arXiv subjects

Xinyi Xie

Publications and source records attributed to Xinyi Xie.

7 recordsLinked to original sources

arXiv2026

Emotion2Skill: Model-Internal Emotion Signals for Adaptive Skill Selection and Evolution

Skill-based LLM agents select reusable procedures from an external library to solve complex tasks, yet their routing decisions rely entirely on text-level signals such as task descriptions, verbal reflections, and experience-derived rules, while the model's own internal representational state remains unobserved. Recent interpretability work has shown that LLMs maintain linear emotion representations that causally influence behavior; however, these representations have been exploited only for post-hoc analysis or direct output steering, and have not been used to inform agent-level decision-making. We propose Emotion2Skill, a framework that extracts LLM-internal emotion vectors and incorporates them into both skill selection and skill evolution. At each decision step, a 27-dimensional emotion state is extracted from the residual stream and mapped to a confidence-gated summary injected into the routing prompt. Beyond online selection, emotion trajectories are analyzed for abrupt internal-state shifts to pinpoint problematic skill invocations, guiding targeted SOP rewriting that replaces the coarse binary outcome signal of prior methods. On WebShop and ALFWorld, Emotion2Skill with Qwen3-8B improves over the Zero-Shot baseline by +26.9% success rate and +25.5% average success respectively, outperforming all baselines on both benchmarks with consistent gains on Qwen3-14B. Co-activation analysis further reveals semantically coherent emotion--skill pairings, confirming that the routing improvements reflect meaningful internal-state signals rather than opaque statistical correlations. These results establish LLM-internal emotion representations as an effective decision-level signal for orchestrating agent skill systems, extending their utility beyond interpretability and output steering. The code is available at https://github.com/BoHan-LIN04/Emotion2Skill.

cs.AI↗

arXiv2026

Look Before You Leap: Distilling Tree Search into Action Evaluation for Frozen VLA Models

Vision-Language-Action (VLA) models acquire broad embodied capabilities through large-scale pretraining, yet their generalization remains far more fragile than that of LLMs and VLMs. The prevailing remedy, post-training via supervised fine-tuning or reinforcement learning, improves task-specific performance but narrows the generalist capability that makes pretraining valuable. We identify a key bottleneck: VLA failures stem not only from action generation but also from action evaluation. A diagnostic pass@k study confirms that frozen VLAs already contain competent behaviors in their output distribution, with overall success rates rising from 33% at pass@1 to 92% at pass@32. Inspired by this, we propose SVA (Search, Value, and Act), a simple framework that equips frozen VLA policies with long-term consequence awareness. SVA first uses Monte-Carlo tree search in simulation to fully explore the VLA's output distribution and collect diverse trajectories annotated with empirical returns; this knowledge is then distilled into a lightweight Q-value model that predicts the expected consequence of candidate actions; at deployment, the frozen VLA proposes multiple candidates and the evaluator selects the one with the highest uncertainty-regularized Q-value, requiring no simulator access. By decoupling action proposal from consequence evaluation, SVA preserves the generalization capacity of the VLA backbone while substantially improving task success rates. Experiments across embodied benchmarks show that SVA consistently improves generalization on unseen tasks and exhibits strong test-time scaling behavior. Strikingly, SVA enables a 9B VLA to outperform a 27B VLA by 7 points at 27% lower inference latency, suggesting that scaling test-time evaluation is more cost-effective than scaling model size.

cs.RO↗

arXiv2025

Text-to-Decision Agent: Offline Meta-Reinforcement Learning from Natural Language Supervision

Offline meta-RL usually tackles generalization by inferring task beliefs from high-quality samples or warmup explorations. The restricted form limits their generality and usability since these supervision signals are expensive and even infeasible to acquire in advance for unseen tasks. Learning directly from the raw text about decision tasks is a promising alternative to leverage a much broader source of supervision. In the paper, we propose \textbf{T}ext-to-\textbf{D}ecision \textbf{A}gent (\textbf{T2DA}), a simple and scalable framework that supervises offline meta-RL with natural language. We first introduce a generalized world model to encode multi-task decision data into a dynamics-aware embedding space. Then, inspired by CLIP, we predict which textual description goes with which decision embedding, effectively bridging their semantic gap via contrastive language-decision pre-training and aligning the text embeddings to comprehend the environment dynamics. After training the text-conditioned generalist policy, the agent can directly realize zero-shot text-to-decision generation in response to language instructions. Comprehensive experiments on MuJoCo and Meta-World benchmarks show that T2DA facilitates high-capacity zero-shot generalization and outperforms various types of baselines. Our code is available at \textcolor{magenta}{\href{https://github.com/NJU-RL/T2DA}{https://github.com/NJU-RL/T2DA}}.

cs.AI↗

arXiv2025

Diagnostic performance of echocardiography in detecting and differentiating cardiac amyloidosis: a meta-analysis

Aims: This meta-analysis aimed to evaluate the diagnostic performance of echocardiographic parameters for cardiac amyloidosis (CA), with a focus on subtype stratification and comparisons with healthy controls. Methods and Results: A comprehensive search identified 26 studies published before February 2025, encompassing 3,802 patients. Compared to healthy individuals, CA patients demonstrated significant echocardiographic abnormalities, including reduced left ventricular ejection fraction (LVEF; WMD = -10.65, 95% CI: [-11.84, -9.46]), increased left atrial volume index (WMD = +15.87, 95% CI: [14.35, 17.38]), and thickened posterior wall (WMD = +5.14, 95% CI: [4.85, 5.42]). Subtype analyses revealed that transthyretin cardiac amyloidosis (ATTR-CA) was associated with more pronounced systolic dysfunction than light-chain cardiac amyloidosis (AL-CA), evidenced by lower global longitudinal strain (WMD = -2.02, 95% CI: [-2.66, -1.37]), reduced LVEF (WMD = -5.31, 95% CI: [-6.63, -3.99]), and diminished tricuspid annular plane systolic excursion (WMD = -1.59, 95% CI: [-2.23, -0.95]). Additionally, ATTR-CA patients exhibited greater ventricular wall thickening in both posterior wall (WMD = +1.87, 95% CI: [1.51, 2.23]) and interventricular septum (WMD = +2.24, 95% CI: [1.85, 2.63]). Conclusion: Echocardiography plays a pivotal role in diagnosing CA and distinguishing between AL-CA and ATTR-CA. Key indices such as LVEF and global longitudinal strain are especially valuable for early detection, while subtype-specific patterns highlight distinct underlying pathophysiologies, offering guidance for tailored diagnostic and therapeutic strategies.

physics.med-ph↗

arXiv2025

VidTwin: Video VAE with Decoupled Structure and Dynamics

Recent advancements in video autoencoders (Video AEs) have significantly improved the quality and efficiency of video generation. In this paper, we propose a novel and compact video autoencoder, VidTwin, that decouples video into two distinct latent spaces: Structure latent vectors, which capture overall content and global movement, and Dynamics latent vectors, which represent fine-grained details and rapid movements. Specifically, our approach leverages an Encoder-Decoder backbone, augmented with two submodules for extracting these latent spaces, respectively. The first submodule employs a Q-Former to extract low-frequency motion trends, followed by downsampling blocks to remove redundant content details. The second averages the latent vectors along the spatial dimension to capture rapid motion. Extensive experiments show that VidTwin achieves a high compression rate of 0.20% with high reconstruction quality (PSNR of 28.14 on the MCL-JCV dataset), and performs efficiently and effectively in downstream generative tasks. Moreover, our model demonstrates explainability and scalability, paving the way for future research in video latent representation and generation. Check our project page for more details: https://vidtwin.github.io/.

cs.CV↗

arXiv2024

Can bidirectional encoder become the ultimate winner for downstream applications of foundation models?

Over the past few decades, Artificial Intelligence(AI) has progressed from the initial machine learning stage to the deep learning stage, and now to the stage of foundational models. Foundational models have the characteristics of pre-training, transfer learning, and self-supervised learning, and pre-trained models can be fine-tuned and applied to various downstream tasks. Under the framework of foundational models, models such as Bidirectional Encoder Representations from Transformers(BERT) and Generative Pre-trained Transformer(GPT) have greatly advanced the development of natural language processing(NLP), especially the emergence of many models based on BERT. BERT broke through the limitation of only using one-way methods for language modeling in pre-training by using a masked language model. It can capture bidirectional context information to predict the masked words in the sequence, this can improve the feature extraction ability of the model. This makes the model very useful for downstream tasks, especially for specialized applications. The model using the bidirectional encoder can better understand the domain knowledge and be better applied to these downstream tasks. So we hope to help understand how this technology has evolved and improved model performance in various natural language processing tasks under the background of foundational models and reveal its importance in capturing context information and improving the model's performance on downstream tasks. This article analyzes one-way and bidirectional models based on GPT and BERT and compares their differences based on the purpose of the model. It also briefly analyzes BERT and the improvements of some models based on BERT. The model's performance on the Stanford Question Answering Dataset(SQuAD) and General Language Understanding Evaluation(GLUE) was compared.

cs.CL↗

arXiv2022

Positive effects of multiplicative noise on the explosion of nonlinear fractional stochastic differential equations

For the nonlinear stochastic partial differential equation which is driven by multiplicative noise of the form \[D_t^βu = \left[ { - {{\left( { - Δ} \right)}^s}u + ζ\left( u \right)} \right]dt + A\sum\limits_{m \in Z_0^d} {\sum\limits_{j = 1}^{d - 1} {{θ_m}{σ_{m,j}}\left( x \right)} } \circ dW_t^{m,j},\;\; s \ge 1,\;\;\frac{1}{2} < β< 1,\] where $D_{t}^β$ denotes the Caputo derivative, $A>0$ is a constant depending on the noise intensity, $\circ$ represent the Stratonovich-type stochastic differential, we consider the blow-up time of its solutions. We find that the introduction of noise can effectively delay the blow-up time of the solution to the deterministic differential equation when $ζ$ in the above equation satisfies some assumptions. A key element in our construction is using the Galerkin approximation and a priori estimates methods to prove the existence and uniqueness of the solutions to the above stochastic equations, which can be regarded as the fractional order extension of the conclusions in \cite{flandoli2021delayed}. We also verify the validation of hypotheses in the time fractional Keller-Segel and time fractional Fisher-KPP equations in 3D case.

math.PR↗