arXiv subjects

Shen Fu

Publications and source records attributed to Shen Fu.

8 recordsLinked to original sources

arXiv2026

DeaMoE: Efficient MoE Structure for Fast Small-Batch Decoding

Mixture-of-Experts (MoE) models have been widely adopted in real-time interactive applications such as coding assistants, real-time audio-video interaction systems. To meet the extremely low response latency requirements of these scenarios, practitioners commonly employ small-batch decoding, under which MoE inference becomes memory-bound and is severely bottlenecked by expert weight loading. However, this bottleneck has received limited attention, and existing solutions such as post-training weight compression or fine-grained expert design during pre-training either degrade model accuracy or introduce additional computation and communication overhead. To tackle this issue, we propose DeaMoE, a decoding-efficient MoE architecture, in which the experts are grouped into several departments, and the experts belonging to the same department share most parameters since they come from the same professional field, and additionally each expert contains a few private parameters to reflect its uniqueness. Moreover, we design customized two-stage routing strategy for DeaMoE to avoid redundant loading, under which DeaMoE greatly improves the efficiency during LLM decoding. Compared with vanilla MoE, DeaMoE reduces per-step loaded weights by up to 50.9% and achieves up to 1.33 end-to-end TPOT speedup for the pre-trained 7B model on A40, and up to 2.00x and 1.97x peak speedup for DeepSeek-V3 on A40 and H100 in microbenchmarks.

cs.LG↗

arXiv2026

Channel Knowledge Map Construction via Guided Flow Matching

The efficient construction of accurate channel knowledge maps (CKMs) is crucial for unleashing the full potential of environment-aware wireless networks, yet it remains a difficult ill-posed problem due to the sparsity of available location-specific channel knowledge data. Although diffusion-based methods such as denoising diffusion probabilistic models (DDPMs) have been exploited for CKM construction, they rely on iterative stochastic sampling, rendering them too slow for real-time wireless applications. To bridge the gap between high fidelity and efficient CKM construction, this letter introduces a novel framework based on linear transport guided flow matching (LT-GFM). Deviating from the noise-removal paradigm of diffusion models, our approach models the CKM generation process as a deterministic ordinary differential equation (ODE) that follows linear optimal transport paths, thereby drastically reducing the number of required inference steps. We propose a unified architecture that is applicable to not only the conventional channel gain map (CGM) construction, but also the more challenging spatial correlation map (SCM) construction. To achieve physics-informed CKM constructions, we integrate environmental semantics (e.g., building masks) for edge recovery and enforce Hermitian symmetry for property of the SCM. Simulation results verify that LT-GFM achieves superior distributional fidelity with significantly lower Fr\'echet Inception Distance (FID) and accelerates inference speed by a factor of 25 compared to DDPMs.

cs.IT↗

arXiv2025

CKMDiff: A Generative Diffusion Model for CKM Construction via Inverse Problems with Learned Priors

Channel knowledge map (CKM) is a promising technology to enable environment-aware wireless communications and sensing with greatly enhanced performance, by offering location-specific channel prior information for future wireless networks. One fundamental problem for CKM-enabled wireless systems lies in how to construct high-quality and complete CKM for all locations of interest, based on only limited and noisy on-site channel knowledge data. This problem resembles the long-standing ill-posed inverse problem, which tries to infer from a set of limited and noisy observations the cause factors that produced them. By utilizing the recent advances of solving inverse problems with learned priors using generative artificial intelligence (AI), we propose CKMDiff, a conditional diffusion model that can be applied to perform various tasks for CKM constructions such as denoising, inpainting, and super-resolution, without having to know the physical environment maps or transceiver locations. Furthermore, we propose an environment-aware data augmentation mechanism to enhance the model's ability to learn implicit relations between electromagnetic propagation patterns and spatial-geometric features. Extensive numerical results are provided based on the CKMImageNet and RadioMapSeer datasets, which demonstrate that the proposed CKMDiff achieves state-of-the-art performance, outperforming various benchmark methods.

eess.SP↗

arXiv2025

CKMImageNet: A Dataset for AI-Based Channel Knowledge Map Towards Environment-Aware Communication and Sensing

With the increasing demand for real-time channel state information (CSI) in sixth-generation (6G) mobile communication networks, channel knowledge map (CKM) emerges as a promising technique, offering a site-specific database that enables environment-awareness and significantly enhances communication and sensing performance by leveraging a priori wireless channel knowledge. However, efficient construction and utilization of CKMs require high-quality, massive, and location-specific channel knowledge data that accurately reflects the real-world environments. Inspired by the great success of ImageNet dataset in advancing computer vision and image understanding in artificial intelligence (AI) community, we introduce CKMImageNet, a dataset developed to bridge AI and environment-aware wireless communications and sensing by integrating location-specific channel knowledge data, high-fidelity environmental maps, and their visual representations. CKMImageNet supports a wide range of AI-driven approaches for CKM construction with spatially consistent and location-specific channel knowledge data, including both supervised and unsupervised, as well as discriminative and generative AI methods.

eess.SP↗

arXiv2024

Generative CKM Construction using Partially Observed Data with Diffusion Model

Channel knowledge map (CKM) is a promising technique that enables environment-aware wireless networks by utilizing location-specific channel prior information to improve communication and sensing performance. A fundamental problem for CKM construction is how to utilize partially observed channel knowledge data to reconstruct a complete CKM for all possible locations of interest. This problem resembles the long-standing ill-posed inverse problem, which tries to infer from a set of limited observations the cause factors that produced them. By utilizing the recent advances of solving inverse problems with generative artificial intelligence (AI), in this paper, we propose generative CKM construction method using partially observed data by solving inverse problems with diffusion models. Simulation results show that the proposed method significantly improves the performance of CKM construction compared with benchmarking schemes.

eess.SP↗

arXiv2024

CKMImageNet: A Comprehensive Dataset to Enable Channel Knowledge Map Construction via Computer Vision

Environment-aware communication and sensing is one of the promising paradigm shifts towards 6G, which fully leverages prior information of the local wireless environment to optimize network performance. One of the key enablers for environment-aware communication and sensing is channel knowledge map (CKM), which provides location-specific channel knowledge that is crucial for channel state information (CSI) acquisition. To support the efficient construction of CKM, large-scale location-specific channel data is essential. However, most existing channel datasets do not have the location information nor visual representations of channel data, making them inadequate for exploring the intrinsic relationship between the channel knowledge and the local environment, nor for applying advanced artificial intelligence (AI) algorithms such as computer vision (CV) for CKM construction. To address such issues, in this paper, a large-scale dataset named CKMImageNet is established, which can provide both location-tagged numerical channel data and visual images, providing a holistic view of the channel and environment. Built using commercial ray tracing software, CKMImageNet captures electromagnetic wave propagation in different scenarios, revealing the relationships between location, environment and channel knowledge. By integrating detailed channel data and the corresponding image, CKMImageNet not only supports the verification of various communication and sensing algorithms, but also enables CKM construction with CV algorithms.

cs.IT↗

arXiv2023

A Comprehensive and Reliable Feature Attribution Method: Double-sided Remove and Reconstruct (DoRaR)

The limited transparency of the inner decision-making mechanism in deep neural networks (DNN) and other machine learning (ML) models has hindered their application in several domains. In order to tackle this issue, feature attribution methods have been developed to identify the crucial features that heavily influence decisions made by these black box models. However, many feature attribution methods have inherent downsides. For example, one category of feature attribution methods suffers from the artifacts problem, which feeds out-of-distribution masked inputs directly through the classifier that was originally trained on natural data points. Another category of feature attribution method finds explanations by using jointly trained feature selectors and predictors. While avoiding the artifacts problem, this new category suffers from the Encoding Prediction in the Explanation (EPITE) problem, in which the predictor's decisions rely not on the features, but on the masks that selects those features. As a result, the credibility of attribution results is undermined by these downsides. In this research, we introduce the Double-sided Remove and Reconstruct (DoRaR) feature attribution method based on several improvement methods that addresses these issues. By conducting thorough testing on MNIST, CIFAR10 and our own synthetic dataset, we demonstrate that the DoRaR feature attribution method can effectively bypass the above issues and can aid in training a feature selector that outperforms other state-of-the-art feature attribution methods. Our code is available at https://github.com/dxq21/DoRaR.

cs.LG↗

arXiv2021

Nanoparticle Radiosensitization: from extended local effect modeling to a survival modificationframework of compound Poisson additive killing and its carbon dots validation

Objective: To construct an analytical model instead of local effect modeling for the prediction of the biological effectiveness of nanoparticle radiosensitization. Approach: An extended local effects model is first proposed with a more comprehensive description of the nanoparticles mediated local killing enhancements, but meanwhile puts forward challenging issues that remain difficult and need to be further studied. As a novel method instead of local effect modeling, a survival modification framework of compound Poisson additive killing is proposed, as the consequence of an independent additive killing by the assumed equivalent uniform doses of individual nanoparticles per cell under the LQ model. A compound Poisson killing (CPK)model based on the framework is thus derived, giving a general expression of nanoparticle mediated LQ parameter modification. For practical use, a simplified form of the model is also derived, as a concentration dependent correction only to the {\alpha} parameter, with the relative correction (alpha"/alpha)) dominated by the mean number, and affected by the agglomeration of nanoparticles per cell. For different agglomeration state, a monodispersion model of the dispersity factor {\eta}=1, and an agglomeration model of 2/3<{\eta}<1,are provided for practical prediction of (alpha"/alpha) value respectively. Main results: Initial validation by the radiosensitization ofHepG2 cells by carbon dots showed a high accuracy of the CPK model. In a safe range of concentration (0.003-0.03 {\mu}g/{\mu}L)of the carbon dots, the prediction errors of the monodispersion and agglomeration models were both within 2%, relative to the clonogenic survival data of the sensitized HepG2 cells.

physics.med-ph↗