arXiv subjects

Xu Dong

Publications and source records attributed to Xu Dong.

At least 19 recordsLinked to original sources

arXiv2026

BoT-Feedback: Grounding Multimodal Reasoning in Biomechanical Evidence for Explainable Human Action Feedback

Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities in visual understanding and multimodal reasoning, yet they remain fundamentally limited in Human Action Feedback Generation. Existing methods infer coaching feedback directly from visual observations, producing generic advice, limited interpretability, and physically implausible hallucinations. In contrast, expert human coaches diagnose performance through explicit biomechanical reasoning over joint kinematics, posture, and body dynamics. We introduce BoT-Feedback, a framework that grounds MLLM reasoning in structured biomechanical evidence. Our key contribution is Biomechanics of Thought (BoT), a four-stage reasoning framework that progressively identifies the action, localises the critical body regions, analyses quantitative biomechanical differences between expert and student performances, and synthesises interpretable coaching feedback. To support this reasoning process, we develop a plug-and-play Biomechanical Data Parser (BDP) that converts videos into structured biomechanical descriptors and an alignment strategy that temporally matches expert and student motions. We further introduce BiomAF, a benchmark containing paired teacher-student videos, 3D skeletons, biomechanical attributes, and expert-coaching annotations. Experiments across twelve open- and closed-source MLLMs demonstrate that grounding reasoning in biomechanical evidence consistently improves feedback quality, interpretability, and robustness while substantially reducing biomechanical hallucinations. BoT-Feedback improves the average expert evaluation score from 2.07 to 2.95 (+40%), enabling compact open-source MLLMs to approach the performance of substantially larger proprietary systems for explainable action feedback generation.

cs.CV↗

arXiv2026

Aero Hand Open: A Simulation-Ready Tendon-Driven Hand for Dexterous Manipulation Learning

Tendon-driven hands are anthropomorphic, and moving the actuators off the joints is what makes a hand of this capability affordable to build. Two effects produce that saving. Routing force through a cable removes the requirement that a motor fit inside the joint it drives, so smaller and cheaper motors suffice, and one motor can drive several joints through a single cable, so fewer motors are needed. They are also harder to learn on than a direct-drive hand. The underactuated transmission that produces the saving is itself difficult to represent in a simulator, and the joints one cable drives are not independently commandable. We present Aero Hand Open, a tendon-driven anthropomorphic hand that is released simulation-ready. Three things ship with it. A simulation model reproduces the cable transmission itself. An identified actuation map connects that model to the motor commands in both directions, including the three-way coupling of the thumb. A reinforcement learning package trains policies for the hand. Together they let a policy be trained entirely in simulation and run on the hand with no fine-tuning and no state estimation. We release the mechanical design, the simulation model, the identified mapping, the training environment and the deployment stack.

cs.RO↗

arXiv2026

Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale

Embedding-based retrieval on user-generated content at the trillion-document scale exposes a sharp conflict between two production demands: deep, expressive personalization for queries with rich user intent, and broad coverage of a massive inventory under fixed latency and resource budgets. We characterize this as the personalization-scale paradox: hosting the full serving inventory in GPU memory is too resource intensive, while CPU compute cannot execute the same interaction-heavy model on the latency-critical path. We present a hybrid GPU-CPU co-serving system that resolves the paradox through orchestration rather than a new model class. A high-depth GPU pathway fuses retrieval and interaction pre-ranking over a curated online pool on the order of a billion documents, while a high-breadth CPU pathway searches an independently selected online inventory roughly twenty times larger with lightweight personalized scoring. Either or both pathways can run per request; candidates are deduplicated before shared downstream ranking. The system is deployed in production. A full-system A/B test against the legacy CPU-only configuration improves model-scored relevance and substantive engagement, while separate pathway experiments show positive value at their own deployment scopes. Retrieval logs show that the pathways contribute structurally distinct candidates, production serving measurements characterize their latency, and a matched capacity plan quantifies the economic rationale for assigning modeling depth to GPUs and inventory breadth to CPUs. Together, these results validate a practical, independently evolvable depth-breadth architecture for ultra-large-scale personalized search.

cs.IR↗

arXiv2026

Moving Beyond More Views: Redundancy-Aware Ego-Exo Fusion for Proficiency Estimation

EgoExo proficiency estimation aims to assess action quality by integrating fine-grained motion cues from egocentric (1st-person) views with spatial context from multiple exocentric (3rd-person) views. Simply adding more exocentric views degrades EgoExo performance, as redundant or noisy perspectives dilute useful motion cues. Our analysis identifies two key causes: (1) Multiview redundancy - From the data perspective, certain views provide limited or noisy information, diluting discriminative cues; (2) Overfitting - From the feature perspective, conventional fusion increases representational complexity, causing the model to memorise view-specific patterns rather than learn generalisable representations. To address these issues, we propose two complementary modules: AdaMVS, which adaptively identifies and fuses the most informative view tokens under weak supervision from the data perspective, and VIB-GB, which combines Gradient Blending and Variational Information Bottleneck regularisation from the feature perspective to compress redundant signals and suppress overfitting during training. Experiments on EgoExo-4D and EgoExo-Fitness demonstrate that our method learns both which view to look at and how to fuse them, achieving new state-of-the-art results. Our source code is available at https://github.com/dx199771/AdaMVS

cs.CV↗

arXiv2026

Size and solvent effects on cellulose transverse anisotropy and toughening design

Cellulose nanocrystals (CNCs) are a promising class of materials derived from the most abundant natural polymer resource on Earth. Hydroxyl-induced polarity is a crucial advantage of CNCs, making them promising for advanced design and application. Side chain hydroxyls, hydrogen bonds, and particular crystal structures of CNCs lead to unique anisotropy. However, the nuanced anisotropy in the transverse section is not sufficiently stressed, and cannot be precisely described experimentally. Although partially covered by previous studies, a quantitative explanation of size dependency and a systematic comparison of solvent influences are still lacking. The manufacturing of cellulose materials requires a better understanding of anisotropy, size dependency, and solvent influences. In this study, the anisotropic performance of CNCs in characteristic directions and a diverse array of solvent environments was carefully inspected and compared using molecular simulations. Furthermore, a data-supported explanation for the size dependency, and transverse arrangement toughness-enhanced designs were both proposed. These systematic comparisons and unique transverse arrangements could aid future applications of cellulose.

physics.comp-ph↗

arXiv2026

PEML: Parameter-efficient Multi-Task Learning with Optimized Continuous Prompts

Parameter-Efficient Fine-Tuning (PEFT) is widely used for adapting Large Language Models (LLMs) for various tasks. Recently, there has been an increasing demand for fine-tuning a single LLM for multiple tasks because it requires overall less data for fine-tuning thanks to the common features shared among tasks. More importantly, LLMs are resource demanding and deploying a single model for multiple tasks facilitates resource consolidation and consumes significantly less resources compared to deploying individual large model for each task. Existing PEFT methods like LoRA and Prefix Tuning are designed to adapt LLMs to a specific task. LoRA and its variation focus on aligning the model itself for tasks, overlooking the importance of prompt tuning in multi-task learning while Prefix Tuning only adopts a simple architecture to optimize prompts, which limits the adaption capabilities for multi-task. To enable efficient fine-tuning for multi-task learning, it is important to co-optimize prompt optimization and model adaptation. In this work, we propose a Parameter-Efficient Multi-task Learning (\PM), which employs a neural architecture engineering method for optimizing the continuous prompts while also performing low-rank adaption for model weights. We prototype PEML by creating an automated framework for optimizing the continuous prompts and adapting model weights. We evaluate PEML against state-of-the-arts multi-task learning methods MTL-LoRA, MultiLoRa, C-Poly, and MoE, on the GLUE, SuperGLUE, Massive Multitask Language Understanding, and commonsense reasoning benchmarks. The evaluation results present an average accuracy improvement of up to 6.67%, with individual tasks showing peak gains of up to 10.75%.

cs.CL↗

arXiv2026

Analytical coarse grained potential parameterization by Reinforcement Learning for anisotropic cellulose

Cellulose nanocrystals (CNCs) are a type of cellulose with excellent mechanical performance and other merit attributes. According to previous reports, hydrogen bonds play a pivotal role in the anisotropic structure of the CNC. Understanding the structure and mechanical behavior of CNC on a mesoscopic scale is critical for the development and manufacture of cellulose materials. However, experimental observations and atomistic simulations are not appropriate on the mesoscopic scale. In this study, we introduce an analytical coarse-grained (CG) potential following an extended bottom-up approach that is directly parameterized using Reinforcement Learning (RL). RL is a powerful tool for industrial and academic applications in various fields. Nevertheless, the potential of RL has not yet been fully exploited in the field of molecular dynamics. The RL and Boltzmann inversion methods were employed to develop a novel CG model of cellulose to represent its anisotropy and polymer stiffness. The resultant CG model is not limited to the target properties for training, and can reproduce the dynamics mechanical properties under other circumstances without additional training. This model confirms that RL can construct a CG potential that is both physically explainable and powerful.

physics.comp-ph↗

arXiv2026

Generative Data Augmentation for Skeleton Action Recognition

Skeleton-based human action recognition is a powerful approach for understanding human behaviour from pose data, but collecting large-scale, diverse, and well-annotated 3D skeleton datasets is both expensive and labor-intensive. To address this challenge, we propose a conditional generative pipeline for data augmentation in skeleton action recognition. Our method learns the distribution of real skeleton sequences under the constraint of action labels, enabling the synthesis of diverse and high-fidelity data. Even with limited training samples, it can effectively generate skeleton sequences and achieve competitive recognition performance in low-data scenarios, demonstrating strong generalisation in downstream tasks. Specifically, we introduce a Transformer-based encoder-decoder architecture, combined with a generative refinement module and a dropout mechanism, to balance fidelity and diversity during sampling. Experiments on HumanAct12 and the refined NTU-RGBD (NTU-VIBE) dataset show that our approach consistently improves the accuracy of multiple skeleton-based action recognition models, validating its effectiveness in both few-shot and full-data settings. The source code can be found at here.

cs.CV↗

arXiv2026

FadeMem: Biologically-Inspired Forgetting for Efficient Agent Memory

Large language models deployed as autonomous agents face critical memory limitations, lacking selective forgetting mechanisms that lead to either catastrophic forgetting at context boundaries or information overload within them. While human memory naturally balances retention and forgetting through adaptive decay processes, current AI systems employ binary retention strategies that preserve everything or lose it entirely. We propose FadeMem, a biologically-inspired agent memory architecture that incorporates active forgetting mechanisms mirroring human cognitive efficiency. FadeMem implements differential decay rates across a dual-layer memory hierarchy, where retention is governed by adaptive exponential decay functions modulated by semantic relevance, access frequency, and temporal patterns. Through LLM-guided conflict resolution and intelligent memory fusion, our system consolidates related information while allowing irrelevant details to fade. Experiments on Multi-Session Chat, LoCoMo, and LTI-Bench demonstrate superior multi-hop reasoning and retrieval with 45\% storage reduction, validating the effectiveness of biologically-inspired forgetting in agent memory systems.

cs.AI↗

arXiv2025

Diffusion-Based, Data-Assimilation-Enabled Super-Resolution of Hub-height Winds

High-quality observations of hub-height winds are valuable but sparse in space and time. Simulations are widely available on regular grids but are generally biased and too coarse to inform wind-farm siting or to assess extreme-weather-related risks (e.g., gusts) at infrastructure scales. To fully utilize both data types for generating high-quality, high-resolution hub-height wind speeds (tens to ~100m above ground), this study introduces WindSR, a diffusion model with data assimilation for super-resolution downscaling of hub-height winds. WindSR integrates sparse observational data with simulation fields during downscaling using state-of-the-art diffusion models. A dynamic-radius blending method is introduced to merge observations with simulations, providing conditioning for the diffusion process. Terrain information is incorporated during both training and inference to account for its role as a key driver of winds. Evaluated against convolutional-neural-network and generative-adversarial-network baselines, WindSR outperforms them in both downscaling efficiency and accuracy. Our data assimilation reduces WindSR's model bias by approximately 20% relative to independent observations.

cs.LG↗

arXiv2025

GCDance: Genre-Controlled Music-Driven 3D Full Body Dance Generation

Music-driven dance generation is a challenging task as it requires strict adherence to genre-specific choreography while ensuring physically realistic and precisely synchronized dance sequences with the music's beats and rhythm. Although significant progress has been made in music-conditioned dance generation, most existing methods struggle to convey specific stylistic attributes in generated dance. To bridge this gap, we propose a diffusion-based framework for genre-specific 3D full-body dance generation, conditioned on both music and descriptive text. To effectively incorporate genre information, we develop a text-based control mechanism that maps input prompts, either explicit genre labels or free-form descriptive text, into genre-specific control signals, enabling precise and controllable text-guided generation of genre-consistent dance motions. Furthermore, to enhance the alignment between music and textual conditions, we leverage the features of a music foundation model, facilitating coherent and semantically aligned dance synthesis. Last, to balance the objectives of extracting text-genre information and maintaining high-quality generation results, we propose a novel multi-task optimization strategy. This effectively balances competing factors such as physical realism, spatial accuracy, and text classification, significantly improving the overall quality of the generated sequences. Extensive experimental results obtained on the FineDance and AIST++ datasets demonstrate the superiority of GCDance over the existing state-of-the-art approaches.

cs.GR↗

arXiv2025

EP241021a: a months-duration X-ray transient with luminous optical and radio emission

We present the discovery of a peculiar X-ray transient, EP241021a, by the Einstein Probe (EP) mission, and the results from multiwavelength follow-up observations. The transient was first detected with the Wide-field X-ray Telescope as an intense flare lasting for ~100 s, reaching a luminosity of L_(0.5-4 keV)~10^48 erg/s at z=0.748. Further observations with EP's Follow-up X-ray Telescope reveal a huge drop in the X-ray flux by a factor of >1000 within 1.5 days. After maintaining a nearly plateau phase for ~7 days, the X-ray flux declines as t^-1.2 over a period of ~30 days, followed by a sudden decrease to an undetectable level by EP and XMM-Newton, making it the longest afterglow emission detected among known fast X-ray transients. A bright counterpart at optical and radio wavelengths was also detected, with high peak luminosities in excess of 10^44 erg/s and 10^41 erg/s, respectively. In addition, EP241021a exhibits a non-thermal X-ray spectrum, red optical color, X-ray and optical rebrightenings in the light curves, and fast radio spectral evolution, suggesting that relativistic jets may have been launched. We discuss possible origins of EP241021a, including a choked jet with supernova shock breakout, a merger-triggered magnetar, a highly structured jet, and a repeating partial tidal disruption event involving an intermediate-mass black hole, but none can perfectly explain the multiwavelength properties. EP241021a may represent a new type of X-ray transients with months-duration evolution timescales, and future EP detections and follow-up observations of similar systems will provide statistical samples to understand the underlying mechanisms at work.

astro-ph.HE↗

arXiv2024

Existence of multiple radial solutions for nonlinear equation involving the mean curvature operator in Lorentz-Minkowski space

We prove existence of multiple radial solutions to the Dirichlet problem for nonlinear equations involving the mean curvature operator in Lorentz-Minkowski space and a nonlinear term of concave-convex type. Solutions are found using Szulkin's critical point theory for non-smooth functional. Multiplicity results are also given for some cases in which the nonlinearity depends also on the gradient of the solution.

math.AP↗

arXiv2024

Interpretable Long-term Action Quality Assessment

Long-term Action Quality Assessment (AQA) evaluates the execution of activities in videos. However, the length presents challenges in fine-grained interpretability, with current AQA methods typically producing a single score by averaging clip features, lacking detailed semantic meanings of individual clips. Long-term videos pose additional difficulty due to the complexity and diversity of actions, exacerbating interpretability challenges. While query-based transformer networks offer promising long-term modeling capabilities, their interpretability in AQA remains unsatisfactory due to a phenomenon we term Temporal Skipping, where the model skips self-attention layers to prevent output degradation. To address this, we propose an attention loss function and a query initialization method to enhance performance and interpretability. Additionally, we introduce a weight-score regression module designed to approximate the scoring patterns observed in human judgments and replace conventional single-score regression, improving the rationality of interpretability. Our approach achieves state-of-the-art results on three real-world, long-term AQA benchmarks. Our code is available at: https://github.com/dx199771/Interpretability-AQA

cs.CV↗

arXiv2023

on some discrete Bonnesen-style isoperimetric inequalities

This article deals with the sharp discrete isoperimetric inequalities in analysis and geometry for planar convex polygons. First, the analytic isoperimetric inequalities based on Schur convex function are established. In the wake of the analytic isoperimetric inequalities, Bonnesen-style isoperimetric inequalities and inverse Bonnesen-style inequalities for the planar convex polygons are obtained.

math.DG↗

arXiv2022

The log-Minkowski inequality of curvature entropy for non-symmetric convex bodies

In an earlier paper \cite{mazeng} the authors introduced the notion of curvature entropy, and proved the plane log-Minkowski inequality of curvature entropy under the symmetry assumption. In this paper we demonstrate the plane log-Minkowski inequality of curvature entropy for general convex bodies. The equivalence of the uniqueness of cone-volume measure, the log-Minkowski inequality of volume, and the log-Minkowski inequality of curvature entropy for general convex bodies in $\mathbb R^{2}$ are shown.

math.MG↗

arXiv2021

Radar Camera Fusion via Representation Learning in Autonomous Driving

Radars and cameras are mature, cost-effective, and robust sensors and have been widely used in the perception stack of mass-produced autonomous driving systems. Due to their complementary properties, outputs from radar detection (radar pins) and camera perception (2D bounding boxes) are usually fused to generate the best perception results. The key to successful radar-camera fusion is the accurate data association. The challenges in the radar-camera association can be attributed to the complexity of driving scenes, the noisy and sparse nature of radar measurements, and the depth ambiguity from 2D bounding boxes. Traditional rule-based association methods are susceptible to performance degradation in challenging scenarios and failure in corner cases. In this study, we propose to address radar-camera association via deep representation learning, to explore feature-level interaction and global reasoning. Additionally, we design a loss sampling mechanism and an innovative ordinal loss to overcome the difficulty of imperfect labeling and to enforce critical human-like reasoning. Despite being trained with noisy labels generated by a rule-based algorithm, our proposed method achieves a performance of 92.2% F1 score, which is 11.6% higher than the rule-based teacher. Moreover, this data-driven method also lends itself to continuous improvement via corner case mining.

cs.CV↗

arXiv2020

Probabilistic Oriented Object Detection in Automotive Radar

Autonomous radar has been an integral part of advanced driver assistance systems due to its robustness to adverse weather and various lighting conditions. Conventional automotive radars use digital signal processing (DSP) algorithms to process raw data into sparse radar pins that do not provide information regarding the size and orientation of the objects. In this paper, we propose a deep-learning based algorithm for radar object detection. The algorithm takes in radar data in its raw tensor representation and places probabilistic oriented bounding boxes around the detected objects in bird's-eye-view space. We created a new multimodal dataset with 102544 frames of raw radar and synchronized LiDAR data. To reduce human annotation effort we developed a scalable pipeline to automatically annotate ground truth using LiDAR as reference. Based on this dataset we developed a vehicle detection pipeline using raw radar data as the only input. Our best performing radar detection model achieves 77.28\% AP under oriented IoU of 0.3. To the best of our knowledge, this is the first attempt to investigate object detection with raw radar data for conventional corner automotive radars.

cs.CV↗