arXiv subjects

Dan Chen

Publications and source records attributed to Dan Chen.

At least 19 recordsLinked to original sources

arXiv2026

Statistical-Uncertainty-Driven Selection of Evaluation Frequency for Time-Dependent Sensing Calibration: A Demonstration with KAGRA Data

Accurate calibration of the gravitational-wave strain h(t) is essential for both detection and astrophysical inference. In operating detectors, slow temporal variations in the sensing response are tracked using calibration lines, but practical constraints can prevent those lines from being injected at frequencies that are favorable for precise estimation of sensing-side parameters. We present a statistical framework for preselecting evaluation frequencies under such constraints. We apply this framework to KAGRA data from the first part of the fourth LIGO-Virgo-KAGRA Observing Run, for which the nominal cavity-pole frequency was about 18 Hz, while the sensing-side calibration line used in practice was injected at 32.7 Hz. For each candidate evaluation frequency, we construct the sensing function, quantify its segment-wise statistical uncertainty from empirical percentiles of the sample distribution, and rank the candidates using a score that combines the interval widths of the amplitude and phase. When a 1% amplitude interval width and a 1 degree phase interval width are weighted equally, 244 Hz is selected in all 4096 s analysis segments throughout the analyzed period. Relative to the reference frequency of 32.7 Hz, the amplitude interval width is reduced to about one quarter over a broad frequency range, while the phase interval width remains broadly comparable. We also assess the discrepancy introduced by frequency translation separately. These results suggest that the proposed method provides a useful statistical preselection framework for evaluation frequencies under practical operational constraints.

gr-qc↗

arXiv2026

Statistical Estimation and Correction of Model-Measurement Bias in Time-Dependent Correction Factors of KAGRA

Calibration of gravitational-wave detectors reconstructs the strain h(t) from the detector output, and bias and uncertainty in this reconstruction directly affect downstream analyses. In ground-based interferometers, time-dependent correction factors (TDCFs) are estimated from calibration lines to track temporal variations of the detector response, while the underlying model parameters are periodically updated using broadband swept-sine calibration measurements (SSCMs). However, if a model-measurement bias exists between the measured transfer function and the reference model, the TDCFs inferred from calibration lines can introduce a systematic deviation into the reconstructed strain. We propose a statistical framework to estimate and correct this bias using repeated measurement-to-model ratios at the calibration-line frequencies. The bias correction factors are estimated with a rolling random-effects model based on restricted maximum likelihood (REML) and incorporated into the TDCF estimation, with their uncertainty propagated to the reconstructed response. Applying the method to KAGRA O4c data, we find that the uncorrected response shows deviations of up to approximately 7% in magnitude and 5 degrees in phase relative to the SSCM-based reference in representative examples. The correction reduces these deviations, with a modest increase in the propagated uncertainty due to the included correction-factor uncertainty. This framework provides a practical way to combine broadband reference models with calibration-line-based tracking when model-measurement bias is present.

astro-ph.IM↗

arXiv2026

Spiral metasurface enables tunable directional edge enhancement

Tunable directional edge enhancement facilitates the acquisition of distinct morphological features from objects, a capability that plays a vital role in enhancing the reliability and safety of autonomous driving systems. However, building a simple, miniature, and switchable directional edge enhancement system remains an urgent challenge. To address this, we propose a compact spiral metasurface capable of achieving tunable vertical and horizontal edge enhancement imaging without requiring a conventional 4f system, relying instead on a single integrated device. This functionality is realized by engineering the metasurface's spiral phase profile and its polarization-dependent response, where the edge enhancement direction is controlled by varying the incident beam's polarization, enabling switching between vertical and horizontal enhancement modes. Simulations demonstrate the metasurface's capability for switchable edge detection of lane markings and barrier contours. Broadband operation is also confirmed through simulation results. Owing to its compactness, switchable functionality, and broadband performance, the proposed spiral metasurface shows significant potential for applications in optical analog computing and autonomous driving.

physics.optics↗

arXiv2026

To Use or Not to Use: Investigating Student Perceptions of Faculty Generative AI Usage in Higher Education

While Generative AI (GenAI) rapidly integrated into higher education, existing research has primarily focused on regulating student use. As a result, student perspectives on faculty adoption of GenAI remained unexplored. In this study, we analyzed survey responses from 156 undergraduate and graduate students to examine their attitudes toward both student and faculty use of GenAI. We classified students into four groups based on their attitudes, including GenAI Optimists, Student Support Group, Faculty Support Group, and Non-supporters. Findings show that 37% of participants do not support GenAI use by either students or faculty, while 31% support GenAI use in both contexts. We also conducted thematic analysis to understand participants' concerns on faculty GenAI usage. Results revealed that (1) a majority of students (79%) questioned the validity and reliability of GenAI-generated responses, and (2) 37% of students feared that faculty overreliance on GenAI created a "futile cycle" that might reduce faculty critical thinking. Our findings showed that students expressed concerns about GenAI use by faculty in teaching and grading contexts, with pedagogical concerns being most prominent. These findings informed the future use of GenAI in teaching and learning in higher education.

cs.CY↗

arXiv2026

Accelerating Multi-Scale Deformable Attention Using Near-Memory-Processing Architecture

Multi Scale Deformable Attention (MSDAttn) has become a fundamental component in various vision tasks due to its effective multi scale grid sampling (MSGS). However, its reliance on random sampling results in highly irregular memory access patterns, making it a memory intensive operation inefficient for GPUs. Near memory processing (NMP) offers a promising solution for accelerating memory bound kernels, yet existing NMP based attention accelerators remain suboptimal for MSDAttn due to incompatible load balancing and data reuse strategies. Specifically, current NMP solutions uniformly distribute processing elements (PEs) across all banks, leading to significant PE underutilization and excessive cross bank data transfers. Moreover, most rely on locality based reuse, which fails under MSDAttn's unpredictable sampling patterns. To address these challenges, this paper presents DANMP, a hardware software co designed NMP based MSDAttn accelerator. On the hardware side, DANMP adopts non uniform NMP integration to handle unbalanced workloads, allocating PEs only in select banks for hot entries, while cold data are processed at the bank group level reducing PE idleness and cross bank transfers. On the software side, it introduces a clustering and packing (CAP) method that leverages clustering to improve temporal locality in query processing, enhancing data reuse. Finally, we implement host NMP co optimization techniques, including an optimized programming model, customized instructions, and a tailored dataflow. Experiments on object detection inference show that DANMP achieves 97.43x speedup and 208.47x energy efficiency improvement over NVIDIA A6000 GPU.

cs.AR↗

arXiv2026

When Scaling Fails: Mitigating Audio Perception Decay of LALMs via Multi-Step Perception-Aware Reasoning

Test-Time Scaling has shown notable efficacy in addressing complex problems through scaling inference compute. However, within Large Audio-Language Models (LALMs), an unintuitive phenomenon exists: post-training models for structured reasoning trajectories results in marginal or even negative gains compared to post-training for direct answering. To investigate it, we introduce CAFE, an evaluation framework designed to precisely quantify audio reasoning errors. Evaluation results reveal LALMs struggle with perception during reasoning and encounter a critical bottleneck: reasoning performance suffers from audio perception decay as reasoning length extends. To address it, we propose MPAR$^2$, a paradigm that encourages dynamic perceptual reasoning and decomposes complex questions into perception-rich sub-problems. Leveraging reinforcement learning, MPAR$^2$ improves perception performance on CAFE from 31.74% to 63.51% and effectively mitigates perception decay, concurrently enhancing reasoning capabilities to achieve a significant 74.59% accuracy on the MMAU benchmark. Further analysis demonstrates that MPAR$^2$ reinforces LALMs to attend to audio input and dynamically adapts reasoning budget to match task complexity.

cs.SD↗

arXiv2025

RESPOND: Risk-Enhanced Structured Pattern for LLM-driven Online Node-level Decision-making

Current LLM-based driving agents that rely on unstructured plain-text memory suffer from low-precision scene retrieval and inefficient reflection. To address this limitation, we present RESPOND, a structured decision-making framework for LLM-driven agents grounded in explicit risk patterns. RESPOND represents each ego-centric scene using a unified 5 by 3 matrix that encodes spatial topology and road constraints, enabling consistent and reliable retrieval of spatial risk configurations. Based on this representation, a hybrid rule and LLM decision pipeline is developed with a two-tier memory mechanism. In high-risk contexts, exact pattern matching enables rapid and safe reuse of verified actions, while in low-risk contexts, sub-pattern matching supports personalized driving style adaptation. In addition, a pattern-aware reflection mechanism abstracts tactical corrections from crash and near-miss frames to update structured memory, achieving one-crash-to-generalize learning. Extensive experiments demonstrate the effectiveness of RESPOND. In highway-env, RESPOND outperforms state-of-the-art LLM-based and reinforcement learning based driving agents while producing substantially fewer collisions. With step-wise human feedback, the agent acquires a Sporty driving style within approximately 20 decision steps through sub-pattern abstraction. For real-world validation, RESPOND is evaluated on 53 high-risk cut-in scenarios extracted from the HighD dataset. For each event, intervention is applied immediately before the cut-in and RESPOND re-decides the driving action. Compared to recorded human behavior, RESPOND reduces subsequent risk in 84.9 percent of scenarios, demonstrating its practical feasibility under real-world driving conditions. These results highlight RESPONDs potential for autonomous driving, personalized driving assistance, and proactive hazard mitigation.

cs.HC↗

arXiv2025

Regression of Suspension Violin Modes in KAGRA O3GK Data with Kalman Filters

Suspension thermal modes in interferometric gravitational-wave detectors produce narrow, high-Q spectral lines that can contaminate gravitational searches and bias parameter estimation. In KAGRA, cryogenic mirrors are held by thick suspension fibers, designed to sustain such a low-temperature environment, which may further affect inharmonicity modes, fiber dimensions, and mechanical behavior compared to typical interferometers. As these modes remain a prominent source of narrowband contamination, we implement a Kalman filter to model and track violin lines, building on the methodology introduced in [1], and apply subtraction to KAGRA O3GK data. Using gravitational-wave template injections, we validate that the subtraction preserves matched-filter SNR while effectively suppressing line power. Comparisons of power spectral densities and residual analyses confirm that the method removes deterministic line contributions without introducing waveform distortions. This approach provides a cleaner strain channel for searches and parameter estimation and will become increasingly important for future low-temperature detectors with higher-Q suspensions, such as the Einstein Telescope.

gr-qc↗

arXiv2025

Preserving spreading dynamics and information flow in complex network reduction

Effectively preserving both the structural and dynamical properties during the reduction of complex networks remains a significant research topic. Existing network reduction methods based on renormalization group or sampling often face challenges such as high computational complexity and the loss of critical dynamic attributes. This paper proposes an efficient network reduction framework based on subgraph extraction, which accurately preserves epidemic spreading dynamics and information flow through a coordinated optimization strategy of node removal and edge pruning. Specifically, a node removal algorithm driven by enhanced degree centrality is introduced to preferentially remove low-centrality nodes, thereby constructing a smaller-scale subnetwork. Subsequently, an edge pruning algorithm is designed to regulate the edge density of the subnetwork, ensuring that its average degree remains approximately consistent with that of the original network. Experimental results on Erd\"os-R\'enyi random graphs, Barab\'asi-Albert scale-free networks, and real-world social contact networks from various domains demonstrate that this proposed method can reduce the size of networks with heterogeneous structures by more than 85\%, while preserving their epidemic dynamics and information flow. More importantly, our method almost always achieves the highest accuracy compared to state-of-the-art techniques. These findings provide valuable insights for predicting the dynamical behavior of large-scale real-world networks, and also reveal that a large number of nodes and edges in real-world networks play redundant roles in information transmission.

cs.SI↗

arXiv2025

Photon Calibration Performance of KAGRA during the 4th Joint Observing Run (O4)

KAGRA is a kilometer-scale cryogenic gravitational-wave (GW) detector in Japan. It joined the 4th joint observing run (O4) in May 2023 in collaboration with the Laser Interferometer GW Observatory (LIGO) in the USA, and Virgo in Italy. After one month of observations, KAGRA entered a break period to enhance its sensitivity to GWs, and it is planned to rejoin O4 before its scheduled end in October 2025. To accurately recover the information encoded in the GW signals, it is essential to properly calibrate the observed signals. We employ a photon calibration (Pcal) system as a reference signal injector to calibrate the output signals obtained from the telescope. In ideal future conditions, the uncertainty in Pcal could dominate the uncertainty in the observed data. In this paper, we present the methods used to estimate the uncertainty in the Pcal systems employed during KAGRA O4 and report an estimated system uncertainty of 0.79%, which is three times lower than the uncertainty achieved in the previous 3rd joint observing run (O3) in 2020. Additionally, we investigate the uncertainty in the Pcal laser power sensors, which had the highest impact on the Pcal uncertainty, and estimate the beam positions on the KAGRA main mirror, which had the second highest impact. The Pcal systems in KAGRA are the first fully functional calibration systems for a cryogenic GW telescope. To avoid interference with the KAGRA cryogenic systems, the Pcal systems incorporate unique features regarding their placement and the use of telephoto cameras, which can capture images of the mirror surface at almost normal incidence. As future GW telescopes, such as the Einstein Telescope, are expected to adopt cryogenic techniques, the performance of the KAGRA Pcal systems can serve as a valuable reference.

astro-ph.IM↗

arXiv2025

Hybrid Photonic-digital Accelerator for Attention Mechanism

The wide adoption and substantial computational resource requirements of attention-based Transformers have spurred the demand for efficient hardware accelerators. Unlike digital-based accelerators, there is growing interest in exploring photonics due to its high energy efficiency and ultra-fast processing speeds. However, the significant signal conversion overhead limits the performance of photonic-based accelerators. In this work, we propose HyAtten, a photonic-based attention accelerator with minimize signal conversion overhead. HyAtten incorporates a signal comparator to classify signals into two categories based on whether they can be processed by low-resolution converters. HyAtten integrates low-resolution converters to process all low-resolution signals, thereby boosting the parallelism of photonic computing. For signals requiring high-resolution conversion, HyAtten uses digital circuits instead of signal converters to reduce area and latency overhead. Compared to state-of-the-art photonic-based Transformer accelerator, HyAtten achieves 9.8X performance/area and 2.2X energy-efficiency/area improvement.

cs.AR↗

arXiv2024

MiLoRA: Efficient Mixture of Low-Rank Adaptation for Large Language Models Fine-tuning

Low-rank adaptation (LoRA) and its mixture-of-experts (MOE) variants are highly effective parameter-efficient fine-tuning (PEFT) methods. However, they introduce significant latency in multi-tenant settings due to the LoRA modules and MOE routers added to multiple linear modules in the Transformer layer. To address this issue, we propose Mixture of Low-Rank Adaptation (MiLoRA), a novel and efficient LoRA variant. MiLoRA differs from previous MOE-style LoRA methods by considering each LoRA module as an expert and employing a prompt-aware routing mechanism. This mechanism calculates expert routing results once before generating the first new token and reuses these results for subsequent tokens, reducing latency. Extensive experiments and analysis on commonsense reasoning tasks, math reasoning tasks, and widely used LLM evaluation benchmarks demonstrate that MiLoRA consistently outperforms strong PEFT baselines with comparable tunable parameter budgets. Additionally, MiLoRA significantly reduces latency in multi-tenant settings compared to previous LoRA-based methods.

cs.CL↗

arXiv2022

A Knowledge-based Learning Framework for Self-supervised Pre-training Towards Enhanced Recognition of Biomedical Microscopy Images

Self-supervised pre-training has become the priory choice to establish reliable neural networks for automated recognition of massive biomedical microscopy images, which are routinely annotation-free, without semantics, and without guarantee of quality. Note that this paradigm is still at its infancy and limited by closely related open issues: 1) how to learn robust representations in an unsupervised manner from unlabelled biomedical microscopy images of low diversity in samples? and 2) how to obtain the most significant representations demanded by a high-quality segmentation? Aiming at these issues, this study proposes a knowledge-based learning framework (TOWER) towards enhanced recognition of biomedical microscopy images, which works in three phases by synergizing contrastive learning and generative learning methods: 1) Sample Space Diversification: Reconstructive proxy tasks have been enabled to embed a priori knowledge with context highlighted to diversify the expanded sample space; 2) Enhanced Representation Learning: Informative noise-contrastive estimation loss regularizes the encoder to enhance representation learning of annotation-free images; 3) Correlated Optimization: Optimization operations in pre-training the encoder and the decoder have been correlated via image restoration from proxy tasks, targeting the need for semantic segmentation. Experiments have been conducted on public datasets of biomedical microscopy images against the state-of-the-art counterparts (e.g., SimCLR and BYOL), and results demonstrate that: TOWER statistically excels in all self-supervised methods, achieving a Dice improvement of 1.38 percentage points over SimCLR. TOWER also has potential in multi-modality medical image analysis and enables label-efficient semi-supervised learning, e.g., reducing the annotation cost by up to 99% in pathological classification.

cs.CV↗

arXiv2022

CPSAA: Accelerating Sparse Attention using Crossbar-based Processing-In-Memory Architecture

The attention mechanism requires huge computational efforts to process unnecessary calculations, significantly limiting the system's performance. Researchers propose sparse attention to convert some DDMM operations to SDDMM and SpMM operations. However, current sparse attention solutions introduce massive off-chip random memory access. We propose CPSAA, a novel crossbar-based PIM-featured sparse attention accelerator. First, we present a novel attention calculation mode. Second, we design a novel PIM-based sparsity pruning architecture. Finally, we present novel crossbar-based methods. Experimental results show that CPSAA has an average of 89.6X, 32.2X, 17.8X, 3.39X, and 3.84X performance improvement and 755.6X, 55.3X, 21.3X, 5.7X, and 4.9X energy-saving when compare with GPU, FPGA, SANGER, ReBERT, and ReTransformer.

cs.AR↗

arXiv2022

From WSI-level to Patch-level: Structure Prior Guided Binuclear Cell Fine-grained Detection

Accurately and quickly binuclear cell (BC) detection plays a significant role in predicting the risk of leukemia and other malignant tumors. However, manual microscopy counting is time-consuming and lacks objectivity. Moreover, with the limitation of staining quality and diversity of morphology features in BC microscopy whole slide images (WSIs), traditional image processing approaches are helpless. To overcome this challenge, we propose a two-stage detection method inspired by the structure prior of BC based on deep learning, which cascades to implement BCs coarse detection at the WSI-level and fine-grained classification in patch-level. The coarse detection network is a multi-task detection framework based on circular bounding boxes for cells detection, and central key points for nucleus detection. The circle representation reduces the degrees of freedom, mitigates the effect of surrounding impurities compared to usual rectangular boxes and can be rotation invariant in WSI. Detecting key points in the nucleus can assist network perception and be used for unsupervised color layer segmentation in later fine-grained classification. The fine classification network consists of a background region suppression module based on color layer mask supervision and a key region selection module based on a transformer due to its global modeling capability. Additionally, an unsupervised and unpaired cytoplasm generator network is firstly proposed to expand the long-tailed distribution dataset. Finally, experiments are performed on BC multicenter datasets. The proposed BC fine detection method outperforms other benchmarks in almost all the evaluation criteria, providing clarification and support for tasks such as cancer screenings.

cs.CV↗

arXiv2022

On the number of tilting modules over a class of Auslander algebras

Let $\Lambda$ be a radical square zero algebra of a Dynkin quiver and let $\Gamma$ be the Auslander algebra of $\Lambda$. Then the number of tilting right $\Gamma$-modules is $2^{m-1}$ if $\Lambda$ is of $A_{m}$ type for $m\geq 1$. Otherwise, the number of tilting right $\Gamma$-modules is $2^{m-3}\times14$ if $\Lambda$ is either of $D_{m}$ type for $m\geq 4$ or of $E_{m}$ type for $m=6,7,8$.

math.RT↗

arXiv2021

Scaling properties of scale-free networks in degree-thresholding renormalization flows

We study the statistical properties of observables of scale-free networks in the degree-thresholding renormalization (DTR) flows. For BA scale-free networks with different sizes, we find that their structural and dynamical observables have similar scaling behavior in the DTR flow. The finite-size scaling analysis confirms this view and reveals a scaling function with a single scaling exponent that collectively captures the changes of these observables. Furthermore, for the scale-free network with a single initial size, we use its DTR snapshots as the original networks in the DTR flows, then perform a similar finite-size scaling analysis. Interestingly, the initial network and its snapshots share the same scaling exponent as the BA synthetic network. Our findings have important guiding significance for analyzing the structure and dynamic behavior of large-scale networks. Such as, in large-scale simulation scenarios with high time complexity, the DTR snapshot could serve as a substitute or guide for the initial network and then quickly explore the scaling behavior of initial networks.

physics.comp-ph↗

arXiv2020

Cryogenic suspension design for a kilometer-scale gravitational-wave detector

We report the mirror suspension design for Large-scale Cryogenic Gravitational wave Telescope, KAGRA, during bKAGRA Phase 1. Mirror thermal noise is one of the fundamental noises for room-temperature gravitational-wave detectors such as Advanced LIGO and Advanced Virgo. Thus, reduction of thermal noise is required for further improvement of their sensitivity. One effective approach for reducing thermal noise is to cool the mirrors. There are many technical challenges that must be overcome to cool the mirrors, such as cryocooler induced vibrations, thermal drift in suspensions, and reduction in duty cycling due to the increased number of potential failure mechanisms. Our mirror suspension has a black coating that makes radiative cooling more efficient. For conduction cooling, we developed ultra high purity aluminum heat links, which yield high thermal conductivity while keeping the spring constant sufficiently small. A unique inclination adjustment system, called moving mass, is used for aligning the mirror orientation in pitch. Photo-reflective displacement sensors, which have a large range, are installed for damping control on marionette recoil mass and intermediate recoil mass. Samarium cobalt magnets are used for coil-magnet actuators to prevent significant change of magnetism between room temperature and cryogenic temperature. In this paper, the design of our first cryogenic payload and its performance during bKAGRA Phase 1 are discussed.

physics.ins-det↗