arXiv subjects

Yifan Zhou

Publications and source records attributed to Yifan Zhou.

At least 109 records · Page 6Linked to original sources

arXiv2025

X-Intelligence 3.0: Training and Evaluating Reasoning LLM for Semiconductor Display

Large language models (LLMs) have recently achieved significant advances in reasoning and demonstrated their advantages in solving challenging problems. Yet, their effectiveness in the semiconductor display industry remains limited due to a lack of domain-specific training and expertise. To bridge this gap, we present X-Intelligence 3.0, the first high-performance reasoning model specifically developed for the semiconductor display industry. This model is designed to deliver expert-level understanding and reasoning for the industry's complex challenges. Leveraging a carefully curated industry knowledge base, the model undergoes supervised fine-tuning and reinforcement learning to enhance its reasoning and comprehension capabilities. To further accelerate development, we implemented an automated evaluation framework that simulates expert-level assessments. We also integrated a domain-specific retrieval-augmented generation (RAG) mechanism, resulting in notable performance gains on benchmark datasets. Despite its relatively compact size of 32 billion parameters, X-Intelligence 3.0 outperforms SOTA DeepSeek-R1-671B across multiple evaluations. This demonstrates its exceptional efficiency and establishes it as a powerful solution to the longstanding reasoning challenges faced by the semiconductor display industry.

cs.CL↗

arXiv2025

Wavelet-GS: 3D Gaussian Splatting with Wavelet Decomposition

3D Gaussian Splatting (3DGS) has revolutionized 3D scene reconstruction, which effectively balances rendering quality, efficiency, and speed. However, existing 3DGS approaches usually generate plausible outputs and face significant challenges in complex scene reconstruction, manifesting as incomplete holistic structural outlines and unclear local lighting effects. To address these issues simultaneously, we propose a novel decoupled optimization framework, which integrates wavelet decomposition into 3D Gaussian Splatting and 2D sampling. Technically, through 3D wavelet decomposition, our approach divides point clouds into high-frequency and low-frequency components, enabling targeted optimization for each. The low-frequency component captures global structural outlines and manages the distribution of Gaussians through voxelization. In contrast, the high-frequency component restores intricate geometric and textural details while incorporating a relight module to mitigate lighting artifacts and enhance photorealistic rendering. Additionally, a 2D wavelet decomposition is applied to the training images, simulating radiance variations. This provides critical guidance for high-frequency detail reconstruction, ensuring seamless integration of details with the global structure. Extensive experiments on challenging datasets demonstrate our method achieves state-of-the-art performance across various metrics, surpassing existing approaches and advancing the field of 3D scene reconstruction.

cs.GR↗

arXiv2025

TokensGen: Harnessing Condensed Tokens for Long Video Generation

Generating consistent long videos is a complex challenge: while diffusion-based generative models generate visually impressive short clips, extending them to longer durations often leads to memory bottlenecks and long-term inconsistency. In this paper, we propose TokensGen, a novel two-stage framework that leverages condensed tokens to address these issues. Our method decomposes long video generation into three core tasks: (1) inner-clip semantic control, (2) long-term consistency control, and (3) inter-clip smooth transition. First, we train To2V (Token-to-Video), a short video diffusion model guided by text and video tokens, with a Video Tokenizer that condenses short clips into semantically rich tokens. Second, we introduce T2To (Text-to-Token), a video token diffusion transformer that generates all tokens at once, ensuring global consistency across clips. Finally, during inference, an adaptive FIFO-Diffusion strategy seamlessly connects adjacent clips, reducing boundary artifacts and enhancing smooth transitions. Experimental results demonstrate that our approach significantly enhances long-term temporal and content coherence without incurring prohibitive computational overhead. By leveraging condensed tokens and pre-trained short video models, our method provides a scalable, modular solution for long video generation, opening new possibilities for storytelling, cinematic production, and immersive simulations. Please see our project page at https://vicky0522.github.io/tokensgen-webpage/ .

cs.CV↗

arXiv2025

Towards Efficient Privacy-Preserving Machine Learning: A Systematic Review from Protocol, Model, and System Perspectives

Privacy-preserving machine learning (PPML) based on cryptographic protocols has emerged as a promising paradigm to protect user data privacy in cloud-based machine learning services. While it achieves formal privacy protection, PPML often incurs significant efficiency and scalability costs due to orders of magnitude overhead compared to the plaintext counterpart. Therefore, there has been a considerable focus on mitigating the efficiency gap for PPML. In this survey, we provide a comprehensive and systematic review of recent PPML studies with a focus on cross-level optimizations. Specifically, we categorize existing papers into protocol level, model level, and system level, and review progress at each level. We also provide qualitative and quantitative comparisons of existing works with technical insights, based on which we discuss future research directions and highlight the necessity of integrating optimizations across protocol, model, and system levels. We hope this survey can provide an overarching understanding of existing approaches and potentially inspire future breakthroughs in the PPML field. As the field is evolving fast, we also provide a public GitHub repository to continuously track the developments, which is available at https://github.com/PKU-SEC-Lab/Awesome-PPML-Papers.

cs.CR↗

arXiv2025

RadFabric: Agentic AI System with Reasoning Capability for Radiology

Chest X ray (CXR) imaging remains a critical diagnostic tool for thoracic conditions, but current automated systems face limitations in pathology coverage, diagnostic accuracy, and integration of visual and textual reasoning. To address these gaps, we propose RadFabric, a multi agent, multimodal reasoning framework that unifies visual and textual analysis for comprehensive CXR interpretation. RadFabric is built on the Model Context Protocol (MCP), enabling modularity, interoperability, and scalability for seamless integration of new diagnostic agents. The system employs specialized CXR agents for pathology detection, an Anatomical Interpretation Agent to map visual findings to precise anatomical structures, and a Reasoning Agent powered by large multimodal reasoning models to synthesize visual, anatomical, and clinical data into transparent and evidence based diagnoses. RadFabric achieves significant performance improvements, with near-perfect detection of challenging pathologies like fractures (1.000 accuracy) and superior overall diagnostic accuracy (0.799) compared to traditional systems (0.229 to 0.527). By integrating cross modal feature alignment and preference-driven reasoning, RadFabric advances AI-driven radiology toward transparent, anatomically precise, and clinically actionable CXR analysis.

cs.CV↗

arXiv2025

MR.NAVI: Mixed-Reality Navigation Assistant for the Visually Impaired

Over 43 million people worldwide live with severe visual impairment, facing significant challenges in navigating unfamiliar environments. We present MR.NAVI, a mixed reality system that enhances spatial awareness for visually impaired users through real-time scene understanding and intuitive audio feedback. Our system combines computer vision algorithms for object detection and depth estimation with natural language processing to provide contextual scene descriptions, proactive collision avoidance, and navigation instructions. The distributed architecture processes sensor data through MobileNet for object detection and employs RANSAC-based floor detection with DBSCAN clustering for obstacle avoidance. Integration with public transit APIs enables navigation with public transportation directions. Through our experiments with user studies, we evaluated both scene description and navigation functionalities in unfamiliar environments, showing promising usability and effectiveness.

cs.CV↗

arXiv2025

IRNN: Innovation-driven Recurrent Neural Network for Time-Series Data Modeling and Prediction

Many real-world datasets are time series that are sequentially collected and contain rich temporal information. Thus, a common interest in practice is to capture dynamics of time series and predict their future evolutions. To this end, the recurrent neural network (RNN) has been a prevalent and effective machine learning option, which admits a nonlinear state-space model representation. Motivated by the resemblance between RNN and Kalman filter (KF) for linear state-space models, we propose in this paper Innovation-driven RNN (IRNN), a novel RNN architecture tailored to time-series data modeling and prediction tasks. By adapting the concept of "innovation" from KF to RNN, past prediction errors are adopted as additional input signals to update hidden states of RNN and boost prediction performance. Since innovation data depend on network parameters, existing training algorithms for RNN do not apply to IRNN straightforwardly. Thus, a tailored training algorithm dubbed input updating-based back-propagation through time (IU-BPTT) is further proposed, which alternates between updating innovations and optimizing network parameters via gradient descent. Experiments on real-world benchmark datasets show that the integration of innovations into various forms of RNN leads to remarkably improved prediction accuracy of IRNN without increasing the training cost substantially.

cs.LG↗

arXiv2025

The JWST weather report from the nearest brown dwarfs II: Consistent variability mechanisms over 7 months revealed by 1-14 $μ$m NIRSpec + MIRI monitoring of WISE 1049AB

We present a new epoch of JWST spectroscopic variability monitoring of the benchmark binary brown dwarf WISE 1049AB, the closest, brightest brown dwarfs known. Our 8-hour MIRI low resolution spectroscopy (LRS) and 7-hour NIRSpec prism observations extended variability measurements for any brown dwarfs beyond 11 $μ$m for the first time, reaching up to 14 $μ$m. Combined with the previous epoch in 2023, they set the longest JWST weather monitoring baseline to date. We found that both WISE 1049AB show wavelength-dependent light curve behaviours. Using a robust k-means clustering algorithm, we identified several clusters of variability behaviours associated with three distinct pressure levels. By comparing to a general circulation model (GCM), we identified the possible mechanisms that drive the variability at these pressure levels: Patchy clouds rotating in and out of view likely shaped the dramatic light curves in the deepest layers between 1-2.5 $μ$m, whereas hot spots arising from temperature / chemical variations of molecular species likely dominate the high-altitude levels between 2.5-3.6 $μ$m and 4.3-8.5 $μ$m. Small-grain silicates potentially contributed to the variability of WISE 1049A at 8.5-11 $μ$m. While distinct atmospheric layers are governed by different mechanisms, we confirmed for the first time that each variability mechanism remains consistent within its layer over the long term. Future multi-period observations will further test the stability of variability mechanisms on this binary, and expanded JWST variability surveys across the L-T-Y sequence will allow us to trace and understand variability mechanisms across a wider population of brown dwarfs and planetary-mass objects.

astro-ph.SR↗

arXiv2025

SAS-Prompt: Large Language Models as Numerical Optimizers for Robot Self-Improvement

We demonstrate the ability of large language models (LLMs) to perform iterative self-improvement of robot policies. An important insight of this paper is that LLMs have a built-in ability to perform (stochastic) numerical optimization and that this property can be leveraged for explainable robot policy search. Based on this insight, we introduce the SAS Prompt (Summarize, Analyze, Synthesize) -- a single prompt that enables iterative learning and adaptation of robot behavior by combining the LLM's ability to retrieve, reason and optimize over previous robot traces in order to synthesize new, unseen behavior. Our approach can be regarded as an early example of a new family of explainable policy search methods that are entirely implemented within an LLM. We evaluate our approach both in simulation and on a real-robot table tennis task. Project website: sites.google.com/asu.edu/sas-llm/

cs.RO↗

arXiv2025

Metasurface-Assisted Adaptive Quantum Phase Contrast Imaging

Quantum imaging employs the nonclassical correlation of photons to break through the noise limitation of classical imaging, realizing high sensitivity, high SNR imaging and multifunctional image processing. To enhance the flexibility and imaging performance of the optical systems, metasurfaces composed of subwavelength structural units provide a powerful optimization approach, enabling advanced applications in quantum state modulation and high-precision imaging. Conventional phase contrast imaging is fundamentally constrained by its single-phase modulation scheme, precluding adaptive switching between imaging modalities. Therefore, the development of high-contrast imaging techniques that can be used in any combination of phases has been a challenge in the field of optical imaging. Here, we propose a novel imaging scheme combining a polarization-entangled light source and a polarization multiplexed metasurface, which realizes remotely switchable bright-dark phase contrast imaging, demonstrating the flexibility and high integration of the system. Experiments show the system can realize high contrast and high SNR imaging under low phase gradient conditions (phase difference as low as π/5) and exhibit excellent phase resolution. In addition, the system is suitable for imaging biological samples under low-throughput light conditions, providing an efficient and non-destructive shooting solution for biomedical imaging and promoting the development of phase-sensitive imaging technology.

physics.optics↗

arXiv2025

EG-Gaussian: Epipolar Geometry and Graph Network Enhanced 3D Gaussian Splatting

In this paper, we explore an open research problem concerning the reconstruction of 3D scenes from images. Recent methods have adopt 3D Gaussian Splatting (3DGS) to produce 3D scenes due to its efficient training process. However, these methodologies may generate incomplete 3D scenes or blurred multiviews. This is because of (1) inaccurate 3DGS point initialization and (2) the tendency of 3DGS to flatten 3D Gaussians with the sparse-view input. To address these issues, we propose a novel framework EG-Gaussian, which utilizes epipolar geometry and graph networks for 3D scene reconstruction. Initially, we integrate epipolar geometry into the 3DGS initialization phase to enhance initial 3DGS point construction. Then, we specifically design a graph learning module to refine 3DGS spatial features, in which we incorporate both spatial coordinates and angular relationships among neighboring points. Experiments on indoor and outdoor benchmark datasets demonstrate that our approach significantly improves reconstruction accuracy compared to 3DGS-based methods.

cs.CV↗

arXiv2025

The Essence of Contextual Understanding in Theory of Mind: A Study on Question Answering with Story Characters

Theory-of-Mind (ToM) is a fundamental psychological capability that allows humans to understand and interpret the mental states of others. Humans infer others' thoughts by integrating causal cues and indirect clues from broad contextual information, often derived from past interactions. In other words, human ToM heavily relies on the understanding about the backgrounds and life stories of others. Unfortunately, this aspect is largely overlooked in existing benchmarks for evaluating machines' ToM capabilities, due to their usage of short narratives without global context, especially personal background of characters. In this paper, we verify the importance of comprehensive contextual understanding about personal backgrounds in ToM and assess the performance of LLMs in such complex scenarios. To achieve this, we introduce CharToM benchmark, comprising 1,035 ToM questions based on characters from classic novels. Our human study reveals a significant disparity in performance: the same group of educated participants performs dramatically better when they have read the novels compared to when they have not. In parallel, our experiments on state-of-the-art LLMs, including the very recent o1 and DeepSeek-R1 models, show that LLMs still perform notably worse than humans, despite that they have seen these stories during pre-training. This highlights the limitations of current LLMs in capturing the nuanced contextual information required for ToM reasoning.

cs.CL↗

arXiv2025

FreeFlux: Understanding and Exploiting Layer-Specific Roles in RoPE-Based MMDiT for Versatile Image Editing

The integration of Rotary Position Embedding (RoPE) in Multimodal Diffusion Transformer (MMDiT) has significantly enhanced text-to-image generation quality. However, the fundamental reliance of self-attention layers on positional embedding versus query-key similarity during generation remains an intriguing question. We present the first mechanistic analysis of RoPE-based MMDiT models (e.g., FLUX), introducing an automated probing strategy that disentangles positional information versus content dependencies by strategically manipulating RoPE during generation. Our analysis reveals distinct dependency patterns that do not straightforwardly correlate with depth, offering new insights into the layer-specific roles in RoPE-based MMDiT. Based on these findings, we propose a training-free, task-specific image editing framework that categorizes editing tasks into three types: position-dependent editing (e.g., object addition), content similarity-dependent editing (e.g., non-rigid editing), and region-preserved editing (e.g., background replacement). For each type, we design tailored key-value injection strategies based on the characteristics of the editing task. Extensive qualitative and quantitative evaluations demonstrate that our method outperforms state-of-the-art approaches, particularly in preserving original semantic content and achieving seamless modifications.

cs.CV↗

arXiv2025

From Task-Specific Models to Unified Systems: A Review of Model Merging Approaches

Model merging has achieved significant success, with numerous innovative methods proposed to enhance capabilities by combining multiple models. However, challenges persist due to the lack of a unified framework for classification and systematic comparative analysis, leading to inconsistencies in terminologies and categorizations. Meanwhile, as an increasing number of fine-tuned models are publicly available, their original training data often remain inaccessible due to privacy concerns or intellectual property restrictions. This makes traditional multi-task learning based on shared training data impractical. In scenarios where direct access to training data is infeasible, merging model parameters to create a unified model with broad generalization across multiple domains becomes crucial, further underscoring the importance of model merging techniques. Despite the rapid progress in this field, a comprehensive taxonomy and survey summarizing recent advances and predicting future directions are still lacking. This paper addresses these gaps by establishing a new taxonomy of model merging methods, systematically comparing different approaches, and providing an overview of key developments. By offering a structured perspective on this evolving area, we aim to help newcomers quickly grasp the field's landscape and inspire further innovations.

cs.LG↗

arXiv2025

Efficient Multi-Task Inferencing: Model Merging with Gromov-Wasserstein Feature Alignment

Automatic scoring of student responses enhances efficiency in education, but deploying a separate neural network for each task increases storage demands, maintenance efforts, and redundant computations. To address these challenges, this paper introduces the Gromov-Wasserstein Scoring Model Merging (GW-SMM) method, which merges models based on feature distribution similarities measured via the Gromov-Wasserstein distance. Our approach begins by extracting features from student responses using individual models, capturing both item-specific context and unique learned representations. The Gromov-Wasserstein distance then quantifies the similarity between these feature distributions, identifying the most compatible models for merging. Models exhibiting the smallest pairwise distances, typically in pairs or trios, are merged by combining only the shared layers preceding the classification head. This strategy results in a unified feature extractor while preserving separate classification heads for item-specific scoring. We validated our approach against human expert knowledge and a GPT-o1-based merging method. GW-SMM consistently outperformed both, achieving a higher micro F1 score, macro F1 score, exact match accuracy, and per-label accuracy. The improvements in micro F1 and per-label accuracy were statistically significant compared to GPT-o1-based merging (p=0.04, p=0.01). Additionally, GW-SMM reduced storage requirements by half without compromising much accuracy, demonstrating its computational efficiency alongside reliable scoring performance.

cs.CL↗

arXiv2025

RGB-Only Gaussian Splatting SLAM for Unbounded Outdoor Scenes

3D Gaussian Splatting (3DGS) has become a popular solution in SLAM, as it can produce high-fidelity novel views. However, previous GS-based methods primarily target indoor scenes and rely on RGB-D sensors or pre-trained depth estimation models, hence underperforming in outdoor scenarios. To address this issue, we propose a RGB-only gaussian splatting SLAM method for unbounded outdoor scenes--OpenGS-SLAM. Technically, we first employ a pointmap regression network to generate consistent pointmaps between frames for pose estimation. Compared to commonly used depth maps, pointmaps include spatial relationships and scene geometry across multiple views, enabling robust camera pose estimation. Then, we propose integrating the estimated camera poses with 3DGS rendering as an end-to-end differentiable pipeline. Our method achieves simultaneous optimization of camera poses and 3DGS scene parameters, significantly enhancing system tracking accuracy. Specifically, we also design an adaptive scale mapper for the pointmap regression network, which provides more accurate pointmap mapping to the 3DGS map representation. Our experiments on the Waymo dataset demonstrate that OpenGS-SLAM reduces tracking error to 9.8\% of previous 3DGS methods, and achieves state-of-the-art results in novel view synthesis. Project Page: https://3dagentworld.github.io/opengs-slam/

cs.CV↗

arXiv2025

H$α$ Variability of AB Aur b with the Hubble Space Telescope: Probing the Nature of a Protoplanet Candidate with Accretion Light Echoes

Giant planets generate accretion luminosity as they form. Much of this energy is radiated in strong H$α$ line emission, which has motivated direct imaging surveys at optical wavelengths to search for accreting protoplanets. However, compact disk structures can mimic accreting planets by scattering emission from the host star. This can complicate the interpretation of H$α$ point sources, especially if the host star itself is accreting. We describe an approach to distinguish accreting protoplanets from scattered-light disk features using "accretion light echoes." This method relies on variable H$α$ emission from a stochastically accreting host star to search for a delayed brightness correlation with a candidate protoplanet. We apply this method to the candidate protoplanet AB Aur b with a dedicated Hubble Space Telescope Wide Field Camera 3 program designed to sequentially sample the host star and the candidate planet in H$α$ while accounting for the light travel time delay and orbital geometry of the source within the protoplanetary disk. Across five epochs spanning 14 months, AB Aur b is over 20 times more variable than its host star; AB Aur's H$α$ emission changes by 15% while AB Aur b varies by 330%. These brightness changes are not correlated, which rules out unobstructed scattered starlight from the host star as the only source of AB Aur b's H$α$ emission and is consistent with tracing emission from an independently accreting protoplanet, inner disk shadowing effects, or a physically evolving compact disk structure. More broadly, accretion light echoes offer a novel tool to explore the nature of protoplanet candidates with well-timed observations of the host star prior to deep imaging in H$α$.

astro-ph.EP↗

arXiv2025

Evidence for Variable Accretion onto PDS 70 c and Implications for Protoplanet Detections

Understanding the processes of planet formation and accretion in young systems is essential to unraveling the initial conditions of planetary systems. The PDS 70 system, which hosts two directly imaged protoplanets, provides a unique laboratory for studying these phenomena, particularly through H-alpha emission a commonly used accretion tracer. We present multi-epoch observations and examine the variability in accretion signatures within this system, focusing on PDS 70 b and c. Using Hubble Space Telescope narrowband H-alpha imaging from 2020 and 2024, we achieve high signal-to-noise ratio detections of these planets and reveal significant changes in H-alpha flux. For PDS 70 c, the H-alpha flux more than doubled between 2020 and 2024. The trend is consistent with the one identified in recently published MagAO-X data, further confirming that PDS 70 c has become significantly brighter in H between 2023 March and 2024 May. The observed variability suggests dynamic accretion processes, possibly modulated by circumplanetary disk properties or transient accretion bursts. High-amplitude variability in PDS 70 c motivates simultaneous monitoring of multiple accretion tracers to probe the mechanisms of mass growth of gas giant planets. We quantify the impact of variability on the detectability of protoplanets in imaging surveys and emphasize the need for continued and regular monitoring to accurately assess the occurrence and characteristics of young, forming planets.

astro-ph.EP↗