arXiv subjects

Zeyu Huang

Publications and source records attributed to Zeyu Huang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Identification of RIS-Assisted Paths for Wireless Integrated Sensing and Communication

Distinguishing between reconfigurable intelligent surface (RIS) assisted paths and non-line-of-sight (NLOS) paths is a fundamental problem for RIS-assisted integrated sensing and communication. In this work, we propose a pattern alternation scheme for the RIS response that uses part of the RIS as a dynamic part to modulate the estimated channel power, which can considerably help the user equipments (UEs) to identify the RIS-assisted paths. Under such a dynamic setup, we formulate the detection framework for a single UE, where we develop a statistical model of the estimated channel power, allowing us to analytically evaluate the performance of the system. We investigate our method under two critical factors: the number of RIS elements allocated for the dynamic part and the allocation of RIS elements among different users. Simulation results verify the accuracy of our analysis.

eess.SP↗

arXiv2025

Leveraging Large Reconfigurable Intelligent Surfaces as Anchors for Near-Field Positioning

In this work, we present a recent investigation on leveraging large reconfigurable intelligent surfaces (RIS) as anchors for positioning in wireless communication systems. Unlike existing approaches, we explicitly address the uncertainty arising from the substantial physical size of the RIS, particularly relevant when a user equipment resides in the near field, and propose a method that ensures accurate positioning under these conditions. We derive the corresponding Cramer-Rao bound for our scheme and validate the effectiveness of our scheme through numerical experiments, highlighting both the feasibility and potential of our approach.

cs.ET↗

arXiv2025

Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free

Gating mechanisms have been widely utilized, from early models like LSTMs and Highway Networks to recent state space models, linear attention, and also softmax attention. Yet, existing literature rarely examines the specific effects of gating. In this work, we conduct comprehensive experiments to systematically investigate gating-augmented softmax attention variants. Specifically, we perform a comprehensive comparison over 30 variants of 15B Mixture-of-Experts (MoE) models and 1.7B dense models trained on a 3.5 trillion token dataset. Our central finding is that a simple modification-applying a head-specific sigmoid gate after the Scaled Dot-Product Attention (SDPA)-consistently improves performance. This modification also enhances training stability, tolerates larger learning rates, and improves scaling properties. By comparing various gating positions and computational variants, we attribute this effectiveness to two key factors: (1) introducing non-linearity upon the low-rank mapping in the softmax attention, and (2) applying query-dependent sparse gating scores to modulate the SDPA output. Notably, we find this sparse gating mechanism mitigates 'attention sink' and enhances long-context extrapolation performance, and we also release related $\href{https://github.com/qiuzh20/gated_attention}{codes}$ and $\href{https://huggingface.co/QwQZh/gated_attention}{models}$ to facilitate future research.

cs.CL↗

arXiv2025

Layerwise Recurrent Router for Mixture-of-Experts

The scaling of large language models (LLMs) has revolutionized their capabilities in various tasks, yet this growth must be matched with efficient computational strategies. The Mixture-of-Experts (MoE) architecture stands out for its ability to scale model size without significantly increasing training costs. Despite their advantages, current MoE models often display parameter inefficiency. For instance, a pre-trained MoE-based LLM with 52 billion parameters might perform comparably to a standard model with 6.7 billion parameters. Being a crucial part of MoE, current routers in different layers independently assign tokens without leveraging historical routing information, potentially leading to suboptimal token-expert combinations and the parameter inefficiency problem. To alleviate this issue, we introduce the Layerwise Recurrent Router for Mixture-of-Experts (RMoE). RMoE leverages a Gated Recurrent Unit (GRU) to establish dependencies between routing decisions across consecutive layers. Such layerwise recurrence can be efficiently parallelly computed for input tokens and introduces negotiable costs. Our extensive empirical evaluations demonstrate that RMoE-based language models consistently outperform a spectrum of baseline models. Furthermore, RMoE integrates a novel computation stage orthogonal to existing methods, allowing seamless compatibility with other MoE architectures. Our analyses attribute RMoE's gains to its effective cross-layer information sharing, which also improves expert selection and diversity. Our code is at https://github.com/qiuzh20/RMoE .

cs.CL↗

arXiv2025

Scintillation and Timing Performance of a 3at% Yttrium-Doped Barium Fluoride Crystal

We report the scintillation and timing performance of a new developed 200 * 20 mm * 20 mm large size barium fluoride crystal doped with 3at% yttrium (BaF2:Y) to enhance the application for high time resolution. This doping effectively suppresses the slow scintillation component while maintaining most of the fast component, as confirmed by X-ray excited luminescence measurements. The BaF2:Y crystal demonstrated a transmittance of near 90% in the visible spectrum and a light response uniformity parameter of delta = (-2.74 +- 1.15)% when coupled with the tail end. The actual yttrium content varied from 2.1at% near the seed end to 3.7at% at the tail end. The assembled large BaF2:Y detector with silicon photomultipliers exhibited a time resolution of (82.2 +- 2.6) ps using constant fraction discrimination method in a cosmic ray test and (140.1 +- 3.8) ps using a low fixed threshold method in a beam test at Shanghai Synchrotron Radiation Facility with an 1.35 GeV electron beam. These results indicate the significant potential of BaF2:Y crystal for various applications, such as detectors for particle physics and nuclear physics.

physics.ins-det↗

arXiv2025

Demons in the Detail: On Implementing Load Balancing Loss for Training Specialized Mixture-of-Expert Models

This paper revisits the implementation of $\textbf{L}$oad-$\textbf{b}$alancing $\textbf{L}$oss (LBL) when training Mixture-of-Experts (MoEs) models. Specifically, LBL for MoEs is defined as $N_E \sum_{i=1}^{N_E} f_i p_i$, where $N_E$ is the total number of experts, $f_i$ represents the frequency of expert $i$ being selected, and $p_i$ denotes the average gating score of the expert $i$. Existing MoE training frameworks usually employ the parallel training strategy so that $f_i$ and the LBL are calculated within a $\textbf{micro-batch}$ and then averaged across parallel groups. In essence, a micro-batch for training billion-scale LLMs normally contains very few sequences. So, the micro-batch LBL is almost at the sequence level, and the router is pushed to distribute the token evenly within each sequence. Under this strict constraint, even tokens from a domain-specific sequence ($\textit{e.g.}$, code) are uniformly routed to all experts, thereby inhibiting expert specialization. In this work, we propose calculating LBL using a $\textbf{global-batch}$ to loose this constraint. Because a global-batch contains much more diverse sequences than a micro-batch, which will encourage load balance at the corpus level. Specifically, we introduce an extra communication step to synchronize $f_i$ across micro-batches and then use it to calculate the LBL. Through experiments on training MoEs-based LLMs (up to $\textbf{42.8B}$ total parameters and $\textbf{400B}$ tokens), we surprisingly find that the global-batch LBL strategy yields excellent performance gains in both pre-training perplexity and downstream tasks. Our analysis reveals that the global-batch LBL also greatly improves the domain specialization of MoE experts.

cs.LG↗

arXiv2024

Be There, Be Together, Be Streamed! AR Scenic Live-Streaming for an Interactive and Collective Experience

Scenic Live-Streaming (SLS), capturing real-world scenic sites from fixed cameras without streamers, combines scene immersion and the social and real-time characteristics of live-streaming into a unique experience. However, existing SLS affords limited audience interactions to engage them in a collective experience compared to many other live-streaming genres. It is also difficult for SLS to recreate important but intangible constituents of in-person trip experiences, such as cultural activities. To offer a more interactive, engaging, and meaningful experience, we propose ARSLS (Augmented Reality Scenic Live-Streaming). Culturally grounded AR objects with awareness of the live-streamed environment can be overlaid over camera views to provide additional interactive features while maintaining consistency with the live-streamed scene. To explore the design space of this new medium, we developed an ARSLS prototype for a famous landscape in China. A preliminary study (N=15) provided initial insights for ARSLS design.

cs.HC↗

arXiv2024

Gravitational waves and cosmic boundary

Space-based gravitational wave detectors have the capability to detect signals from very high redshifts. It is interesting to know if such capability can be used to study the global structure of the cosmic space. In this paper, we focus on one particular question: if there exists a reflective cosmic boundary at the high redshift ($z>15$), is it possible to find it? We find that, with the current level of technology: 1) gravitational waves appear to be the only means with which that signatures from the cosmic boundary can possibly be detected; 2) a large variety of black holes, with masses roughly in the range $(10^3\sim 10^6) {\rm~M_\odot}$, can be used for the task; 3) in the presumably rare but physically possible case that two merger events from the growth history of a massive black hole are detected coincidentally, a detector network like TianQin+LISA is essential in help improving the chance to determine the orientation of the cosmic boundary; 4) the possibility to prove or disprove the presence of the cosmic boundary largely depends on how likely one can detect multiple pairs of coincident gravitational wave events.

gr-qc↗

arXiv2024

Hybrid Channel Modeling and Environment Reconstruction for Terahertz Monostatic Sensing

THz ISAC aims to integrate novel functionalities, such as positioning and environmental sensing, into communication systems. Accurate channel modeling is crucial for the design and performance evaluation of future ISAC systems. In this paper, a THz measurement campaign for monostatic sensing is presented. VNA-based channel measurements are conducted in a laboratory scenario, where the transmitter and receiver are positioned together to mimic monostatic sensing. The centering frequency and measured bandwidth for these measurements are 300 GHz and 20 GHz, respectively. A DSS scheme is employed to capture spatial sensing channel profiles. Measurements are conducted across 28 transceiver locations arranged along an 'L'-shaped route. Then, an element-wise SAGE algorithm is used to estimate the MPC parameters, i.e., amplitude and delay. Specular and diffuse reflections are analyzed based on geometric principles and the estimated MPC parameters, where the effects from the radiation pattern are observed. A geometry-based MPC trajectory tracking algorithm is then proposed to classify the MPCs and de-embed the effects of the radiation pattern. Following this algorithm, a hybrid channel model is proposed based on the de-embedded MPC parameters. In this hybrid channel model for monostatic sensing, the MPCs are categorized into target-related and environment-related components. The target-related components are utilized for target detection and identification, while the environment-related ones focus on geometrical scenario reconstruction. A demonstration of geometrical environment reconstruction, along with an analysis of reflection loss for target identification, is subsequently presented. This work offers valuable insights into THz monostatic sensing channel modeling and the design of future THz ISAC systems.

eess.SP↗

arXiv2024

Stacking Your Transformers: A Closer Look at Model Growth for Efficient LLM Pre-Training

LLMs are computationally expensive to pre-train due to their large scale. Model growth emerges as a promising approach by leveraging smaller models to accelerate the training of larger ones. However, the viability of these model growth methods in efficient LLM pre-training remains underexplored. This work identifies three critical $\underline{\textit{O}}$bstacles: ($\textit{O}$1) lack of comprehensive evaluation, ($\textit{O}$2) untested viability for scaling, and ($\textit{O}$3) lack of empirical guidelines. To tackle $\textit{O}$1, we summarize existing approaches into four atomic growth operators and systematically evaluate them in a standardized LLM pre-training setting. Our findings reveal that a depthwise stacking operator, called $G_{\text{stack}}$, exhibits remarkable acceleration in training, leading to decreased loss and improved overall performance on eight standard NLP benchmarks compared to strong baselines. Motivated by these promising results, we conduct extensive experiments to delve deeper into $G_{\text{stack}}$ to address $\textit{O}$2 and $\textit{O}$3. For $\textit{O}$2 (untested scalability), our study shows that $G_{\text{stack}}$ is scalable and consistently performs well, with experiments up to 7B LLMs after growth and pre-training LLMs with 750B tokens. For example, compared to a conventionally trained 7B model using 300B tokens, our $G_{\text{stack}}$ model converges to the same loss with 194B tokens, resulting in a 54.6\% speedup. We further address $\textit{O}$3 (lack of empirical guidelines) by formalizing guidelines to determine growth timing and growth factor for $G_{\text{stack}}$, making it practical in general LLM pre-training. We also provide in-depth discussions and comprehensive ablation studies of $G_{\text{stack}}$. Our code and pre-trained model are available at https://llm-stacking.github.io.

cs.CL↗

arXiv2024

Unlocking Continual Learning Abilities in Language Models

Language models (LMs) exhibit impressive performance and generalization capabilities. However, LMs struggle with the persistent challenge of catastrophic forgetting, which undermines their long-term sustainability in continual learning (CL). Existing approaches usually address the issue by incorporating old task data or task-wise inductive bias into LMs. However, old data and accurate task information are often unavailable or costly to collect, hindering the availability of current CL approaches for LMs. To address this limitation, we introduce $\textbf{MIGU}$ ($\textbf{M}$agn$\textbf{I}$tude-based $\textbf{G}$radient $\textbf{U}$pdating for continual learning), a rehearsal-free and task-label-free method that only updates the model parameters with large magnitudes of output in LMs' linear layers. MIGU is based on our observation that the L1-normalized magnitude distribution of the output in LMs' linear layers is different when the LM models deal with different task data. By imposing this simple constraint on the gradient update process, we can leverage the inherent behaviors of LMs, thereby unlocking their innate CL abilities. Our experiments demonstrate that MIGU is universally applicable to all three LM architectures (T5, RoBERTa, and Llama2), delivering state-of-the-art or on-par performance across continual finetuning and continual pre-training settings on four CL benchmarks. For example, MIGU brings a 15.2% average accuracy improvement over conventional parameter-efficient finetuning baselines in a 15-task CL benchmark. MIGU can also seamlessly integrate with all three existing CL types to further enhance performance. Code is available at https://github.com/wenyudu/MIGU.

cs.LG↗

arXiv2024

DiscipLink: Unfolding Interdisciplinary Information Seeking Process via Human-AI Co-Exploration

Interdisciplinary studies often require researchers to explore literature in diverse branches of knowledge. Yet, navigating through the highly scattered knowledge from unfamiliar disciplines poses a significant challenge. In this paper, we introduce DiscipLink, a novel interactive system that facilitates collaboration between researchers and large language models (LLMs) in interdisciplinary information seeking (IIS). Based on users' topics of interest, DiscipLink initiates exploratory questions from the perspectives of possible relevant fields of study, and users can further tailor these questions. DiscipLink then supports users in searching and screening papers under selected questions by automatically expanding queries with disciplinary-specific terminologies, extracting themes from retrieved papers, and highlighting the connections between papers and questions. Our evaluation, comprising a within-subject comparative experiment and an open-ended exploratory study, reveals that DiscipLink can effectively support researchers in breaking down disciplinary boundaries and integrating scattered knowledge in diverse fields. The findings underscore the potential of LLM-powered tools in fostering information-seeking practices and bolstering interdisciplinary research.

cs.HC↗

arXiv2024

FRI-Net: Floorplan Reconstruction via Room-wise Implicit Representation

In this paper, we introduce a novel method called FRI-Net for 2D floorplan reconstruction from 3D point cloud. Existing methods typically rely on corner regression or box regression, which lack consideration for the global shapes of rooms. To address these issues, we propose a novel approach using a room-wise implicit representation with structural regularization to characterize the shapes of rooms in floorplans. By incorporating geometric priors of room layouts in floorplans into our training strategy, the generated room polygons are more geometrically regular. We have conducted experiments on two challenging datasets, Structured3D and SceneCAD. Our method demonstrates improved performance compared to state-of-the-art methods, validating the effectiveness of our proposed representation for floorplan reconstruction.

cs.CV↗

arXiv2024

Purification Of Contaminated Convolutional Neural Networks Via Robust Recovery: An Approach with Theoretical Guarantee in One-Hidden-Layer Case

Convolutional neural networks (CNNs), one of the key architectures of deep learning models, have achieved superior performance on many machine learning tasks such as image classification, video recognition, and power systems. Despite their success, CNNs can be easily contaminated by natural noises and artificially injected noises such as backdoor attacks. In this paper, we propose a robust recovery method to remove the noise from the potentially contaminated CNNs and provide an exact recovery guarantee on one-hidden-layer non-overlapping CNNs with the rectified linear unit (ReLU) activation function. Our theoretical results show that both CNNs' weights and biases can be exactly recovered under the overparameterization setting with some mild assumptions. The experimental results demonstrate the correctness of the proofs and the effectiveness of the method in both the synthetic environment and the practical neural network setting. Our results also indicate that the proposed method can be extended to multiple-layer CNNs and potentially serve as a defense strategy against backdoor attacks.

cs.LG↗

arXiv2024

PFME: A Modular Approach for Fine-grained Hallucination Detection and Editing of Large Language Models

Large Language Models (LLMs) excel in fluency but risk producing inaccurate content, called "hallucinations." This paper outlines a standardized process for categorizing fine-grained hallucination types and proposes an innovative framework--the Progressive Fine-grained Model Editor (PFME)--specifically designed to detect and correct fine-grained hallucinations in LLMs. PFME consists of two collaborative modules: the Real-time Fact Retrieval Module and the Fine-grained Hallucination Detection and Editing Module. The former identifies key entities in the document and retrieves the latest factual evidence from credible sources. The latter further segments the document into sentence-level text and, based on relevant evidence and previously edited context, identifies, locates, and edits each sentence's hallucination type. Experimental results on FavaBench and FActScore demonstrate that PFME outperforms existing methods in fine-grained hallucination detection tasks. Particularly, when using the Llama3-8B-Instruct model, PFME's performance in fine-grained hallucination detection with external knowledge assistance improves by 8.7 percentage points (pp) compared to ChatGPT. In editing tasks, PFME further enhances the FActScore of FActScore-Alpaca13B and FActScore-ChatGPT datasets, increasing by 16.2pp and 4.6pp, respectively.

cs.CL↗

arXiv2024

Spatial and Surface Correspondence Field for Interaction Transfer

In this paper, we introduce a new method for the task of interaction transfer. Given an example interaction between a source object and an agent, our method can automatically infer both surface and spatial relationships for the agent and target objects within the same category, yielding more accurate and valid transfers. Specifically, our method characterizes the example interaction using a combined spatial and surface representation. We correspond the agent points and object points related to the representation to the target object space using a learned spatial and surface correspondence field, which represents objects as deformed and rotated signed distance fields. With the corresponded points, an optimization is performed under the constraints of our spatial and surface interaction representation and additional regularization. Experiments conducted on human-chair and hand-mug interaction transfer tasks show that our approach can handle larger geometry and topology variations between source and target shapes, significantly outperforming state-of-the-art methods.

cs.CV↗

arXiv2024

Unlocking Emergent Modularity in Large Language Models

Modular Neural Networks (MNNs) demonstrate various advantages over monolithic models. Existing MNNs are generally $\textit{explicit}$: their modular architectures are pre-defined, with individual modules expected to implement distinct functions. Recent works reveal that there exists $\textit{implicit}$ modularity in standard pre-trained transformers, namely $\textit{Emergent Modularity}$. They indicate that such modular structures spontaneously exhibit during the early pre-training phase. Despite the benefits of modularity, most Language Models (LMs) are still treated as monolithic models in the pre-train and fine-tune paradigm, with their emergent modularity locked and underutilized. In this work, focusing on unlocking the emergent modularity in LMs, we showcase that standard LMs could be fine-tuned as their Mixture-of-Expert (MoEs) counterparts without introducing any extra parameters. Such MoEs are derived from emergent modularity and are referred to as Emergent MoEs (EMoE). Our experiments demonstrate that fine-tuning EMoE effectively improves downstream in-domain and out-of-domain generalization compared with vanilla fine-tuning. Our analysis and ablation studies further illustrate that it is robust to various configurations and can scale up to Large Language Models (i.e., Llama2-7B and Llama-30B). Code is available at https://github.com/qiuzh20/EMoE.

cs.LG↗

arXiv2024

Sharing Frissons among Online Video Viewers: Exploring the Design of Affective Communication for Aesthetic Chills

On online video platforms, viewers often lack a channel to sense others' and express their affective state on the fly compared to co-located group-viewing. This study explored the design of complementary affective communication specifically for effortless, spontaneous sharing of frissons during video watching. Also known as aesthetic chills, frissons are instant psycho-physiological reactions like goosebumps and shivers to arousing stimuli. We proposed an approach that unobtrusively detects viewers' frissons using skin electrodermal activity sensors and presents the aggregated data alongside online videos. Following a design process of brainstorming, focus group interview (N=7), and design iterations, we proposed three different designs to encode viewers' frisson experiences, namely, ambient light, icon, and vibration. A mixed-methods within-subject study (N=48) suggested that our approach offers a non-intrusive and efficient way to share viewers' frisson moments, increases the social presence of others as if watching together, and can create affective contagion among viewers.

cs.HC↗