arXiv subjects

Haoran Qin

Publications and source records attributed to Haoran Qin.

12 recordsLinked to original sources

arXiv2026

LinCa: Accelerating Diffusion Models via Learnable Decomposed Feature Caching

Diffusion models have achieved remarkable success in image and video generation, yet the high computational cost of iterative sampling remains a critical bottleneck for practical deployment. Feature caching has emerged as a promising acceleration paradigm by reusing or predicting intermediate features across timesteps. However, existing training-free methods apply uniform prediction strategies that cannot adapt to the heterogeneous feature dynamics, causing significant quality degradation under high acceleration ratios. We propose LinCa, a feature caching framework based on learnable invertible networks. LinCa decomposes cached features into sub-components with distinct continuity properties via a lightweight invertible network and applies differentiated prediction orders matched to each component. The strict invertibility guarantees lossless reconstruction back to the original feature space, forming a unified Decompose-Predict-Reconstruct pipeline. By training separate predictors for different models and timestep segments, LinCa adapts to heterogeneous feature dynamics. Experiments on FLUX, Qwen-Image, and HunyuanVideo demonstrate that LinCa, with less than 0.2% additional parameters, significantly outperforms existing methods and maintains near-lossless quality at 5-7x speedup. Code: https://github.com/QHR69/LinCa

cs.CV↗

arXiv2026

AViTS: Adaptive Spatiotemporal Token Selection for Efficient Dynamic-Resolution Generation

Diffusion Transformers (DiTs) achieve high-quality generation but are costly due to iterative sampling. Dynamic-resolution sampling reduces early-stage cost by denoising at low resolution; however, uniformly upsampling all latent tokens at resolution transitions incurs redundant computation and may degrade fine-detail consistency. Existing partial upsampling strategies typically rely on local latent structure cues or single-step statistics, making it difficult to jointly capture token-text semantic relevance and token-wise representation dynamics across diffusion steps. We propose AViTS, an adaptive spatiotemporal token selection framework for dynamic-resolution DiTs. AViTS models spatial importance via latent-text attention and temporal importance via token-level feature variation across diffusion timesteps, and fuses them to enable spatiotemporal importance-aware selective upsampling: it prioritizes resolution refinement for critical tokens while deferring less important ones, thereby reducing redundant high-resolution computation and improving the quality-efficiency trade-off. AViTS achieves up to 6.34x on FLUX and nearly 9x FLOPs reduction on Qwen-Image-Edit and FLUX.1-Kontext-dev, orthogonal to distillation, quantization, and feature caching, and reaching 14.76x with distilled models. Code: https://github.com/QHR69/AViTS

cs.CV↗

arXiv2026

IO-SVD: Input-Output Whitened SVD for Adaptive-Rank LLM Compression

Large language models deliver strong performance across language and reasoning tasks, but their storage and compute costs remain major barriers to deployment in resource-constrained and latency-sensitive settings. SVD-based post-training compression offers a hardware-agnostic way to reduce model size and improve inference efficiency through low-rank factorization. However, existing methods often rely on input-only whitening spaces, homogeneous rank allocation, or loss-agnostic allocation heuristics, limiting their ability to preserve model quality under aggressive compression. We propose Input-Output Whitened SVD (IO-SVD), a post-training compression method that forms a KL-aware double-sided whitening space for model weights. Using a second-order expansion of the KL loss over the top-K token probabilities, IO-SVD constructs an output-side metric that captures predictive sensitivity, while input whitening captures activation statistics. We further introduce an efficient heterogeneous rank-allocation strategy that scores whitened singular components using first-order calibration loss estimates and prunes the least sensitive components under a global budget. Inspired by prior work that combines SVD truncation with quantization, we improve hybrid SVD-quantization compression through loss-aware remapping, which selects low-rank factor rows for 8-bit quantization based on the predicted loss change incurred by quantizing them. Extensive experiments across diverse LLM and VLM families, and inference-time analysis shows that IO-SVD compresses LLMs with minimal performance degradation while delivering practical inference speedups. Code is available at https://github.com/mint-vu/IO-SVD.git

cs.LG↗

arXiv2026

SpecEdit: Training-Free Acceleration for Diffusion based Image Editing via Semantic Locking

Diffusion-based image editing offers strong semantic controllability, but remains computationally expensive due to iterative high-resolution denoising over all spatial tokens. Dynamic-resolution sampling reduces this cost by performing early steps at reduced resolution. However, existing approaches prioritize upsampling using low-level heuristics such as edge detection or channel variance, which are weakly aligned with editing semantics and may lead to structural inconsistency. Moreover, spatial regions are often upsampled without verifying whether semantic modification is actually required, resulting in redundant high-resolution computation and accumulated errors. Therefore, we propose SpecEdit, a training-free dynamic-resolution framework tailored for diffusion-based image editing. SpecEdit follows a draft-and-verify scheme: a low-resolution draft first estimates the semantic outcome, after which token-level discrepancies are used to identify edit-relevant tokens for high-resolution denoising, while the remaining tokens stay at a coarse resolution. Experiments on Qwen-Image-Edit and FLUX.1-Kontext-dev demonstrate up to 10x and 7x acceleration, while maintaining strong quality. SpecEdit is complementary to step distillation and other acceleration techniques, achieving up to 13x speedup when combined with existing methods. Our code is in supplementary material and will be released on GitHub.

cs.CV↗

arXiv2026

Zero Sum SVD: Balancing Loss Sensitivity for Low Rank LLM Compression

Advances in large language models have driven strong performance across many tasks, but their memory and compute costs still hinder deployment. SVD-based compression reduces storage and can speed up inference via low-rank factors, yet performance depends on how rank is allocated under a global compression ratio. Prior methods often use homogeneous ranks for similarly sized matrices, despite large differences in loss sensitivity, or rely on expensive iterative pre-truncation optimization to determine per matrix ranks. We propose \textbf{Zero Sum SVD} (\textbf{ZS-SVD}), a post-training method that performs \emph{global} singular component selection using activation whitening and first-order calibration loss estimates in whitened coordinates. \textbf{ZS-SVD} prunes components across the whole model with a \textbf{zero sum} rule that keeps the cumulative predicted loss change near zero, automatically yielding heterogeneous ranks without solving a rank allocation optimization. Motivated by evidence that gradients near pretrained solutions exhibit low rank structure, we also introduce an optional lightweight correction that applies a \textbf{single} projected gradient update after truncation, followed by re-truncation. Extensive experiments across multiple LLM architectures show consistent gains across diverse benchmarks and compression ratios. Code is available at https://github.com/mint-vu/Zero-Sum-SVD

cs.LG↗

arXiv2025

Low-Rank Prehab: Preparing Neural Networks for SVD Compression

Low-rank approximation methods such as singular value decomposition (SVD) and its variants (e.g., Fisher-weighted SVD, Activation SVD) have recently emerged as effective tools for neural network compression. In this setting, decomposition acts as a "surgical" intervention, followed by fine-tuning that serves as "rehab" to recover accuracy. Inspired by prehabilitation in surgery, we introduce a pre-compression fine-tuning stage, Low-Rank Prehab, that explicitly encourages low-rank structure in weight matrices while preserving task performance. By conditioning the model before SVD, Prehab steers weights toward spectrally compact regions of the parameter space, enabling smoother low-rank approximation and improved recovery. Experiments on large language models (LLMs) and other Transformer-based architectures, including Vision Transformers (ViTs), show that Prehab substantially reduces the immediate accuracy drop after compression and consistently improves post-finetuning performance. Across a wide range of compression ratios, our method outperforms state-of-the-art SVD-based techniques such as SVD-LLM, highlighting the importance of preparing models for compression rather than only improving the compression and recovery stages. Source code is available at https://github.com/niqretnuh/PREHAB-SVD

cs.LG↗

arXiv2025

Analysis and Algorithm for Multi IRS Collaborative Localization via Hybrid Time Angle Estimation

This paper proposes a novel multiple intelligent reflecting surfaces (IRSs) collaborative hybrid localization system, which involves deploying multiple IRSs near the target area and achieving target localization through joint time delay and angle estimation. Specifically, echo signals from all reflective elements are received by each sensor and jointly processed to estimate the time delay and angle parameters. Based on the above model, we derive the Fisher Information Matrix (FIM) for cascaded delay, Angle of Arrival (AOA), and Angle of Departure (AOD) estimation in semi passive passive models, along with the corresponding Cramer Rao Bound (CRB). To achieve precise estimation close to the CRB, we design efficient algorithms for angle and location estimation. For angle estimation, reflective signals are categorized into three cases based on their rank, with different signal preprocessing. By constructing an atomic norm set and minimizing the atomic norm, the joint angle estimation problem is transformed into a convex optimization problem, and low-complexity estimation of multiple AOA and AOD pairs is achieved using the Alternating Direction Method of Multipliers (ADMM). For location estimation, we propose a three-stage localization algorithm that combines weighted least squares, total least squares, and quadratic correction to handle errors in the coefficient matrix and observation vector, thus improving accuracy. Numerical simulations validate the superiority of the proposed system, demonstrating that the system's collaboration, hybrid localization, and distributed deployment provide substantial benefits, as well as the accuracy of the proposed estimation algorithms, particularly in low signal to noise ratio (SNR) condition.

eess.SP↗

arXiv2025

Towards Automated Smart Contract Generation: Evaluation, Benchmarking, and Retrieval-Augmented Repair

Smart contracts, predominantly written in Solidity and deployed on blockchains such as Ethereum, are immutable after deployment, making functional correctness critical. However, existing evaluations of Solidity code generation rely largely on surface-level metrics (e.g., BLEU, CrystalBLEU) or manual inspection, which correlate poorly with functional correctness. In contrast to Python, Solidity lacks large-scale, execution-based benchmarks, limiting systematic evaluation of large language models for smart contract development. We introduce SolBench, a comprehensive benchmark and automated testing pipeline for Solidity that emphasizes functional correctness via differential fuzzing. SolBench consists of 28825 functions extracted from 7604 real-world smart contracts collected from Etherscan (genesis-2024), spanning ten application domains. We benchmark 14 diverse LLMs, covering open and closed models, 1.3B-671B parameters, and both general-purpose and code-specialized architectures. The dominant failure mode is missing critical intra-contract information, such as state variables and type definitions. Providing full-contract context improves accuracy but incurs prohibitive inference costs. To address this, we propose Retrieval-Augmented Repair (RAR), a cost-effective framework that integrates execution feedback into code repair. RAR uses compiler and runtime error messages to retrieve only the minimal contract snippets needed to correct a target function, avoiding full-context inference. This significantly reduces input length while improving functional correctness. We further analyze retrieval and repair strategies within RAR, demonstrating consistent gains in accuracy and efficiency. SolBench and RAR enable principled, execution-based evaluation and economical improvement of Solidity code generation. Dataset and code are publicly available at https://github.com/ZaoyuChen/SolBench.

cs.SE↗

arXiv2024

Towards TMA-Based Transmissive RIS Transceiver Enabled Downlink Communication Networks: A Consensus-ADMM Approach

This paper presents a novel multi-stream downlink communication system that utilizes a transmissive reconfigurable intelligent surface (RIS) transceiver. Specifically, we elaborate the downlink communication scheme using time-modulated array (TMA) technology, which enables high order modulation and multi-stream beamforming. Then, an optimization problem is formulated to maximize the minimum signal-to-interference-plusnoise ratio (SINR) with user fairness, which takes into account the constraint of the maximum available power for each transmissive element. Due to the non-convex nature of the formulated problem,finding optimal solution is challenging. To mitigate the complexity,we propose a linear-complexity beamforming algorithm based on consensus alternating direction method of multipliers (ADMM).Specifically, by introducing a set of auxiliary variables, the problem can be decomposed into multiple sub-problems that are amenable to parallel computation, where the each sub-problem can yield closed-form expressions, bringing a significant reduction in the computational complexity. The overall problem achieves convergence by iteratively addressing these sub-problems in an alternating manner. Finally, the convergence of the proposed algorithm and the impact of various parameter configurations on the system performance are validated through numerical simulations.

eess.SP↗

arXiv2024

Cramer-Rao Bound Minimization for Movable Antenna-Assisted Multiuser Integrated Sensing and Communications

This paper investigates a movable antenna (MA)-assisted multiuser integrated sensing and communication (ISAC) system, where the base station (BS) and communication users are all equipped with MA for improving both the sensing and communication performance. We employ the Cramer-Rao bound (CRB) as the performance metric of sensing, thus a joint beamforming design and MAs' position optimizing problem is formulated to minimize the CRB. However the resulting optimization problem is NP-hard and the variables are highly coupled. To tackle this problem, we propose an alternating optimization (AO) framework by adopting semidefinite relaxation (SDR) and successive convex approximation (SCA) technique. Numerical results reveal that the proposed MA-assisted ISAC system achieves lower estimation CRB compared to the fixed-position antenna (FPA) counterpart.

eess.SP↗

arXiv2024

Antenna Positioning and Beamforming Design for Fluid-Antenna Enabled Multi-user Downlink Communications

This paper investigates a multiple input single output (MISO) downlink communication system in which users are equipped with fluid antennas (FAs). First, we adopt a field-response based channel model to characterize the downlink channel with respect to FAs' positions. Then, we aim to minimize the total transmit power by jointly optimizing the FAs' positions and beamforming matrix. To solve the resulting non-convex problem, we employ an alternating optimization (AO) algorithm based on penalty method and successive convex approximation (SCA) to obtain a sub-optimal solution. Numerical results demonstrate that the FA-assisted communication system performs better than conventional fixed position antennas system.

cs.IT↗

arXiv2020

Optimized Loss Functions for Object detection: A Case Study on Nighttime Vehicle Detection

Loss functions is a crucial factor that affecting the detection precision in object detection task. In this paper, we optimize both two loss functions for classification and localization simultaneously. Firstly, by multiplying an IoU-based coefficient by the standard cross entropy loss in classification loss function, the correlation between localization and classification is established. Compared to the existing studies, in which the correlation is only applied to improve the localization accuracy for positive samples, this paper utilizes the correlation to obtain the really hard negative samples and aims to decrease the misclassified rate for negative samples. Besides, a novel localization loss named MIoU is proposed by incorporating a Mahalanobis distance between predicted box and target box, which eliminate the gradients inconsistency problem in the DIoU loss, further improving the localization accuracy. Finally, sufficient experiments for nighttime vehicle detection have been done on two datasets. Our results show than when train with the proposed loss functions, the detection performance can be outstandingly improved. The source code and trained models are available at https://github.com/therebellll/NegIoU-PosIoU-Miou.

cs.CV↗