arXiv subjects

Yuqing Zhu

Publications and source records attributed to Yuqing Zhu.

At least 19 recordsLinked to original sources

arXiv2026

SeekBrain: An Autonomous Multi-Agent System for Accelerating Neuroscience Discovery

Modern neuroscience relies on integrating multi-scale, multimodal datasets to uncover the neural principles underlying intelligence. However, analytical challenges posed by highly heterogeneous data and fragmented workflows increasingly constrain discoveries. Here we introduce SeekBrain, an autonomous multi-agent framework designed to accelerate neuroscience discovery through domain-grounded hierarchical planning and cross-modal data analysis. SeekBrain dynamically constructs a repertoire of analysis recipes extracted from code-paper pairs. By coupling this codified expertise with agentic planning and execution engines, the framework scalably generates hypotheses and analytical pipelines on demand. Systematic evaluation on the expert-annotated BrainArena benchmark demonstrates that SeekBrain substantially outperforms state-of-the-art agent baselines across various analysis tasks. Crucially, when deployed in real-world research, SeekBrain integrated behavioral, neural, and anatomical data to reveal structured, distributed neural representations of larval zebrafish behavior and a shared axis of regional decoding strength across the brain in a mouse decision-making task. These results establish SeekBrain as a scalable and practical tool for accelerating data-driven discoveries in neuroscience.

cs.MA↗

arXiv2026

Ultrafast Light-Induced Magnetoelectric Effect in van der Waals Magnetic Semiconductor Heterostructures

Atomic-scale heterostructures of van der Waals (vdW) magnets and semiconductors provide a unique environment for exploring magnetic dynamics. In contrast to typical photothermal excitation of precessional magnetization dynamics by a pump laser pulse, we find that ultrafast optical excitation of a WS$_2$/CrGeTe$_3$ (CGT) bilayer produces an opposite sign of magnetic torque compared to an isolated CGT film. Experimental observations by time-resolved magneto-optic Kerr effect (TR-MOKE) and theoretical analysis by density functional theory (DFT) and Landau-Lifshitz-Gilbert (LLG) simulations support a mechanism in which charge transfer of photoexcited carriers across the interface alters the perpendicular magnetic anisotropy, which in turn generates a torque on the magnetic layer to trigger precessional magnetization dynamics. These results provide new avenues for ultrafast manipulation of magnetization in vdW heterostructures with type-II band alignments. Lastly, we show that optically-generated spin currents from WS$_2$ into CGT can also trigger precessional dynamics via angular momentum transfer.

cond-mat.mes-hall↗

arXiv2026

PRIME: Efficient Algorithm for Token Graph Routing Problem

Optimizing asset exchanges on blockchain-driven platforms poses a novel and challenging graph query optimization problem. In this model, assets represent vertices and exchanges form edges, recasting the graph query task as a routing problem over a large-scale, dynamic graph. However, the existing solutions fail to solve the problem efficiently due to the non-linear nature of the edge weights defined by a concave swap function. To address the challenge, we propose PRIME, a two-stage iterative graph algorithm designed for the Token Graph Routing Problem (TGRP). The first stage employs a pruned graph search to efficiently identify a set of high-potential routing paths. The second stage formulates the allocation task as a strongly convex optimization problem, which we solve using our novel Adaptive Sign Gradient Method (ASGM) with a linear convergence rate. Extensive experiments on real-world Ethereum data confirm PRIME's advantages over industry baselines. PRIME consistently outperforms the widely-used Uniswap routing algorithm, achieving up to 8.42 basis points (bps) better execution prices on large trades while reducing computation up to 96.7%. The practicality of PRIME is further validated by its deployment in hedge fund production environments, demonstrating its viability as a scalable graph query processing solution for high-frequency decentralized markets.

cs.DB↗

arXiv2025

Age of Information for Constrained Scheduling with Imperfect Feedback

This paper considers a downlink system where an access point sends the monitored status of multiple sources to multiple users. By jointly accounting for imperfect feedback and constrained transmission rate, which are key limited factors in practical systems, we aim to design scheduling algorithms to optimize the age of information (AoI) over the infinite time horizon. For zero feedback under the generate-at-will traffic, we derive a closed-form lower bound of achievable AoI, which, to the best of our knowledge, reflects the impact of zero feedback for the first time, and propose a policy that achieves this bound in many cases by jointly applying rate splitting and modular arithmetic. For zero feedback under the Bernoulli traffic, we develop a drift-plus-penalty (DPP) policy with a threshold structure based on the theory of Lyapunov optimization and provide a closed-form performance guarantee. Furthermore, we extend the design of this DPP policy to support general imperfect feedback without increasing the online computational complexity. Numerical results verify our theoretical analysis and the AoI advantage of the proposed policies over state-of-the-art policies.

cs.IT↗

arXiv2025

Agentic Reinforcement Learning for Real-World Code Repair

We tackle the challenge of training reliable code-fixing agents in real repositories, where complex builds and shifting dependencies make evaluation unstable. We developed a verifiable pipeline with success defined as post-fix build validation and improved reproducibility across ~1K real issues by pinning dependencies and disabling automatic upgrades. Building on this, we introduced a scalable simplified pipeline for large-scale reinforcement learning (RL). Using this setup, we supervised fine-tuned Qwen3-32B in the full pipeline and applied RL on top of the SFT model in the simplified environment. The SFT model distilled from GPT-4.1 trajectories performs on par while being 56x smaller, and RL added 7-20% absolute gains under matched train-test conditions. "Thinking mode" was on par or worse in our experiments. Both SFT and RL models failed to generalize across environments, highlighting the importance of matching train-test environments for building reliable real-world code-fixing agents.

cs.LG↗

arXiv2025

TokenShapley: Token Level Context Attribution with Shapley Value

Large language models (LLMs) demonstrate strong capabilities in in-context learning, but verifying the correctness of their generated responses remains a challenge. Prior work has explored attribution at the sentence level, but these methods fall short when users seek attribution for specific keywords within the response, such as numbers, years, or names. To address this limitation, we propose TokenShapley, a novel token-level attribution method that combines Shapley value-based data attribution with KNN-based retrieval techniques inspired by recent advances in KNN-augmented LLMs. By leveraging a precomputed datastore for contextual retrieval and computing Shapley values to quantify token importance, TokenShapley provides a fine-grained data attribution approach. Extensive evaluations on four benchmarks show that TokenShapley outperforms state-of-the-art baselines in token-level attribution, achieving an 11-23% improvement in accuracy.

cs.CL↗

arXiv2025

Adapting to Linear Separable Subsets with Large-Margin in Differentially Private Learning

This paper studies the problem of differentially private empirical risk minimization (DP-ERM) for binary linear classification. We obtain an efficient $(\varepsilon,\delta)$-DP algorithm with an empirical zero-one risk bound of $\tilde{O}\left(\frac{1}{\gamma^2\varepsilon n} + \frac{|S_{\mathrm{out}}|}{\gamma n}\right)$ where $n$ is the number of data points, $S_{\mathrm{out}}$ is an arbitrary subset of data one can remove and $\gamma$ is the margin of linear separation of the remaining data points (after $S_{\mathrm{out}}$ is removed). Here, $\tilde{O}(\cdot)$ hides only logarithmic terms. In the agnostic case, we improve the existing results when the number of outliers is small. Our algorithm is highly adaptive because it does not require knowing the margin parameter $\gamma$ or outlier subset $S_{\mathrm{out}}$. We also derive a utility bound for the advanced private hyperparameter tuning algorithm.

cs.LG↗

arXiv2025

Training People to Reward Robots

Learning from demonstration (LfD) is a technique that allows expert teachers to teach task-oriented skills to robotic systems. However, the most effective way of guiding novice teachers to approach expert-level demonstrations quantitatively for specific teaching tasks remains an open question. To this end, this paper investigates the use of machine teaching (MT) to guide novice teachers to improve their teaching skills based on reinforcement learning from demonstration (RLfD). The paper reports an experiment in which novices receive MT-derived guidance to train their ability to teach a given motor skill with only 8 demonstrations and generalise this to previously unseen ones. Results indicate that the MT-guidance not only enhances robot learning performance by 89% on the training skill but also causes a 70% improvement in robot learning performance on skills not seen by subjects during training. These findings highlight the effectiveness of MT-guidance in upskilling human teaching behaviours, ultimately improving demonstration quality in RLfD.

cs.RO↗

arXiv2025

LEDD: Large Language Model-Empowered Data Discovery in Data Lakes

Data discovery in data lakes with ever increasing datasets has long been recognized as a big challenge in the realm of data management, especially for semantic search of and hierarchical global catalog generation of tables. While large language models (LLMs) facilitate the processing of data semantics, challenges remain in architecting an end-to-end system that comprehensively exploits LLMs for the two semantics-related tasks. In this demo, we propose LEDD, an end-to-end system with an extensible architecture that leverages LLMs to provide hierarchical global catalogs with semantic meanings and semantic table search for data lakes. Specifically, LEDD can return semantically related tables based on natural-language specification. These features make LEDD an ideal foundation for downstream tasks such as model training and schema linking for text-to-SQL tasks. LEDD also provides a simple Python interface to facilitate the extension and the replacement of data discovery algorithms.

cs.DB↗

arXiv2024

Neural Collapse Meets Differential Privacy: Curious Behaviors of NoisyGD with Near-perfect Representation Learning

A recent study by De et al. (2022) has reported that large-scale representation learning through pre-training on a public dataset significantly enhances differentially private (DP) learning in downstream tasks, despite the high dimensionality of the feature space. To theoretically explain this phenomenon, we consider the setting of a layer-peeled model in representation learning, which results in interesting phenomena related to learned features in deep learning and transfer learning, known as Neural Collapse (NC). Within the framework of NC, we establish an error bound indicating that the misclassification error is independent of dimension when the distance between actual features and the ideal ones is smaller than a threshold. Additionally, the quality of the features in the last layer is empirically evaluated under different pre-trained models within the framework of NC, showing that a more powerful transformer leads to a better feature representation. Furthermore, we reveal that DP fine-tuning is less robust compared to fine-tuning without DP, particularly in the presence of perturbations. These observations are supported by both theoretical analyses and experimental evaluation. Moreover, to enhance the robustness of DP fine-tuning, we suggest several strategies, such as feature normalization or employing dimension reduction methods like Principal Component Analysis (PCA). Empirically, we demonstrate a significant improvement in testing accuracy by conducting PCA on the last-layer features.

cs.LG↗

arXiv2024

Using Machine Teaching to Boost Novices' Robot Teaching Skill

Recent evidence has shown that, contrary to expectations, it is difficult for users, especially novices, to teach robots tasks through LfD. This paper introduces a framework that leverages MT algorithms to train novices to become better teachers of robots, and verifies whether such teaching ability is retained beyond the period of training and generalises such that novices teach robots more effectively, even for skills for which training has not been received. A between-subjects study is reported, in which novice teachers are asked to teach simple motor skills to a robot. The results demonstrate that subjects that receive training show average 78.83% improvement in teaching ability (as measured by accuracy of the skill learnt by the robot), and average 63.69% improvement in the teaching of new skills not included as part of the training.

cs.RO↗

arXiv2024

Jammer Mitigation in Absorptive RIS-Assisted Uplink NOMA

Non-orthogonal multiple access (NOMA) is a promising technology for next-generation wireless communication systems due to its enhanced spectral efficiency. In this paper, we consider an uplink NOMA system operating together with a high-dimensional absorptive reconfigurable intelligent surface (A-RIS). We aim to minimize the total power transmitted by the users in order to meet signal-to-interference-plus-noise constraints at the base station in the presence of a jammer. We propose an iterative algorithm to solve the high-dimensional non-convex optimization problem using linear programming to find the transmit powers and a fractional programming algorithm based on the Dinkelbach algorithm with a sequential convex relaxation procedure to optimize the reflection coefficients. We show that our algorithm converges on large optimization problems, with a jammer comprising as many as $64$ antennas, and an A-RIS with $128$ elements. Our numerical results show that, compared with a standard RIS that reflects all impinging energy, the A-RIS can dramatically reduce the users' required transmit power and successfully mitigate interference from the jammer. The absorption capability of the A-RIS is in particular useful in cases when the number of jammer antennas is of the same order as the number of A-RIS elements.

cs.IT↗

arXiv2024

Age-Gain-Dependent Random Access for Event-Driven Periodic Updating

This paper considers utilizing the knowledge of age gains to reduce the network average age of information (AoI) in random access with event-driven periodic updating for the first time. Built on the form of slotted ALOHA, we require each device to determine its age gain threshold and transmission probability in an easily implementable decentralized manner, so that the unavoided contention can be limited to devices with age gains as high as possible. For the basic case that each device utilizes its knowledge of age gain of only itself, we provide an analytical modeling approach by a multi-layer discrete-time Markov chains (DTMCs), where an external infinite-horizon DTMC manages the jumps between the beginnings of frames and an internal finite-horizon DTMC manages the evolution during an arbitrary frame. Such modelling enables that optimal access parameters can be obtained offline. For the enhanced case that each device utilizes its knowledge of age gains of all the devices, we require each device to adjust its access parameters for maximizing the estimated network \textit{expected AoI reduction} (EAR) per slot, which captures the essential for improving the contribution of the throughput to the AoI performance. To estimate the network EAR, we require each device to use Bayes' rule to keep a posteriori joint probability distribution of local age and age gain of an arbitrary device based on the channel observations. Numerical results validate our theoretical analysis and demonstrate the advantage of the proposed schemes over the existing schemes in a wide range of network configurations.

cs.IT↗

arXiv2024

Sub-Adjacent Transformer: Improving Time Series Anomaly Detection with Reconstruction Error from Sub-Adjacent Neighborhoods

In this paper, we present the Sub-Adjacent Transformer with a novel attention mechanism for unsupervised time series anomaly detection. Unlike previous approaches that rely on all the points within some neighborhood for time point reconstruction, our method restricts the attention to regions not immediately adjacent to the target points, termed sub-adjacent neighborhoods. Our key observation is that owing to the rarity of anomalies, they typically exhibit more pronounced differences from their sub-adjacent neighborhoods than from their immediate vicinities. By focusing the attention on the sub-adjacent areas, we make the reconstruction of anomalies more challenging, thereby enhancing their detectability. Technically, our approach concentrates attention on the non-diagonal areas of the attention matrix by enlarging the corresponding elements in the training stage. To facilitate the implementation of the desired attention matrix pattern, we adopt linear attention because of its flexibility and adaptability. Moreover, a learnable mapping function is proposed to improve the performance of linear attention. Empirically, the Sub-Adjacent Transformer achieves state-of-the-art performance across six real-world anomaly detection benchmarks, covering diverse fields such as server monitoring, space exploration, and water treatment.

cs.LG↗

arXiv2024

Electrically Switchable Circular Photogalvanic Effect in Methylammonium Lead Iodide Microcrystals

We investigate the circular photogalvanic effect (CPGE) in single-crystalline methylammonium lead iodide microcrystals under a static electric field. The external electric field can enhance the magnitude of the helicity dependent photocurrent (HDPC) by two orders of magnitude and flip its sign, which we attribute to magnetic shift currents induced by the Rashba-Edelstein effect. This HDPC induced by the static electric field may be viewed as an unusually strong third-order photoresponse, which produces a current two orders of magnitude larger than second-order injection current. Furthermore, the HDPC is highly nonlocal and can be created by photoexcitation out of the device channel, indicating a spin diffusion length up to 50 $μ$m at 78 K.

cond-mat.mes-hall↗

arXiv2023

Threshold KNN-Shapley: A Linear-Time and Privacy-Friendly Approach to Data Valuation

Data valuation aims to quantify the usefulness of individual data sources in training machine learning (ML) models, and is a critical aspect of data-centric ML research. However, data valuation faces significant yet frequently overlooked privacy challenges despite its importance. This paper studies these challenges with a focus on KNN-Shapley, one of the most practical data valuation methods nowadays. We first emphasize the inherent privacy risks of KNN-Shapley, and demonstrate the significant technical difficulties in adapting KNN-Shapley to accommodate differential privacy (DP). To overcome these challenges, we introduce TKNN-Shapley, a refined variant of KNN-Shapley that is privacy-friendly, allowing for straightforward modifications to incorporate DP guarantee (DP-TKNN-Shapley). We show that DP-TKNN-Shapley has several advantages and offers a superior privacy-utility tradeoff compared to naively privatized KNN-Shapley in discerning data quality. Moreover, even non-private TKNN-Shapley achieves comparable performance as KNN-Shapley. Overall, our findings suggest that TKNN-Shapley is a promising alternative to KNN-Shapley, particularly for real-world applications involving sensitive data.

cs.LG↗

arXiv2023

Age-of-Information Dependent Random Access for Periodic Updating

This paper considers an uplink Internet of Things system with synchronous periodic traffic, where multiple devices generate their status updates at the beginning of each global frame and attempt to send them to a common access point. To achieve a low network-wide age of information (AoI) in an easily implementable manner, we require each device to adopt an age-dependent random access protocol, i.e., to transmit with a certain probability only when its corresponding AoI reaches a certain threshold. We analyze the time-average expected AoI by a multi-layer Markov model where an external infinite-horizon Markov chain manages the jumps between the beginnings of frames, while two internal finite-horizon Markov chains manage the evolution during an arbitrary frame for different cases. Simulation results verify the accuracy of the modeling and the AoI advantage over age-independent schemes.

cs.IT↗

arXiv2023

"Private Prediction Strikes Back!'' Private Kernelized Nearest Neighbors with Individual Renyi Filter

Most existing approaches of differentially private (DP) machine learning focus on private training. Despite its many advantages, private training lacks the flexibility in adapting to incremental changes to the training dataset such as deletion requests from exercising GDPR's right to be forgotten. We revisit a long-forgotten alternative, known as private prediction, and propose a new algorithm named Individual Kernelized Nearest Neighbor (Ind-KNN). Ind-KNN is easily updatable over dataset changes and it allows precise control of the Rényi DP at an individual user level -- a user's privacy loss is measured by the exact amount of her contribution to predictions; and a user is removed if her prescribed privacy budget runs out. Our results show that Ind-KNN consistently improves the accuracy over existing private prediction methods for a wide range of $ε$ on four vision and language tasks. We also illustrate several cases under which Ind-KNN is preferable over private training with NoisySGD.

cs.LG↗