arXiv subjects

Yang Gao

Publications and source records attributed to Yang Gao.

At least 253 records · Page 14Linked to original sources

arXiv2025

Learning to Plan Before Answering: Self-Teaching LLMs to Learn Abstract Plans for Problem Solving

In the field of large language model (LLM) post-training, the effectiveness of utilizing synthetic data generated by the LLM itself has been well-presented. However, a key question remains unaddressed: what essential information should such self-generated data encapsulate? Existing approaches only produce step-by-step problem solutions, and fail to capture the abstract meta-knowledge necessary for generalization across similar problems. Drawing insights from cognitive science, where humans employ high-level abstraction to simplify complex problems before delving into specifics, we introduce a novel self-training algorithm: LEarning to Plan before Answering (LEPA). LEPA trains the LLM to formulate anticipatory plans, which serve as abstract meta-knowledge for problem-solving, before engaging with the intricacies of problems. This approach not only outlines the solution generation path but also shields the LLM from the distraction of irrelevant details. During data generation, LEPA first crafts an anticipatory plan based on the problem, and then generates a solution that aligns with both the plan and the problem. LEPA refines the plan through self-reflection, aiming to acquire plans that are instrumental in yielding correct solutions. During model optimization, the LLM is trained to predict both the refined plans and the corresponding solutions. By efficiently extracting and utilizing the anticipatory plans, LEPA demonstrates remarkable superiority over conventional algorithms on various challenging natural language reasoning benchmarks.

cs.CL↗

arXiv2025

Mamba-Sea: A Mamba-based Framework with Global-to-Local Sequence Augmentation for Generalizable Medical Image Segmentation

To segment medical images with distribution shifts, domain generalization (DG) has emerged as a promising setting to train models on source domains that can generalize to unseen target domains. Existing DG methods are mainly based on CNN or ViT architectures. Recently, advanced state space models, represented by Mamba, have shown promising results in various supervised medical image segmentation. The success of Mamba is primarily owing to its ability to capture long-range dependencies while keeping linear complexity with input sequence length, making it a promising alternative to CNNs and ViTs. Inspired by the success, in the paper, we explore the potential of the Mamba architecture to address distribution shifts in DG for medical image segmentation. Specifically, we propose a novel Mamba-based framework, Mamba-Sea, incorporating global-to-local sequence augmentation to improve the model's generalizability under domain shift issues. Our Mamba-Sea introduces a global augmentation mechanism designed to simulate potential variations in appearance across different sites, aiming to suppress the model's learning of domain-specific information. At the local level, we propose a sequence-wise augmentation along input sequences, which perturbs the style of tokens within random continuous sub-sequences by modeling and resampling style statistics associated with domain shifts. To our best knowledge, Mamba-Sea is the first work to explore the generalization of Mamba for medical image segmentation, providing an advanced and promising Mamba-based architecture with strong robustness to domain shifts. Remarkably, our proposed method is the first to surpass a Dice coefficient of 90% on the Prostate dataset, which exceeds previous SOTA of 88.61%. The code is available at https://github.com/orange-czh/Mamba-Sea.

cs.CV↗

arXiv2025

Observation and branching fraction measurement of the decay $J\!/\!ψ\rightarrow \bar{p} Σ^{+} K_{S}^{0} + c.c.$

The first observation of the decays $J\!/\!ψ\rightarrow \bar{p} Σ^{+} K_{S}^{0}$ and $J\!/\!ψ\rightarrow p \barΣ^{-} K_{S}^{0}$ is reported using $(10087\pm44)\times10^{6}$ $J\!/\!ψ$ events recorded by the BESIII detector at the BEPCII storage ring. The branching fractions of each channel are determined to be $\mathcal{B}(J\!/\!ψ\rightarrow \bar{p} Σ^{+} K_{S}^{0})=(1.361 \pm 0.006 \pm 0.025) \times 10^{-4}$ and $\mathcal{B}(J\!/\!ψ\rightarrow p \barΣ^{-} K_{S}^{0})=(1.352 \pm 0.006 \pm 0.025) \times 10^{-4}$. The combined result is $\mathcal{B}(J\!/\!ψ\rightarrow \bar{p} Σ^{+} K_{S}^{0} +c.c.)=(2.725 \pm 0.009 \pm 0.050) \times 10^{-4}$, where the first uncertainty is statistical and the second systematic. The results presented are in good agreement with the branching fractions of the isospin partner decay $J\!/\!ψ\rightarrow p K^- \barΣ^0 + c.c.$.

hep-ex↗

arXiv2025

Organization of Historical Oceanic Overturnings on Cross-Sphere Climate Signals

The global ocean meridional overturning circulation (GMOC) is central for ocean transport and climate variations. However, a comprehensive picture of its historical mean state and variability remains vague due to limitations in modelling and observing systems. Incorporating observations into models offers a viable approach to reconstructing climate history, yet achieving coherent estimates of GMOC has proven challenging due to difficulties in harmonizing ocean stratification. Here, we demonstrate that applying multiscale data assimilation scheme that integrates atmospheric and oceanic observations into multiple coupled models in a dynamically consistent way, the global ocean currents and GMOC over the past 80 years are retrieved. While the major historic events are printed in variability of the rebuilt GMOC, the timeseries of multisphere 3-dimensional physical variables representing the realistic historical evolution enable us to advance understanding of mechanisms of climate signal propagation cross spheres and give birth to Artificial Intelligence coupled big models, thus advancing the Earth science.

physics.ao-ph↗

arXiv2025

StyleRec: A Benchmark Dataset for Prompt Recovery in Writing Style Transformation

Prompt Recovery, reconstructing prompts from the outputs of large language models (LLMs), has grown in importance as LLMs become ubiquitous. Most users access LLMs through APIs without internal model weights, relying only on outputs and logits, which complicates recovery. This paper explores a unique prompt recovery task focused on reconstructing prompts for style transfer and rephrasing, rather than typical question-answering. We introduce a dataset created with LLM assistance, ensuring quality through multiple techniques, and test methods like zero-shot, few-shot, jailbreak, chain-of-thought, fine-tuning, and a novel canonical-prompt fallback for poor-performing cases. Our results show that one-shot and fine-tuning yield the best outcomes but highlight flaws in traditional sentence similarity metrics for evaluating prompt recovery. Contributions include (1) a benchmark dataset, (2) comprehensive experiments on prompt recovery strategies, and (3) identification of limitations in current evaluation metrics, all of which advance general prompt recovery research, where the structure of the input prompt is unrestricted.

cs.CL↗

arXiv2025

Balancing Multi-Target Semi-Supervised Medical Image Segmentation with Collaborative Generalist and Specialists

Despite the promising performance achieved by current semi-supervised models in segmenting individual medical targets, many of these models suffer a notable decrease in performance when tasked with the simultaneous segmentation of multiple targets. A vital factor could be attributed to the imbalanced scales among different targets: during simultaneously segmenting multiple targets, large targets dominate the loss, leading to small targets being misclassified as larger ones. To this end, we propose a novel method, which consists of a Collaborative Generalist and several Specialists, termed CGS. It is centered around the idea of employing a specialist for each target class, thus avoiding the dominance of larger targets. The generalist performs conventional multi-target segmentation, while each specialist is dedicated to distinguishing a specific target class from the remaining target classes and the background. Based on a theoretical insight, we demonstrate that CGS can achieve a more balanced training. Moreover, we develop cross-consistency losses to foster collaborative learning between the generalist and the specialists. Lastly, regarding their intrinsic relation that the target class of any specialized head should belong to the remaining classes of the other heads, we introduce an inter-head error detection module to further enhance the quality of pseudo-labels. Experimental results on three popular benchmarks showcase its superior performance compared to state-of-the-art methods.

cs.CV↗

arXiv2025

SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines

Large language models (LLMs) have demonstrated remarkable proficiency in mainstream academic disciplines such as mathematics, physics, and computer science. However, human knowledge encompasses over 200 specialized disciplines, far exceeding the scope of existing benchmarks. The capabilities of LLMs in many of these specialized fields-particularly in light industry, agriculture, and service-oriented disciplines-remain inadequately evaluated. To address this gap, we present SuperGPQA, a comprehensive benchmark that evaluates graduate-level knowledge and reasoning capabilities across 285 disciplines. Our benchmark employs a novel Human-LLM collaborative filtering mechanism to eliminate trivial or ambiguous questions through iterative refinement based on both LLM responses and expert feedback. Our experimental results reveal significant room for improvement in the performance of current state-of-the-art LLMs across diverse knowledge domains (e.g., the reasoning-focused model DeepSeek-R1 achieved the highest accuracy of 61.82% on SuperGPQA), highlighting the considerable gap between current model capabilities and artificial general intelligence. Additionally, we present comprehensive insights from our management of a large-scale annotation process, involving over 80 expert annotators and an interactive Human-LLM collaborative system, offering valuable methodological guidance for future research initiatives of comparable scope.

cs.CL↗

arXiv2025

Model-Based Offline Reinforcement Learning with Adversarial Data Augmentation

Model-based offline Reinforcement Learning (RL) constructs environment models from offline datasets to perform conservative policy optimization. Existing approaches focus on learning state transitions through ensemble models, rollouting conservative estimation to mitigate extrapolation errors. However, the static data makes it challenging to develop a robust policy, and offline agents cannot access the environment to gather new data. To address these challenges, we introduce Model-based Offline Reinforcement learning with AdversariaL data augmentation (MORAL). In MORAL, we replace the fixed horizon rollout by employing adversaria data augmentation to execute alternating sampling with ensemble models to enrich training data. Specifically, this adversarial process dynamically selects ensemble models against policy for biased sampling, mitigating the optimistic estimation of fixed models, thus robustly expanding the training data for policy optimization. Moreover, a differential factor is integrated into the adversarial process for regularization, ensuring error minimization in extrapolations. This data-augmented optimization adapts to diverse offline tasks without rollout horizon tuning, showing remarkable applicability. Extensive experiments on D4RL benchmark demonstrate that MORAL outperforms other model-based offline RL methods in terms of policy learning and sample efficiency.

cs.LG↗

arXiv2025

Cross section measurement of $e^+e^-\to ηψ(2S)$ and search for $e^+e^-\toη\tilde{X}(3872)$

The energy-dependent cross section for $e^+e^-\to ηψ(2S)$ is measured at eighteen center of mass energies from 4.288 GeV to 4.951 GeV using the BESIII detector. Using the same data samples, we also perform the first search for the reaction $e^+e^-\toη\tilde{X}(3872)$, but no evidence is found for the $\tilde{X}(3872)$ in the $π^+π^- J/ψ$ mass distribution. At each of the eighteen center of mass energies, upper limits at the 90\% confidence level on the cross section for $e^+e^-\toηψ(2S)$ and on the product of the $e^+e^-\toη\tilde{X}(3872)$ cross section with the branching fraction of $\tilde{X}(3872)\toπ^+π^- J/ψ$ are reported.

hep-ex↗

arXiv2025

Observation of $D^+\toη^\primeμ^+ν_μ$ and First Study of $D^+\to η^\prime \ell^+ν_\ell$ Decay Dynamics

Using $20.3\,\rm fb^{-1}$ of $e^+e^-$ collision data collected at the center-of-mass energy 3.773\,GeV with the BESIII detector, we report the first observation of the semileptonic decay $D^+\to η^\prime μ^+ν_μ$ with significance of $8.6σ$ including systematic uncertainties, and an improved measurement of $D^+\to η^\prime e^+ν_e$. The branching fractions of $D^+\to η^\prime μ^+ν_μ$ and $D^+\to η^\prime e^+ν_e$ are determined to be $(1.92\pm0.28_{\rm stat}\pm 0.08_{\rm syst})\times 10^{-4}$ and $(1.79\pm0.19_{\rm stat}\pm 0.07_{\rm syst})\times 10^{-4}$, respectively. The ratio of the two branching fractions is determined to be ${\mathcal R}_{μ/e}=1.07\pm0.19_{\rm stat}\pm 0.03_{\rm syst}$, which agrees with the theoretical expectation of lepton flavor universality within the Standard Model.From an analysis of the $D^+\to η^\prime \ell^+ν_\ell$ decay dynamics, the product of the hadronic form factor $f_+^{η^{\prime}}(0)$ and the CKM matrix element $|V_{cd}|$ is measured for the first time, giving $f^{η^\prime}_+(0)|V_{cd}| = (5.92\pm0.56_{\rm stat}\pm0.13_{\rm syst})\times 10^{-2}$. The $η-η^\prime$ mixing angle in the quark flavor basis is determined to be $ϕ_{\rm P} =(39.8\pm0.8_{\rm stat}\pm0.3_{\rm syst})^\circ$.

hep-ex↗

arXiv2025

Efficient Last-iterate Convergence Algorithms in Solving Games

To establish last-iterate convergence for Counterfactual Regret Minimization (CFR) algorithms in learning a Nash equilibrium (NE) of extensive-form games (EFGs), recent studies reformulate learning an NE of the original EFG as learning the NEs of a sequence of (perturbed) regularized EFGs. Consequently, proving last-iterate convergence in solving the original EFG reduces to proving last-iterate convergence in solving (perturbed) regularized EFGs. However, the empirical convergence rates of the algorithms in these studies are suboptimal, since they do not utilize Regret Matching (RM)-based CFR algorithms to solve perturbed EFGs, which are known the exceptionally fast empirical convergence rates. Additionally, since solving multiple perturbed regularized EFGs is required, fine-tuning across all such games is infeasible, making parameter-free algorithms highly desirable. In this paper, we prove that CFR$^+$, a classical parameter-free RM-based CFR algorithm, achieves last-iterate convergence in learning an NE of perturbed regularized EFGs. Leveraging CFR$^+$ to solve perturbed regularized EFGs, we get Reward Transformation CFR$^+$ (RTCFR$^+$). Importantly, we extend prior work on the parameter-free property of CFR$^+$, enhancing its stability, which is crucial for the empirical convergence of RTCFR$^+$. Experiments show that RTCFR$^+$ significantly outperforms existing algorithms with theoretical last-iterate convergence guarantees.

cs.GT↗

arXiv2025

MDTeamGPT: A Self-Evolving LLM-based Multi-Agent Framework for Multi-Disciplinary Team Medical Consultation

Large Language Models (LLMs) have made significant progress in various fields. However, challenges remain in Multi-Disciplinary Team (MDT) medical consultations. Current research enhances reasoning through role assignment, task decomposition, and accumulation of medical experience. Multi-role collaboration in MDT consultations often results in excessively long dialogue histories. This increases the model's cognitive burden and degrades both efficiency and accuracy. Some methods only store treatment histories. They do not extract effective experience or reflect on errors. This limits knowledge generalization and system evolution. We propose a multi-agent MDT medical consultation framework based on LLMs to address these issues. Our framework uses consensus aggregation and a residual discussion structure for multi-round consultations. It also employs a Correct Answer Knowledge Base (CorrectKB) and a Chain-of-Thought Knowledge Base (ChainKB) to accumulate consultation experience. These mechanisms enable the framework to evolve and continually improve diagnosis rationality and accuracy. Experimental results on the MedQA and PubMedQA datasets demonstrate that our framework achieves accuracies of 90.1% and 83.9%, respectively, and that the constructed knowledge bases generalize effectively across test sets from both datasets.

cs.AI↗

arXiv2025

GravMAD: Grounded Spatial Value Maps Guided Action Diffusion for Generalized 3D Manipulation

Robots' ability to follow language instructions and execute diverse 3D manipulation tasks is vital in robot learning. Traditional imitation learning-based methods perform well on seen tasks but struggle with novel, unseen ones due to variability. Recent approaches leverage large foundation models to assist in understanding novel tasks, thereby mitigating this issue. However, these methods lack a task-specific learning process, which is essential for an accurate understanding of 3D environments, often leading to execution failures. In this paper, we introduce GravMAD, a sub-goal-driven, language-conditioned action diffusion framework that combines the strengths of imitation learning and foundation models. Our approach breaks tasks into sub-goals based on language instructions, allowing auxiliary guidance during both training and inference. During training, we introduce Sub-goal Keypose Discovery to identify key sub-goals from demonstrations. Inference differs from training, as there are no demonstrations available, so we use pre-trained foundation models to bridge the gap and identify sub-goals for the current task. In both phases, GravMaps are generated from sub-goals, providing GravMAD with more flexible 3D spatial guidance compared to fixed 3D positions. Empirical evaluations on RLBench show that GravMAD significantly outperforms state-of-the-art methods, with a 28.63% improvement on novel tasks and a 13.36% gain on tasks encountered during training. Evaluations on real-world robotic tasks further show that GravMAD can reason about real-world tasks, associate them with relevant visual information, and generalize to novel tasks. These results demonstrate GravMAD's strong multi-task learning and generalization in 3D manipulation. Video demonstrations are available at: https://gravmad.github.io.

cs.RO↗

arXiv2025

Robust Dataset Distillation by Matching Adversarial Trajectories

Dataset distillation synthesizes compact datasets that enable models to achieve performance comparable to training on the original large-scale datasets. However, existing distillation methods overlook the robustness of the model, resulting in models that are vulnerable to adversarial attacks when trained on distilled data. To address this limitation, we introduce the task of ``robust dataset distillation", a novel paradigm that embeds adversarial robustness into the synthetic datasets during the distillation process. We propose Matching Adversarial Trajectories (MAT), a method that integrates adversarial training into trajectory-based dataset distillation. MAT incorporates adversarial samples during trajectory generation to obtain robust training trajectories, which are then used to guide the distillation process. As experimentally demonstrated, even through natural training on our distilled dataset, models can achieve enhanced adversarial robustness while maintaining competitive accuracy compared to existing distillation methods. Our work highlights robust dataset distillation as a new and important research direction and provides a strong baseline for future research to bridge the gap between efficient training and adversarial robustness.

cs.CV↗

arXiv2025

Measurement of Branching Fractions for $Λ_{c}^{+} \rightarrow n K_{S}^{0} π^{+}$ and $Λ_{c}^{+} \rightarrow n K_{S}^{0} K^{+}$

Based on 4.5 fb$^{-1}$ of $e^{+}e^{-}$ collision data accumulated at center-of-mass energies between $4.600\,\mathrm{GeV}$ and $4.699\,\mathrm{GeV}$ with the BESIII detector, we measure the absolute branching fraction of the Cabibbo-favored decay $Λ_{c}^{+} \rightarrow n K_{S}^{0} π^{+}$ with the precision improved by a factor of 2.8 and report the first evidence for the singly-Cabibbo-suppressed decay $Λ_{c}^{+} \rightarrow n K_{S}^{0} K^{+}$. The branching fractions for $Λ_{c}^{+} \rightarrow n K_{S}^{0} π^{+}$ and $Λ_{c}^{+} \rightarrow n K_{S}^{0} K^{+}$ are determined to be $(1.86\pm0.08\pm0.04)\times10^{-2}$ and $\left(3.9^{+1.7}_{-1.4}\pm0.3\right)\times10^{-4}$, respectively, where the first uncertainties are statistical and the second ones are systematic.

hep-ex↗

arXiv2025

FP3: A 3D Foundation Policy for Robotic Manipulation

Following its success in natural language processing and computer vision, foundation models that are pre-trained on large-scale multi-task datasets have also shown great potential in robotics. However, most existing robot foundation models rely solely on 2D image observations, ignoring 3D geometric information, which is essential for robots to perceive and reason about the 3D world. In this paper, we introduce FP3, a first large-scale 3D foundation policy model for robotic manipulation. FP3 builds on a scalable diffusion transformer architecture and is pre-trained on 60k trajectories with point cloud observations. With the model design and diverse pre-training data, FP3 can be efficiently fine-tuned for downstream tasks while exhibiting strong generalization capabilities. Experiments on real robots demonstrate that with only 80 demonstrations, FP3 is able to learn a new task with over 90% success rates in novel environments with unseen objects, significantly surpassing existing robot foundation models.

cs.RO↗

arXiv2025

Unified Human Localization and Trajectory Prediction with Monocular Vision

Conventional human trajectory prediction models rely on clean curated data, requiring specialized equipment or manual labeling, which is often impractical for robotic applications. The existing predictors tend to overfit to clean observation affecting their robustness when used with noisy inputs. In this work, we propose MonoTransmotion (MT), a Transformer-based framework that uses only a monocular camera to jointly solve localization and prediction tasks. Our framework has two main modules: Bird's Eye View (BEV) localization and trajectory prediction. The BEV localization module estimates the position of a person using 2D human poses, enhanced by a novel directional loss for smoother sequential localizations. The trajectory prediction module predicts future motion from these estimates. We show that by jointly training both tasks with our unified framework, our method is more robust in real-world scenarios made of noisy inputs. We validate our MT network on both curated and non-curated datasets. On the curated dataset, MT achieves around 12% improvement over baseline models on BEV localization and trajectory prediction. On real-world non-curated dataset, experimental results indicate that MT maintains similar performance levels, highlighting its robustness and generalization capability. The code is available at https://github.com/vita-epfl/MonoTransmotion.

cs.CV↗

arXiv2025

Extrapolation Merging: Keep Improving With Extrapolation and Merging

Large Language Models (LLMs) require instruction fine-tuning to perform different downstream tasks. However, the instruction fine-tuning phase still demands significant computational resources and labeled data, lacking a paradigm that can improve model performance without additional computational power and data. Model merging aims to enhance performance by combining the parameters of different models, but the lack of a clear optimization direction during the merging process does not always guarantee improved performance. In this paper, we attempt to provide a clear optimization direction for model merging. We first validate the effectiveness of the model extrapolation method during the instruction fine-tuning phase. Then, we propose Extrapolation Merging, a paradigm that can continue improving model performance without requiring extra computational resources or data. Using the extrapolation method, we provide a clear direction for model merging, achieving local optimization search, and consequently enhancing the merged model's performance. We conduct experiments on seven different tasks, and the results show that our method can consistently improve the model's performance after fine-tuning.

cs.CL↗