations to a privileged teacher in simulation via contrastive objectives, our student policy learns structured latent spaces and exhibits robust generalization to Out-of-Distribution (OOD) scenarios, surpassing the fully privileged "Teacher". Results showed accelerated training by 2x compared to state-of-the-art baselines to achieve peak performance. OOD scenarios showed better generalization by 40% on average compared to existing methods. Moreover, TAR transitions seamlessly into learning during deployment without requiring privileged states, setting a new benchmark in sample-efficient, adaptive locomotion and enabling continual fine-tuning in real-world scenarios. Open-source code and videos are available at https://amrmousa.com/TARLoco/.

cs.RO↗

arXiv2025

Reimagining Voltage-Controlled Cryogenic Boolean Logic Paradigm with Quantum-Enhanced Josephson Junction FETs

The growing demand for ultra low power computing and the emergence of quantum technologies have intensified interest in cryogenic electronics, particularly superconducting devices. Despite their promise, current controlled superconducting components face fundamental challenges in cascadability, limiting their effectiveness in complex logic architectures. To overcome this, recent efforts have focused on developing gate tunable superconducting devices, such as Josephson junction field effect transistors (JJFETs). However, achieving robust control and sufficient supercurrent gain, both critical for transistor-like performance in logic circuits remains a key challenge. A recent advancement in JJFET design, based on InAs and GaSb heterostructures, demonstrates enhanced gain and favorable device characteristics suitable for circuit integration. Building on this innovation, we propose and analyze fundamental voltage controlled logic topologies using the quantum enhanced JJFET. We develop a Verilog A based circuit compatible compact model of the quantum enhanced JJFET which accurately captures the experimentally observed device characteristics. To ensure cascadability, our logic circuits incorporate the multilayered Heater Nanocryotron (nTron), a superconducting nanowire-based thermal switch. Through simulation based analysis, we demonstrate the successful implementation of fundamental logic gates, including NOT, NAND, and NOR. Furthermore, we design a 3 input majority gate, which plays a pivotal role in quantum and reversible computing due to its universality. Finally, to demonstrate the cascadability of our proposed logic topology, we demonstrate the operation of a 2 input XOR gate based on our designed JJFET based NOT, NAND, and NOR gate.

cs.ET↗

arXiv2025

Reinforcement Learning for Multi-Objective Multi-Echelon Supply Chain Optimisation

This study develops a generalised multi-objective, multi-echelon supply chain optimisation model with non-stationary markets based on a Markov decision process, incorporating economic, environmental, and social considerations. The model is evaluated using a multi-objective reinforcement learning (RL) method, benchmarked against an originally single-objective RL algorithm modified with weighted sum using predefined weights, and a multi-objective evolutionary algorithm (MOEA)-based approach. We conduct experiments on varying network complexities, mimicking typical real-world challenges using a customisable simulator. The model determines production and delivery quantities across supply chain routes to achieve near-optimal trade-offs between competing objectives, approximating Pareto front sets. The results demonstrate that the primary approach provides the most balanced trade-off between optimality, diversity, and density, further enhanced with a shared experience buffer that allows knowledge transfer among policies. In complex settings, it achieves up to 75\% higher hypervolume than the MOEA-based method and generates solutions that are approximately eleven times denser, signifying better robustness, than those produced by the modified single-objective RL method. Moreover, it ensures stable production and inventory levels while minimising demand loss.

cs.AI↗

arXiv2025

Deep Koopman Learning of Nonlinear Time-Varying Systems

This paper presents a data-driven approach to approximate the dynamics of a nonlinear time-varying system (NTVS) by a linear time-varying system (LTVS), which is resulted from the Koopman operator and deep neural networks. Analysis of the approximation error between states of the NTVS and the resulting LTVS is presented. Simulations on a representative NTVS show that the proposed method achieves small approximation errors, even when the system changes rapidly. Furthermore, simulations in an example of quadcopters demonstrate the computational efficiency of the proposed approach.

eess.SY↗