Searcharxiv⌕ Search

arXiv subjects

Yifan Zhao

Publications and source records attributed to Yifan Zhao.

At least 37 records · Page 2Linked to original sources
arXiv2026

The Right Inference Strategy Is All You Need: Nearly Training-Free Domain-Wise Inference for EgoCross Challenge

EgoCross evaluates multimodal large language models on egocentric video question answering under substantial domain shift, where test videos come from surgery, industrial assembly, extreme sports, and animal-mounted cameras rather than ordinary daily-life scenes. In the source-limited track, the base model is fixed to Qwen3-VL-4B, while the official task-specific support set contains only 20 training samples. This setting makes the challenge less about model scaling and more about exposing the right visual, temporal, and answer-selection cues to a constrained model. Our key observation is that the frozen baseline model is not simply incapable of these rare scenarios; rather, it often fails to transfer its existing visual-language knowledge to the new task format without an appropriate interface. We therefore use a domain-wise inference strategy that treats the four target domains separately and designs different input, prompting, and answer-mapping procedures according to each domain's task characteristics. These strategies make the rare egocentric scenes more interpretable to the VLM by emphasizing the cues that matter for each domain. The resulting system is nearly training-free: surgery, and animal questions are answered with the base Qwen3-VL-4B model, while XSports and industry use only the official SFT checkpoint trained for two epochs on the provided 20 training samples. On the final evaluation, this simple strategy reaches 66.98\% overall accuracy, suggesting that careful domain-aware inference can compensate for limited base-model strength and recover much of the ability already present in the baseline model.

cs.CV↗
arXiv2026

Envisioning Beyond the Few: Disentangled Semantics and Primitives for Few-Shot Atypical Layout-to-Image Generation

The layout-to-image (L2I) task enables fine-grained control over image generation via object categories and spatial layouts. However, existing L2I methods yield fragmented and distorted generations under few-shot atypical settings. We term this failure as representation fragmentation, arising from a granularity mismatch that entangles semantic identity with visual details. To address this issue, we propose a representation-driven framework that disentangles semantics from primitives for robust few-shot adaptation. Specifically, Semantic Anchoring aggregates categorical semantics into anchors for stable identity, while Primitive Imbuing models recomposable primitives for robust local detail modeling. Conceptual Steering further regulates optimization with a saliency-aware objective to preserve foreground semantic consistency. Extensive experiments demonstrate consistent improvements in the 5-shot regime over state-of-the-art L2I methods in both visual fidelity and alignment across diverse atypical domains. The source code is publicly available at https://github.com/iCVTEAM/DSP.

cs.CV↗
arXiv2026

ARQ: A Mixed-Precision Quantization Framework for Accurate and Certifiably Robust DNNs

Mixed precision quantization has become an important technique for optimizing the execution of deep neural networks (DNNs). Certified robustness, which provides provable guarantees about a model's ability to withstand different adversarial perturbations, has rarely been addressed in quantization due to the unacceptably high cost of certifying robustness. This paper introduces ARQ, an innovative mixed-precision quantization method that not only preserves the clean accuracy of the smoothed classifiers, but also maintains their certified robustness. ARQ uses reinforcement learning to find accurate and robust DNN quantization, while efficiently leveraging randomized smoothing, a popular class of statistical DNN verification algorithms. ARQ consistently performs better than multiple state-of-the-art quantization techniques across all the benchmarks and the input perturbation levels. The performance of ARQ quantized networks reaches that of the original DNN with floating-point weights, while using only 1.5% instructions and the highest certified radius. ARQ's code is available at https://github.com/uiuc-arc/ARQ.

cs.LG↗
arXiv2026

Neptune: Advanced ML Operator Fusion for Locality and Parallelism on GPUs

Operator fusion has become a key optimization for deep learning, which combines multiple deep learning operators to improve data reuse and reduce global memory transfers. However, existing tensor compilers struggle to fuse complex reduction computations involving loop-carried dependencies, such as attention mechanisms. This paper introduces Neptune, a tensor compiler for advanced operator fusion for sequences of reduction operators. Neptune presents a new approach for advanced operator fusion, which intentionally breaks some existing dependencies and compensates by constructing algebraic correction expressions that allow the kernel to produce the correct result. Applying Neptune's advanced operator fusion to a plain attention operator generates operators equivalent to FlashAttention and FlashDecoding. On ten attention-based benchmarks, Neptune, starting from a plain attention code and a high-level scheduling template, outperforms existing compilers like Triton, TVM, and FlexAttention, including Triton-based implementations of FlashAttention. Across four different GPU architectures from NVIDIA and AMD, Neptune-generated kernels have an average speedup of $1.35\times$ over the next best alternative, with up to $2.65\times$ speedup on Nvidia GPUs and up to $3.32\times$ on AMD GPUs, demonstrating its effectiveness for deep learning workloads.

cs.PL↗
arXiv2026

Nautilus: An Auto-Scheduling Tensor Compiler for Efficient Tiled GPU Kernels

We present Nautilus, a novel tensor compiler that moves toward fully automated math-to-kernel optimization. Nautilus compiles a high-level algebraic specification of tensor operators into efficient tiled GPU kernels. Nautilus's successive lowering design allows high-level optimizations, expression rewrites, and tile optimizations to be jointly applied in a single end-to-end system. Nautilus presents a novel auto-scheduler that discovers sequences of high-level optimizations, while preserving the regular program structure needed by tile optimizers. Nautilus's auto-scheduler captures complex interactions and trade-offs in the high-level optimizations, including aggressive global transformations like advanced reduction fusion. Nautilus is the first end-to-end tensor compiler capable of starting from a math-like description of attention and automatically discovering FlashAttention-3-like kernels, offloading the entire burden of optimization from the programmer to the compiler. Across five transformer-based models and 150 evaluation configurations on NVIDIA GH200 and RTX 5090 GPUs, Nautilus achieves up to 23% higher throughput than state-of-the-art compilers on GH200 and up to 42% on RTX 5090, while matching or exceeding manually written cuDNN kernels on many long-sequence configurations.

cs.PL↗
arXiv2026

Terahertz Generation and Detection through Gain-Enhanced Interband Photomixing in Quantum Well Structures

Terahertz waves hold immense potential across diverse fields, including healthcare monitoring, biomedical imaging, precision navigation, high-speed communication, security screening, industrial quality control, and space exploration. However, the widespread adoption of terahertz technology has been hindered by the bulky, complex, and costly nature of existing systems. Here, we demonstrate gain-enhanced interband photomixing in quantum well (QW) PIN photodiodes as an efficient mechanism for frequency-tunable terahertz generation and detection, achieving significant improvements in power efficiency and sensitivity over the state-of-the-art. QWs embedded in PIN photodiodes - key elements of commercially available photonic integrated circuits (PICs) - enable monolithic integration of lasers, semiconductor optical amplifiers (SOAs), modulators, filters, demultiplexers, and other passive optical components. By establishing QW PIN photodiodes as the foundation of a Monolithically Integrated Terahertz Optoelectronic (MITO) platform, this work paves the way for compact, scalable terahertz optoelectronic systems with applications in high-speed data transfer, spectroscopy, and hyperspectral imaging. This advancement positions terahertz technology for widespread use, facilitating practical applications across remote sensing, communications, and medical diagnostics within portable devices.

physics.optics↗
arXiv2026

Stable maps, multiplicities, and compactified Jacobians

Let $C$ be a complex projective integral curve with planar singularities. In this note, we study numerical relations among its versal deformation space, moduli space of stable maps, and compactified Jacobian. In particular, we correct a statement by Fantechi--Göttsche--van Straten on the multiplicity of the $δ$-constant stratum of the versal deformation space at $[C]$. We also give a necessary and sufficient condition for the original claim to hold.

math.AG↗
arXiv2026

Seeing through Light and Darkness: Sensor-Physics Grounded Deblurring HDR NeRF from Single-Exposure Images and Events

Novel view synthesis from low dynamic range (LDR) blurry images, which are common in the wild, struggles to recover high dynamic range (HDR) and sharp 3D representations in extreme lighting conditions. Although existing methods employ event data to address this issue, they ignore the sensor-physics mismatches between the camera output and physical world radiance, resulting in suboptimal HDR and deblurring results. To cope with this problem, we propose a unified sensor-physics grounded NeRF framework for sharp HDR novel view synthesis from single-exposure blurry LDR images and corresponding events. We employ NeRF to directly represent the actual radiance of the 3D scene in the HDR domain and model raw HDR scene rays hitting the sensor pixels as in the physical world. A 2D pixel-wise RGB CRF model is introduced to align the NeRF rendered pixel values with the sensor-recorded LDR pixel values of the input images. A novel event CRF model is also designed to bridge the gap between physical scene dynamics and event sensor output. The two models are jointly optimized with the NeRF network, leveraging the spatial and temporal dynamic information in events to enhance the sharp HDR 3D representation learning. Experiments on the collected and public datasets demonstrate that our method achieves state-of-the-art HDR and deblurring novel view synthesis results with single-exposure blurry LDR images and corresponding events.

cs.CV↗
arXiv2026

Search-Induced Issues in Web-Augmented LLM Code Generation: Detecting and Repairing Error-Inducing Pages

Web-augmented large language models (LLMs) offer promising capabilities for automatic code generation. However, integrating live web search exposes models to unreliable or malicious content, leading to Search-Induced Issues (SII), a novel failure mode in which external pages mislead LLMs into producing incorrect code. This paper presents a comprehensive empirical study of the prevalence and impact of SII across three commercial search APIs and six advanced LLMs. Our analysis reveals that all evaluated web-augmented LLMs are vulnerable to SII, with root causes arising from either misaligned specifications or flawed code implementations in the searched Error-Inducing Pages (EIPs). To address this challenge, we propose Sherlock, an automated framework that enables LLM service providers to proactively safeguard web-augmented generation systems at scale. Sherlock operates as a continuous pipeline that first detects potential SII instances, then debugs them to identify the responsible EIPs and pinpoint their root causes, and finally repairs them by either annotating misaligned content or replacing erroneous code snippets with evaluated solutions from trusted sources. Experiments show that Sherlock identifies EIPs with an F1 score of up to 95% and repairs 71% to 100% of affected generations across the evaluated models, with modest computational overhead. Our findings and framework provide practical guidance for improving the reliability of web-augmented LLM-based code generation systems in real-world software engineering scenarios.

cs.SE↗
arXiv2026

Diffusion-Classifier Synergy: Reward-Aligned Learning via Mutual Boosting Loop for FSCIL

Few-Shot Class-Incremental Learning (FSCIL) challenges models to sequentially learn new classes from minimal examples without forgetting prior knowledge, a task complicated by the stability-plasticity dilemma and data scarcity. Current FSCIL methods often struggle with generalization due to their reliance on limited datasets. While diffusion models offer a path for data augmentation, their direct application can lead to semantic misalignment or ineffective guidance. This paper introduces Diffusion-Classifier Synergy (DCS), a novel framework that establishes a mutual boosting loop between diffusion model and FSCIL classifier. DCS utilizes a reward-aligned learning strategy, where a dynamic, multi-faceted reward function derived from the classifier's state directs the diffusion model. This reward system operates at two levels: the feature level ensures semantic coherence and diversity using prototype-anchored maximum mean discrepancy and dimension-wise variance matching, while the logits level promotes exploratory image generation and enhances inter-class discriminability through confidence recalibration and cross-session confusion-aware mechanisms. This co-evolutionary process, where generated images refine the classifier and an improved classifier state yields better reward signals, demonstrably achieves state-of-the-art performance on FSCIL benchmarks, significantly enhancing both knowledge retention and new class learning.

cs.CV↗
arXiv2026

Re-coding for Uncertainties: Edge-awareness Semantic Concordance for Resilient Event-RGB Segmentation

Semantic segmentation has achieved great success in ideal conditions. However, when facing extreme conditions (e.g., insufficient light, fierce camera motion), most existing methods suffer from significant information loss of RGB, severely damaging segmentation results. Several researches exploit the high-speed and high-dynamic event modality as a complement, but event and RGB are naturally heterogeneous, which leads to feature-level mismatch and inferior optimization of existing multi-modality methods. Different from these researches, we delve into the edge secret of both modalities for resilient fusion and propose a novel Edge-awareness Semantic Concordance framework to unify the multi-modality heterogeneous features with latent edge cues. In this framework, we first propose Edge-awareness Latent Re-coding, which obtains uncertainty indicators while realigning event-RGB features into unified semantic space guided by re-coded distribution, and transfers event-RGB distributions into re-coded features by utilizing a pre-established edge dictionary as clues. We then propose Re-coded Consolidation and Uncertainty Optimization, which utilize re-coded edge features and uncertainty indicators to solve the heterogeneous event-RGB fusion issues under extreme conditions. We establish two synthetic and one real-world event-RGB semantic segmentation datasets for extreme scenario comparisons. Experimental results show that our method outperforms the state-of-the-art by a 2.55% mIoU on our proposed DERS-XS, and possesses superior resilience under spatial occlusion. Our code and datasets are publicly available at https://github.com/iCVTEAM/ESC.

cs.CV↗
arXiv2026

WorldTree: Towards 4D Dynamic Worlds from Monocular Video using Tree-Chains

Dynamic reconstruction has achieved remarkable progress, but there remain challenges in monocular input for more practical applications. The prevailing works attempt to construct efficient motion representations, but lack a unified spatiotemporal decomposition framework, suffering from either holistic temporal optimization or coupled hierarchical spatial composition. To this end, we propose WorldTree, a unified framework comprising Temporal Partition Tree (TPT) that enables coarse-to-fine optimization based on the inheritance-based partition tree structure for hierarchical temporal decomposition, and Spatial Ancestral Chains (SAC) that recursively query ancestral hierarchical structure to provide complementary spatial dynamics while specializing motion representations across ancestral nodes. Experimental results on different datasets indicate that our proposed method achieves 8.26% improvement of LPIPS on NVIDIA-LS and 9.09% improvement of mLPIPS on DyCheck compared to the second-best method. Code: https://github.com/iCVTEAM/WorldTree.

cs.CV↗
arXiv2026

"I Was Told to Come Back and Share This": Social Media-Based Near-Death Experience Disclosures as Expressions of Spiritual Beliefs

People who experienced near-death events often turn to personal expression as a way of processing trauma and articulating beliefs. While scholars have examined how individuals share near-death experiences (NDEs), limited research has explored how these narratives are communicated collaboratively on today's social media platforms. We analyzed 200 randomly sampled TikTok videos tagged with #nde and related hashtags. Content analysis revealed that individuals often use NDE narratives to articulate personal meaning, with spiritual and religious themes appearing in the majority of posts and serving as a means of exploring and making sense of personal spiritual perspectives. Consistent with this, analyses of comment sections reveal that videos containing spiritual themes tend to attract more engagement and foster deeper conversations around faith and meaning. Our findings offer insights into how online platforms facilitate community-level engagement with spirituality, and suggest implications for design of spaces that support shared expression and connection in specialized communities.

cs.HC↗
arXiv2025

DNA-HHE: Dual-mode Near-network Accelerator for Hybrid Homomorphic Encryption on the Edge

Fully homomorphic encryption (FHE) schemes like RNS-CKKS enable privacy-preserving outsourced computation (PPOC) but suffer from high computational latency and ciphertext expansion, especially on the resource-constrained edge side. Hybrid Homomorphic Encryption (HHE) mitigates these issues on the edge side by replacing HE with lightweight symmetric encryption for plaintext encryption, such as the Rubato cipher for the HHE variant of RNS-CKKS, yet it introduces transciphering overhead on the cloud. The respective strengths and limitations of FHE and HHE call for a dual-mode HHE solution with flexible algorithm switching ability. This paper presents DNA-HHE, the first dual-mode HHE accelerator with near-network coupling for edge devices. DNA-HHE supports both edge-side RNS-CKKS and Rubato within a unified architecture driven by flexible custom instructions. To realize a compact implementation for the edge side, we propose a DSP-efficient modular reduction design, a compact multi-field-adaptive butterfly unit, and parallel scheduling schemes of Rubato with a high degree of resource sharing. DNA-HHE is designed with network protocol packaging and transmission capacities and directly coupled to the network interface controller, achieving reduced overall latency of edge-side PPOC by 1.09$\times$ to 1.56$\times$. Our evaluations on the ASIC and FPGA platforms demonstrate that DNA-HHE outperforms the state-of-the-art single-mode designs in both edge-side RNS-CKKS and symmetric cipher with better computation latency and area efficiency, while offering dual-mode functionality.

cs.CR↗
arXiv2025

Atomically-precise synthesis and simultaneous integration of 2D transition metal dichalcogenides enabled by nano-confinement

Two-dimensional (2D) materials, such as graphene, transition metal dichalcogenides (TMDs), and hBN, exhibit intriguing properties that are sensitive to their atomic-scale structures and can be further enriched through van der Waals (vdW) integration. However, the precise synthesis and clean integration of 2D materials remain challenging. Here, using graphene or hBN as a vdW capping layer, we create a nano-confined environment that directs the growth kinetics of 2D TMDs (e.g., NbSe2 and MoS2), enabling precise formation of TMD monolayers with tailored morphologies, from isolated monolayer domains to large-scale continuous films and intrinsically-patterned rings. Moreover, Janus S-Mo-Se monolayers are synthesized with atomic precision via vdW-protected bottom-plane chalcogen substitution. Importantly, our approach simultaneously produces ultraclean vdW interfaces. This in situ encapsulation reliably preserves air-sensitive materials, as evidenced by the enhanced superconductivity of nano-confined NbSe2 monolayers. Altogether, our study establishes a versatile platform for the controlled synthesis and integration of 2D TMDs for advanced applications.

cond-mat.mtrl-sci↗
arXiv2025

Towards Unified Co-Speech Gesture Generation via Hierarchical Implicit Periodicity Learning

Generating 3D-based body movements from speech shows great potential in extensive downstream applications, while it still suffers challenges in imitating realistic human movements. Predominant research efforts focus on end-to-end generation schemes to generate co-speech gestures, spanning GANs, VQ-VAE, and recent diffusion models. As an ill-posed problem, in this paper, we argue that these prevailing learning schemes fail to model crucial inter- and intra-correlations across different motion units, i.e. head, body, and hands, thus leading to unnatural movements and poor coordination. To delve into these intrinsic correlations, we propose a unified Hierarchical Implicit Periodicity (HIP) learning approach for audio-inspired 3D gesture generation. Different from predominant research, our approach models this multi-modal implicit relationship by two explicit technique insights: i) To disentangle the complicated gesture movements, we first explore the gesture motion phase manifolds with periodic autoencoders to imitate human natures from realistic distributions while incorporating non-period ones from current latent states for instance-level diversities. ii) To model the hierarchical relationship of face motions, body gestures, and hand movements, driving the animation with cascaded guidance during learning. We exhibit our proposed approach on 3D avatars and extensive experiments show our method outperforms the state-of-the-art co-speech gesture generation methods by both quantitative and qualitative evaluations. Code and models will be publicly available.

cs.AI↗
arXiv2025

CupCleaner: A Hybrid Data Cleaning Approach for Comment Updating

Comment updating is an emerging task in software evolution that aims to automatically revise source code comments in accordance with code changes. This task plays a vital role in maintaining code-comment consistency throughout software development. Recently, deep learning-based approaches have shown great potential in addressing comment updating by learning complex patterns between code edits and corresponding comment modifications. However, the effectiveness of these learning-based approaches heavily depends on the quality of training data. Existing datasets are typically constructed by mining version histories from open-source repositories such as GitHub, where there is often a lack of quality control over comment edits. As a result, these datasets may contain noisy or inconsistent samples that hinder model learning and generalization. In this paper, we focus on cleaning existing comment updating datasets, considering both the data's characteristics in the updating scenario and their implications on the model training process. We propose a hybrid statistical approach named CupCleaner (Comment UPdating's CLEANER) to achieve this purpose. Specifically, we combine static semantic information within data samples and dynamic loss information during the training process to clean the dataset. Experimental results demonstrate that, on the same test set, both the individual static strategy and the dynamic strategy can significantly filter out a portion of the data and enhance the performance of the model. Furthermore, employing a model ensemble approach can combine the characteristics of static and dynamic cleaning, further enhancing the performance of the model and the reliability of its output results.

cs.SE↗
arXiv2025

Directional Diffusion-Style Code Editing Pre-training

Code pre-trained models have shown promising effectiveness in various software engineering tasks. Among these tasks, many tasks are related to software evolution and/or code editing. However, existing code pre-trained models often overlook the real-world code editing data and the evolutionary nature of the editing process. In this paper, to simulate the step-by-step code editing process of human developers, we propose DivoT5, a pre-trained model based on directional diffusion at the data level. In DivoT5, we adopt two categories of pre-training tasks. The first category is mask and denoising tasks augmented with a diffusion direction representing code evolution. That is, we first apply a noising process to the code snippets before evolution, and then ask the pre-training process to restore the snippets with noise into the code snippets after evolution. The second category is tasks aiming to reinforce the evolutionary direction. That is, we first generate various intermediate versions for each pair of snippets before and after evolution, and then ask the pre-training process to transform the intermediate versions into the snippet after evolution for each pair. We evaluate DivoT5 for two code-editing scenarios and one non-editing scenario using five downstream tasks. Given each downstream task, we fine-tune the pre-trained DivoT5 to evaluate its effectiveness. Our experimental results show that DivoT5 achieves state-of-the-art (SOTA) performance on most tasks in comparison to models of the same scale (220M), large scale (770M) models in fine-tuning, and billion-scale (6.7B, 8B, ChatGPT) models in few-shot settings. For one code-editing task (i.e., automated code review), DivoT5 pre-trained on top of CodeT5-small (60M) can even outperform CodeT5-base (220M) and other pre-trained models with 220M parameters except for DivoT5 pre-trained on top of CodeT5-base (220M).

cs.SE↗