arXiv subjects

Yi Yang

Publications and source records attributed to Yi Yang.

At least 397 records · Page 22Linked to original sources

arXiv2025

TarPro: Targeted Protection against Malicious Image Editing

The rapid advancement of image editing techniques has raised concerns about their misuse for generating Not-Safe-for-Work (NSFW) content. This necessitates a targeted protection mechanism that blocks malicious edits while preserving normal editability. However, existing protection methods fail to achieve this balance, as they indiscriminately disrupt all edits while still allowing some harmful content to be generated. To address this, we propose TarPro, a targeted protection framework that prevents malicious edits while maintaining benign modifications. TarPro achieves this through a semantic-aware constraint that only disrupts malicious content and a lightweight perturbation generator that produces a more stable, imperceptible, and robust perturbation for image protection. Extensive experiments demonstrate that TarPro surpasses existing methods, achieving a high protection efficacy while ensuring minimal impact on normal edits. Our results highlight TarPro as a practical solution for secure and controlled image editing.

cs.CR↗

arXiv2025

The Virasoro symmetries of the bigraded modified Toda hierarchy

Modified Toda hierarchy is a two-component generalization of the 1st modified KP hierarchy, which has been widely applied to analyze constraints of the Toda hierarchy, including the B--Toda and C--Toda hierarchies. In this paper, we construct additional symmetries for the modified Toda hierarchy and derive the corresponding Adler-Shiota-van Moerbeke formula. In addition, we also show that $(N,M)$--bigraded modified Toda hierarchy(BMTH), which is a special reduction for modified Toda, possess additional symmetries that form a subalgebra of the Virasoro algebra.

nlin.SI↗

arXiv2025

Bridging the LLM Accessibility Divide? Performance, Fairness, and Cost of Closed versus Open LLMs for Automated Essay Scoring

Closed large language models (LLMs) such as GPT-4 have set state-of-the-art results across a number of NLP tasks and have become central to NLP and machine learning (ML)-driven solutions. Closed LLMs' performance and wide adoption has sparked considerable debate about their accessibility in terms of availability, cost, and transparency. In this study, we perform a rigorous comparative analysis of nine leading LLMs, spanning closed, open, and open-source LLM ecosystems, across text assessment and generation tasks related to automated essay scoring. Our findings reveal that for few-shot learning-based assessment of human generated essays, open LLMs such as Llama 3 and Qwen2.5 perform comparably to GPT-4 in terms of predictive performance, with no significant differences in disparate impact scores when considering age- or race-related fairness. Moreover, Llama 3 offers a substantial cost advantage, being up to 37 times more cost-efficient than GPT-4. For generative tasks, we find that essays generated by top open LLMs are comparable to closed LLMs in terms of their semantic composition/embeddings and ML assessed scores. Our findings challenge the dominance of closed LLMs and highlight the democratizing potential of open LLMs, suggesting they can effectively bridge accessibility divides while maintaining competitive performance and fairness.

cs.CL↗

arXiv2025

Chart-HQA: A Benchmark for Hypothetical Question Answering in Charts

Multimodal Large Language Models (MLLMs) have garnered significant attention for their strong visual-semantic understanding. Most existing chart benchmarks evaluate MLLMs' ability to parse information from charts to answer questions. However, they overlook the inherent output biases of MLLMs, where models rely on their parametric memory to answer questions rather than genuinely understanding the chart content. To address this limitation, we introduce a novel Chart Hypothetical Question Answering (HQA) task, which imposes assumptions on the same question to compel models to engage in counterfactual reasoning based on the chart content. Furthermore, we introduce HAI, a human-AI interactive data synthesis approach that leverages the efficient text-editing capabilities of LLMs alongside human expert knowledge to generate diverse and high-quality HQA data at a low cost. Using HAI, we construct Chart-HQA, a challenging benchmark synthesized from publicly available data sources. Evaluation results on 18 MLLMs of varying model sizes reveal that current models face significant generalization challenges and exhibit imbalanced reasoning performance on the HQA task.

cs.CL↗

arXiv2025

DoraemonGPT: Toward Understanding Dynamic Scenes with Large Language Models (Exemplified as A Video Agent)

Recent LLM-driven visual agents mainly focus on solving image-based tasks, which limits their ability to understand dynamic scenes, making it far from real-life applications like guiding students in laboratory experiments and identifying their mistakes. Hence, this paper explores DoraemonGPT, a comprehensive and conceptually elegant system driven by LLMs to understand dynamic scenes. Considering the video modality better reflects the ever-changing nature of real-world scenarios, we exemplify DoraemonGPT as a video agent. Given a video with a question/task, DoraemonGPT begins by converting the input video into a symbolic memory that stores task-related attributes. This structured representation allows for spatial-temporal querying and reasoning by well-designed sub-task tools, resulting in concise intermediate results. Recognizing that LLMs have limited internal knowledge when it comes to specialized domains (e.g., analyzing the scientific principles underlying experiments), we incorporate plug-and-play tools to assess external knowledge and address tasks across different domains. Moreover, a novel LLM-driven planner based on Monte Carlo Tree Search is introduced to explore the large planning space for scheduling various tools. The planner iteratively finds feasible solutions by backpropagating the result's reward, and multiple solutions can be summarized into an improved final answer. We extensively evaluate DoraemonGPT's effectiveness on three benchmarks and several in-the-wild scenarios. The code will be released at https://github.com/z-x-yang/DoraemonGPT.

cs.CV↗

arXiv2025

GaussianGraph: 3D Gaussian-based Scene Graph Generation for Open-world Scene Understanding

Recent advancements in 3D Gaussian Splatting(3DGS) have significantly improved semantic scene understanding, enabling natural language queries to localize objects within a scene. However, existing methods primarily focus on embedding compressed CLIP features to 3D Gaussians, suffering from low object segmentation accuracy and lack spatial reasoning capabilities. To address these limitations, we propose GaussianGraph, a novel framework that enhances 3DGS-based scene understanding by integrating adaptive semantic clustering and scene graph generation. We introduce a "Control-Follow" clustering strategy, which dynamically adapts to scene scale and feature distribution, avoiding feature compression and significantly improving segmentation accuracy. Additionally, we enrich scene representation by integrating object attributes and spatial relations extracted from 2D foundation models. To address inaccuracies in spatial relationships, we propose 3D correction modules that filter implausible relations through spatial consistency verification, ensuring reliable scene graph construction. Extensive experiments on three datasets demonstrate that GaussianGraph outperforms state-of-the-art methods in both semantic segmentation and object grounding tasks, providing a robust solution for complex scene understanding and interaction.

cs.CV↗

arXiv2025

EP240801a/XRF 240801B: An X-ray Flash Detected by the Einstein Probe and Implications of its Multiband Afterglow

We present multiband observations and analysis of EP240801a, a low-energy, extremely soft gamma-ray burst (GRB) discovered on August 1, 2024 by the Einstein Probe (EP) satellite, with a weak contemporaneous signal also detected by Fermi/GBM. Optical spectroscopy of the afterglow, obtained by GTC and Keck, identified the redshift of $z = 1.6734$. EP240801a exhibits a burst duration of 148 s in X-rays and 22.3 s in gamma-rays, with X-rays leading by 80.61 s. Spectral lag analysis indicates the gamma-ray signal arrived 8.3 s earlier than the X-rays. Joint spectral fitting of EP/WXT and Fermi/GBM data yields an isotropic energy $E_{γ,\rm{iso}} = (5.57^{+0.54}_{-0.50})\times 10^{51}\,\rm{erg}$, a peak energy $E_{\rm{peak}} = 14.90^{+7.08}_{-4.71}\,\rm{keV}$, a fluence ratio $\rm S(25-50\,\rm{keV})/S(50-100\,\rm{keV}) = 1.67^{+0.74}_{-0.46}$, classifying EP240801a as an X-ray flash (XRF). The host-galaxy continuum spectrum, inferred using Prospector, was used to correct its contribution for the observed outburst optical data. Unusual early $R$-band behavior and EP/FXT observations suggest multiple components in the afterglow. Three models are considered: two-component jet model, forward-reverse shock model and forward-shock model with energy injection. Both three provide reasonable explanations. The two-component jet model and the energy injection model imply a relatively small initial energy and velocity of the jet in the line of sight, while the forward-reverse shock model remains typical. Under the two-component jet model, EP240801a may resemble GRB 221009A (BOAT) if the bright narrow beam is viewed on-axis. Therefore, EP240801a can be interpreted as an off-beam (narrow) jet or an intrinsically weak GRB jet. Our findings provide crucial clues for uncovering the origin of XRFs.

astro-ph.HE↗

arXiv2025

Stringy scaling of multi-tensor hard string scattering amplitudes and the K-identities

We calculate n-point hard string scattering amplitudes (HSSA) with n-2 tachyons and 2 tensor states at arbitrary mass levels. We discover the stringy scaling behavior of these HSSA. It is found that for HSSA with more than 2 transverse directions, the degree of stringy scaling dimM2 decreases comparing to the degree of stringy scaling dimM1 of the n-1 tachyons and 1 tensor HSSA calculated previously. Moreover, we propose a set of K-identities which is the key to demonstrate the stringy scaling behavior of HSSA. We explicitly prove both the diagonal and off-diagonal K-identities for the 4-point HSSA and give numerical proofs of these K-identities for some higher point HSSA.

hep-th↗

arXiv2025

Normalization through Fine-tuning: Understanding Wav2vec 2.0 Embeddings for Phonetic Analysis

Phonetic normalization plays a crucial role in speech recognition and analysis, ensuring the comparability of features derived from raw audio data. However, in the current paradigm of fine-tuning pre-trained large transformer models, phonetic normalization is not deemed a necessary step; instead, it is implicitly executed within the models. This study investigates the normalization process within transformer models, especially wav2vec 2.0. Through a comprehensive analysis of embeddings from models fine-tuned for various tasks, our results demonstrate that fine-tuning wav2vec 2.0 effectively achieves phonetic normalization by selectively suppressing task-irrelevant information. We found that models fine-tuned for multiple tasks retain information for both tasks without compromising performance, and that suppressing task-irrelevant information is not necessary for effective classification. These findings provide new insights into how phonetic normalization can be flexibly achieved in speech models and how it is realized in human speech perception.

cs.CL↗

arXiv2025

Vista-LLaMA: Reducing Hallucination in Video Language Models via Equal Distance to Visual Tokens

Recent advances in large video-language models have displayed promising outcomes in video comprehension. Current approaches straightforwardly convert video into language tokens and employ large language models for multi-modal tasks. However, this method often leads to the generation of irrelevant content, commonly known as "hallucination", as the length of the text increases and the impact of the video diminishes. To address this problem, we propose Vista-LLaMA, a novel framework that maintains the consistent distance between all visual tokens and any language tokens, irrespective of the generated text length. Vista-LLaMA omits relative position encoding when determining attention weights between visual and text tokens, retaining the position encoding for text and text tokens. This amplifies the effect of visual tokens on text generation, especially when the relative distance is longer between visual and text tokens. The proposed attention mechanism significantly reduces the chance of producing irrelevant text related to the video content. Furthermore, we present a sequential visual projector that projects the current video frame into tokens of language space with the assistance of the previous frame. This approach not only captures the temporal relationship within the video, but also allows less visual tokens to encompass the entire video. Our approach significantly outperforms various previous methods (e.g., Video-ChatGPT, MovieChat) on four challenging open-ended video question answering benchmarks. We reach an accuracy of 60.7 on the zero-shot NExT-QA and 60.5 on the zero-shot MSRVTT-QA, setting a new state-of-the-art performance. This project is available at https://jinxxian.github.io/Vista-LLaMA.

cs.CV↗

arXiv2025

OpenGS-SLAM: Open-Set Dense Semantic SLAM with 3D Gaussian Splatting for Object-Level Scene Understanding

Recent advancements in 3D Gaussian Splatting have significantly improved the efficiency and quality of dense semantic SLAM. However, previous methods are generally constrained by limited-category pre-trained classifiers and implicit semantic representation, which hinder their performance in open-set scenarios and restrict 3D object-level scene understanding. To address these issues, we propose OpenGS-SLAM, an innovative framework that utilizes 3D Gaussian representation to perform dense semantic SLAM in open-set environments. Our system integrates explicit semantic labels derived from 2D foundational models into the 3D Gaussian framework, facilitating robust 3D object-level scene understanding. We introduce Gaussian Voting Splatting to enable fast 2D label map rendering and scene updating. Additionally, we propose a Confidence-based 2D Label Consensus method to ensure consistent labeling across multiple views. Furthermore, we employ a Segmentation Counter Pruning strategy to improve the accuracy of semantic scene representation. Extensive experiments on both synthetic and real-world datasets demonstrate the effectiveness of our method in scene understanding, tracking, and mapping, achieving 10 times faster semantic rendering and 2 times lower storage costs compared to existing methods. Project page: https://young-bit.github.io/opengs-github.github.io/.

cs.CV↗

arXiv2025

Lateral Contrast Enhancement in Tomographic Volumetric 3D-Printing via Binary Photoinhibition

Tomographic volumetric 3D-printing (TVP) utilizes a nonlinear photoresponse of polymer precursor to cure all points in a three-dimensional (3D) object in parallel. A key challenge in TVP is to build up dose contrast between in-part and out-of-part points in a lateral plane, which relies on coordinated illumination from various projecting angles. This challenge has mainly been tackled by projection optimization. Here we show that designing material responses to photo-excitation can be a more effective way of addressing this challenge. By introducing a secondary photo-inhibitory species that reacts to external ultraviolet (UV) stimulus, we create a binary photoinhibition (BPI) system that greatly enhances the achievable dose contrast in a lateral plane. We first show that, in theory, combining dose subtraction with sufficient projection angles can guarantee an exact mathematical reconstruction of any greyscale design. We then propose a theoretical framework for BPI, in which a single stationary state with swappable stability can be used to realize dose subtraction. We use oxygen-lophyl radical pair as an approximation to show improvements in print quality enabled by enhanced dose contrast. In situ shadowgraphy shows that BPI improves the lateral patterning with various geometric features, creating differentiable changes in refractive index within 54 um or less. We show qualitative improvements in surface features and internal hollowness in physical prints. The direct impacts of UV light on the formation of positive and negative features on vertical and lateral planes of 5 workpieces are analyzed quantitatively. We conclude that introducing BPI with UV irradiation grants us direct control over the formation of negative features on the lateral plane.

physics.optics↗

arXiv2025

ReFu: Recursive Fusion for Exemplar-Free 3D Class-Incremental Learning

We introduce a novel Recursive Fusion model, dubbed ReFu, designed to integrate point clouds and meshes for exemplar-free 3D Class-Incremental Learning, where the model learns new 3D classes while retaining knowledge of previously learned ones. Unlike existing methods that either rely on storing historical data to mitigate forgetting or focus on single data modalities, ReFu eliminates the need for exemplar storage while utilizing the complementary strengths of both point clouds and meshes. To achieve this, we introduce a recursive method which continuously accumulates knowledge by updating the regularized auto-correlation matrix. Furthermore, we propose a fusion module, featuring a Pointcloud-guided Mesh Attention Layer that learns correlations between the two modalities. This mechanism effectively integrates point cloud and mesh features, leading to more robust and stable continual learning. Experiments across various datasets demonstrate that our proposed framework outperforms existing methods in 3D class-incremental learning.

cs.CV↗

arXiv2025

Hydra-SGG: Hybrid Relation Assignment for One-stage Scene Graph Generation

DETR introduces a simplified one-stage framework for scene graph generation (SGG) but faces challenges of sparse supervision and false negative samples. The former occurs because each image typically contains fewer than 10 relation annotations, while DETR-based SGG models employ over 100 relation queries. Each ground truth relation is assigned to only one query during training. The latter arises when one ground truth relation may have multiple queries with similar matching scores, leading to suboptimally matched queries being treated as negative samples. To address these, we propose Hydra-SGG, a one-stage SGG method featuring a Hybrid Relation Assignment. This approach combines a One-to-One Relation Assignment with an IoU-based One-to-Many Relation Assignment, increasing positive training samples and mitigating sparse supervision. In addition, we empirically demonstrate that removing self-attention between relation queries leads to duplicate predictions, which actually benefits the proposed One-to-Many Relation Assignment. With this insight, we introduce Hydra Branch, an auxiliary decoder without self-attention layers, to further enhance One-to-Many Relation Assignment by promoting different queries to make the same relation prediction. Hydra-SGG achieves state-of-the-art performance on multiple datasets, including VG150 (16.0 mR@50), Open Images V6 (50.1 weighted score), and GQA (12.7 mR@50).

cs.CV↗

arXiv2025

SE(3)-Equivariant Ternary Complex Prediction Towards Target Protein Degradation

Targeted protein degradation (TPD) induced by small molecules has emerged as a rapidly evolving modality in drug discovery, targeting proteins traditionally considered "undruggable". Proteolysis-targeting chimeras (PROTACs) and molecular glue degraders (MGDs) are the primary small molecules that induce TPD. Both types of molecules form a ternary complex linking an E3 ligase with a target protein, a crucial step for drug discovery. While significant advances have been made in binary structure prediction for proteins and small molecules, ternary structure prediction remains challenging due to obscure interaction mechanisms and insufficient training data. Traditional methods relying on manually assigned rules perform poorly and are computationally demanding due to extensive random sampling. In this work, we introduce DeepTernary, a novel deep learning-based approach that directly predicts ternary structures in an end-to-end manner using an encoder-decoder architecture. DeepTernary leverages an SE(3)-equivariant graph neural network (GNN) with both intra-graph and ternary inter-graph attention mechanisms to capture intricate ternary interactions from our collected high-quality training dataset, TernaryDB. The proposed query-based Pocket Points Decoder extracts the 3D structure of the final binding ternary complex from learned ternary embeddings, demonstrating state-of-the-art accuracy and speed in existing PROTAC benchmarks without prior knowledge from known PROTACs. It also achieves notable accuracy on the more challenging MGD benchmark under the blind docking protocol. Remarkably, our experiments reveal that the buried surface area calculated from predicted structures correlates with experimentally obtained degradation potency-related metrics. Consequently, DeepTernary shows potential in effectively assisting and accelerating the development of TPDs for previously undruggable targets.

q-bio.BM↗

arXiv2025

LLM-based MOFs Synthesis Condition Extraction using Few-Shot Demonstrations

The extraction of Metal-Organic Frameworks (MOFs) synthesis route from literature has been crucial for the logical MOFs design with desirable functionality. The recent advent of large language models (LLMs) provides disruptively new solution to this long-standing problem. While the latest researches mostly stick to primitive zero-shot LLMs lacking specialized material knowledge, we introduce in this work the few-shot LLM in-context learning paradigm. First, a human-AI interactive data curation approach is proposed to secure high-quality demonstrations. Second, an information retrieval algorithm is applied to pick and quantify few-shot demonstrations for each extraction. Over three datasets randomly sampled from nearly 90,000 well-defined MOFs, we conduct triple evaluations to validate our method. The synthesis extraction, structure inference, and material design performance of the proposed few-shot LLMs all significantly outplay zero-shot LLM and baseline methods. The lab-synthesized material guided by LLM surpasses 91.1% high-quality MOFs of the same class reported in the literature, on the key physical property of specific surface area.

cs.CL↗

arXiv2025

VideoGrain: Modulating Space-Time Attention for Multi-grained Video Editing

Recent advancements in diffusion models have significantly improved video generation and editing capabilities. However, multi-grained video editing, which encompasses class-level, instance-level, and part-level modifications, remains a formidable challenge. The major difficulties in multi-grained editing include semantic misalignment of text-to-region control and feature coupling within the diffusion model. To address these difficulties, we present VideoGrain, a zero-shot approach that modulates space-time (cross- and self-) attention mechanisms to achieve fine-grained control over video content. We enhance text-to-region control by amplifying each local prompt's attention to its corresponding spatial-disentangled region while minimizing interactions with irrelevant areas in cross-attention. Additionally, we improve feature separation by increasing intra-region awareness and reducing inter-region interference in self-attention. Extensive experiments demonstrate our method achieves state-of-the-art performance in real-world scenarios. Our code, data, and demos are available at https://knightyxp.github.io/VideoGrain_project_page/

cs.CV↗

arXiv2025

OpenVox: Real-time Instance-level Open-vocabulary Probabilistic Voxel Representation

In recent years, vision-language models (VLMs) have advanced open-vocabulary mapping, enabling mobile robots to simultaneously achieve environmental reconstruction and high-level semantic understanding. While integrated object cognition helps mitigate semantic ambiguity in point-wise feature maps, efficiently obtaining rich semantic understanding and robust incremental reconstruction at the instance-level remains challenging. To address these challenges, we introduce OpenVox, a real-time incremental open-vocabulary probabilistic instance voxel representation. In the front-end, we design an efficient instance segmentation and comprehension pipeline that enhances language reasoning through encoding captions. In the back-end, we implement probabilistic instance voxels and formulate the cross-frame incremental fusion process into two subtasks: instance association and live map evolution, ensuring robustness to sensor and segmentation noise. Extensive evaluations across multiple datasets demonstrate that OpenVox achieves state-of-the-art performance in zero-shot instance segmentation, semantic segmentation, and open-vocabulary retrieval. Furthermore, real-world robotics experiments validate OpenVox's capability for stable, real-time operation.

cs.RO↗