arXiv subjects

Hui Xie

Publications and source records attributed to Hui Xie.

At least 19 recordsLinked to original sources

arXiv2026

PTQ4SNN: Membrane-Aware Post-Training Quantization for Spiking Neural Networks

Spiking neural networks (SNNs) enable sparse and event-driven computation, but their low-bit deployment remains incomplete because recurrent membrane states are commonly retained in floating point even after weight quantization. Quantizing these states is challenging because their distributions differ across channels and from the preceding weights, while small perturbations near the firing threshold may alter spike decisions and accumulate over time. We propose PTQ4SNN, a membrane-aware post-training quantization framework that jointly quantizes weights and recurrent membrane states using only a small calibration set. First, a channel-wise Unified Scale Bridge constrains the membrane scale as s_mem,c = s_w,c * 2^k_c, adapting to membrane distributions while enabling shift-compatible scale conversion. Second, Mixed-Precision Bit Allocation assigns 2/4/8-bit precision to membrane channels according to firing activity and quantization sensitivity under an average-bit budget. The framework operates on reusable projection-LIF pairs and supports both convolutional SNNs and spike-driven Transformers without backbone retraining. Experiments on static and event-based classification and semantic segmentation show that PTQ4SNN effectively preserves model accuracy under W4 quantization and approximately 4-bit membrane precision.

cs.AI↗

arXiv2026

SemPIC: Learning Semantic Position-Independent KV Caches

Long-context retrieval and agentic workloads repeatedly reuse the same documents under changing instructions, histories, and document orders. Prefix caching cannot exploit this reuse, while position-independent caching (PIC) remains unreliable because independently compiled KV states lack the future context in which they will be consumed. Our diagnostics show that a learned boundary-conditioned baseline sharply reduces attention deviation near reusable-block boundaries but leaves interior and task-level residuals, motivating adaptation of the document representation itself. We present \emph{SemPIC}, which trains a LoRA-enabled Writer to compile native per-layer document KVs through behavioral distillation while retaining the pretrained decoder as an unchanged Reader. Adaptation is confined to offline cache construction, preserving the standard KV interface and cache-hit decoding path. We further introduce KV Gradient Checkpointing, which reduces peak training memory without severing gradients through cached KVs. Across three models and four tasks, SemPIC raises mean micro-F1 over KV Packet from 0.53 to 0.60, approaching Full Recompute at 0.62. Code: https://github.com/jn12-29/SemPIC

cs.AI↗

arXiv2026

A pilot study examining transcranial photobiomodulation therapy intervention in college students with insomnia

College students commonly report insufficient sleep and poor sleep quality, with ~30% meeting insomnia criteria, posing significant threats to their physical growth, cognitive development, and overall well-being, as well as imposing a substantial economic burden on society [1]. The hyperarousal model of insomnia [2] emphasizes that hyperarousal across cognitive, emotional, and physiological domains mutually reinforces one another. Neuroimaging studies have further identified prefrontal hypoactivity as a key neural substrate underlying these dysfunctional cognitions and elevated arousal, reflecting a failure of top-down modulatory control over both limbic reactivity [3] and brainstem arousal nuclei [4]. Moreover, transcranial photobiomodulation (tPBM) therapy targeting the prefrontal cortex has demonstrated therapeutic efficacy across neuropsychiatric disorders with insomnia comorbidities [5,6], providing preliminary support for its application in insomnia. However, the neuro mechanisms underlying tPBM's therapeutic effects on insomnia remain to be elucidated.

q-bio.NC↗

arXiv2026

An Empirical Study of openPangu Quantization on Ascend NPUs

openPangu models are attractive targets for private and domestic large-language-model deployment, yet their robustness under aggressive post-training quantization on Ascend NPUs has not been systematically characterized. This paper conducts a controlled empirical study of openPangu 1B and 7B models on Huawei Ascend 910B1 NPUs. We evaluate representative weight-only and weight-activation post-training quantization methods, including RTN, GPTQ, AWQ, SmoothQuant, GPTAQ, BiLLM, and SliM-LLM, under a unified calibration and evaluation protocol. Across 18 evaluation tasks, we find that 8-bit weight-only quantization is effectively lossless for both models, while 4-bit quantization remains practical for the 7B model but is visibly more harmful for the 1B model on reasoning, math, and code tasks. Ultra-low precision remains challenging: most 2-bit and binary settings collapse to near-random behavior, and W4A4 SmoothQuant produces non-finite perplexity in our evaluation. These results provide an NPU-oriented accuracy map for selecting openPangu quantization settings and highlight the persistent difficulty of extreme low-bit compression.

cs.LG↗

arXiv2026

Selective Latent Thinking: Adaptive Compression of LLM Reasoning Chains

Explicit chain-of-thought (CoT) reasoning substantially improves the reasoning ability of large language models (LLMs), but incurs high inference cost due to lengthy autoregressive traces. Existing latent reasoning methods offer a promising alternative, yet they often treat reasoning as uniformly compressible, causing precision-critical intermediate steps to be overly compressed and thereby degrading reasoning accuracy. In this work, we propose Selective Latent Thinking (SLT), a framework that selectively compresses redundant reasoning spans into latent representations while preserving precision-critical spans as explicit CoT within the same reasoning trajectory. Specifically, SLT first uses a lightweight decoder to anticipate a short upcoming reasoning span, and then applies confidence-based gating to determine the longest span that can be reliably compressed. The accepted span is encoded into a compact latent representation to improve reasoning efficiency, while uncertain or precision-critical reasoning remains in explicit CoT form to preserve accuracy. To learn this selective compression policy, SLT adopts a three-stage training strategy that combines span-level latent compression, reliability-aware future reasoning prediction, and trajectory-level reinforcement learning to optimize the trade-off between answer correctness and reasoning cost. Extensive experiments across four mathematical reasoning benchmarks demonstrate that SLT achieves 22.7\% higher accuracy than latent reasoning baselines at comparable compression ratios, while reducing reasoning chain length by 58.4\% with only 2.8\% accuracy degradation compared to explicit CoT,Our code can be found in https://github.com/hunshi34/SLT.

cs.CL↗

arXiv2026

DiffKD-DCIS: Predicting Upgrade of Ductal Carcinoma In Situ with Diffusion Augmentation and Knowledge Distillation

Accurately predicting the upgrade of ductal carcinoma in situ (DCIS) to invasive ductal carcinoma (IDC) is crucial for surgical planning. However, traditional deep learning methods face challenges due to limited ultrasound data and poor generalization ability. This study proposes the DiffKD-DCIS framework, integrating conditional diffusion modeling with teacher-student knowledge distillation. The framework operates in three stages: First, a conditional diffusion model generates high-fidelity ultrasound images using multimodal conditions for data augmentation. Then, a deep teacher network extracts robust features from both original and synthetic data. Finally, a compact student network learns from the teacher via knowledge distillation, balancing generalization and computational efficiency. Evaluated on a multi-center dataset of 1,435 cases, the synthetic images were of good quality. The student network had fewer parameters and faster inference. On external test sets, it outperformed partial combinations, and its accuracy was comparable to senior radiologists and superior to junior ones, showing significant clinical potential.

cs.CV↗

arXiv2025

EEG Dynamic Microstate Patterns Induced by Pulsed Wave Transcranial Photobiomodulation Therapy

Transcranial photobiomodulation (tPBM) therapy is an emerging, non-invasive neuromodulation technique that has demonstrated considerable potential in the field of neuropsychiatric disorders. Several studies have found that pulsed wave (PW) tPBM therapy yields superior biomodulatory effects. However, its neural mechanisms are still unknown which poses a significant barrier to the development of an optimized protocol. A randomized, single-blind study including 29 participants was conducted using a crossover design, with sham and continuous wave (CW) groups as controls. The EEG microstate analysis was utilized to explore the relative variations in temporal parameters and brain functional connectivity. To further elucidate the dynamic activity patterns of microstates, a 10-repeat 10-fold cross-validation with nine machine learning algorithms and kernel Shapley additive explanations analysis was employed. Results indicated that the pulsed wave mode enhanced the global efficiency, local efficiency, and betweenness centrality of microstate C in brain functional networks as well as the mean durations parameter achieving a middle to large effect size, with superior effects compared to the sham and continuous wave groups. Furthermore, the support vector machine based on the radial basis function method with kernel Shapley additive explanations analysis demonstrated the best performance with an area under the curve (AUC) reaching 0.956, and found that the 8 of top-10 microstate features related to microstate C contributed most significantly to the PW mode. In conclusion, the EEG microstate analysis found that PW tPBM therapy modulates the microstate C-specific patterns in the human brain, suggesting that microstate dynamics may serve as a state-dependent biomarker for the optimization of tPBM protocol.

q-bio.NC↗

arXiv2025

Conditional Diffusion Model with Anatomical-Dose Dual Constraints for End-to-End Multi-Tumor Dose Prediction

Radiotherapy treatment planning often relies on time-consuming, trial-and-error adjustments that heavily depend on the expertise of specialists, while existing deep learning methods face limitations in generalization, prediction accuracy, and clinical applicability. To tackle these challenges, we propose ADDiff-Dose, an Anatomical-Dose Dual Constraints Conditional Diffusion Model for end-to-end multi-tumor dose prediction. The model employs LightweightVAE3D to compress high-dimensional CT data and integrates multimodal inputs, including target and organ-at-risk (OAR) masks and beam parameters, within a progressive noise addition and denoising framework. It incorporates conditional features via a multi-head attention mechanism and utilizes a composite loss function combining MSE, conditional terms, and KL divergence to ensure both dosimetric accuracy and compliance with clinical constraints. Evaluation on a large-scale public dataset (2,877 cases) and three external institutional cohorts (450 cases in total) demonstrates that ADDiff-Dose significantly outperforms traditional baselines, achieving an MAE of 0.101-0.154 (compared to 0.316 for UNet and 0.169 for GAN models), a DICE coefficient of 0.927 (a 6.8% improvement), and limiting spinal cord maximum dose error to within 0.1 Gy. The average plan generation time per case is reduced to 22 seconds. Ablation studies confirm that the structural encoder enhances compliance with clinical dose constraints by 28.5%. To our knowledge, this is the first study to introduce a conditional diffusion model framework for radiotherapy dose prediction, offering a generalizable and efficient solution for automated treatment planning across diverse tumor sites, with the potential to substantially reduce planning time and improve clinical workflow efficiency.

cs.CV↗

arXiv2025

SPEAR: Structured Pruning for Spiking Neural Networks via Synaptic Operation Estimation and Reinforcement Learning

While deep spiking neural networks (SNNs) demonstrate superior performance, their deployment on resource-constrained neuromorphic hardware still remains challenging. Network pruning offers a viable solution by reducing both parameters and synaptic operations (SynOps) to facilitate the edge deployment of SNNs, among which search-based pruning methods search for the SNNs structure after pruning. However, existing search-based methods fail to directly use SynOps as the constraint because it will dynamically change in the searching process, resulting in the final searched network violating the expected SynOps target. In this paper, we introduce a novel SNN pruning framework called SPEAR, which leverages reinforcement learning (RL) technique to directly use SynOps as the searching constraint. To avoid the violation of SynOps requirements, we first propose a SynOps prediction mechanism called LRE to accurately predict the final SynOps after search. Observing SynOps cannot be explicitly calculated and added to constrain the action in RL, we propose a novel reward called TAR to stabilize the searching. Extensive experiments show that our SPEAR framework can effectively compress SNN under specific SynOps constraint.

cs.NE↗

arXiv2024

Significant Impact of Quantum and Anharmonic Effects on the Structural Stability and Superconductivity of NbH3 at High Pressures

First-principles calculations combined with the stochastic self-consistent harmonic approximation reveal significant effects of the quantum ionic fluctuations and lattice anharmonicity on the dynamical stability of NbH3 under high pressures. Previous theoretical predictions, which ignored ionic fluctuations and relied on the harmonic approximation, suggested that the I43d phase is the most thermodynamically favorable structure between 33 and 400 GPa, with the Fm3m phase considered thermodynamically metastable. However, recent experiments at 187 GPa identified the Fm3m phase, conflicting with the prediction. In contrast, the present study indicates that the Fm3m phase remains dynamically stable down to at least 145 GPa, approximately 145 GPa lower than harmonic estimates, while the I43d phase is dynamically unstable at 187 GPa, consistent with the experimental findings. Furthermore, systematic calculations are performed on the structural, vibrational and superconducting properties of Fm3m NbH3 under pressures ranging from 100 to 300 GPa, revealing dramatic modifications due to the quantum and anharmonic effects. The calculated superconducting critical temperature (Tc) from the McMillan equation for Fm3m NbH3 at 187 GPa is 44 K, with mu set at 0.15, close to the measured value. These findings highlight the crucial role of quantum anharmonic effects in stabilizing the Fm3m phase.

cond-mat.mtrl-sci↗

arXiv2024

Feature compression is the root cause of adversarial fragility in neural network classifiers

In this paper, we uniquely study the adversarial robustness of deep neural networks (NN) for classification tasks against that of optimal classifiers. We look at the smallest magnitude of possible additive perturbations that can change a classifier's output. We provide a matrix-theoretic explanation of the adversarial fragility of deep neural networks for classification. In particular, our theoretical results show that a neural network's adversarial robustness can degrade as the input dimension $d$ increases. Analytically, we show that neural networks' adversarial robustness can be only $1/\sqrt{d}$ of the best possible adversarial robustness of optimal classifiers. Our theories match remarkably well with numerical experiments of practically trained NN, including NN for ImageNet images. The matrix-theoretic explanation is consistent with an earlier information-theoretic feature-compression-based explanation for the adversarial fragility of neural networks.

cs.LG↗

arXiv2024

Toward Efficient Deep Spiking Neuron Networks:A Survey On Compression

With the rapid development of deep learning, Deep Spiking Neural Networks (DSNNs) have emerged as promising due to their unique spike event processing and asynchronous computation. When deployed on neuromorphic chips, DSNNs offer significant power advantages over Deep Artificial Neural Networks (DANNs) and eliminate time and energy consuming multiplications due to the binary nature of spikes (0 or 1). Additionally, DSNNs excel in processing temporal information, making them potentially superior for handling temporal data compared to DANNs. However, their deep network structure and numerous parameters result in high computational costs and energy consumption, limiting real-life deployment. To enhance DSNNs efficiency, researchers have adapted methods from DANNs, such as pruning, quantization, and knowledge distillation, and developed specific techniques like reducing spike firing and pruning time steps. While previous surveys have covered DSNNs algorithms, hardware deployment, and general overviews, focused research on DSNNs compression and efficiency has been lacking. This survey addresses this gap by concentrating on efficient DSNNs and their compression methods. It begins with an exploration of DSNNs' biological background and computational units, highlighting differences from DANNs. It then delves into various compression methods, including pruning, quantization, knowledge distillation, and reducing spike firing, and concludes with suggestions for future research directions.

cs.NE↗

arXiv2023

Distance Guided Generative Adversarial Network for Explainable Binary Classifications

Despite the potential benefits of data augmentation for mitigating the data insufficiency, traditional augmentation methods primarily rely on the prior intra-domain knowledge. On the other hand, advanced generative adversarial networks (GANs) generate inter-domain samples with limited variety. These previous methods make limited contributions to describing the decision boundaries for binary classification. In this paper, we propose a distance guided GAN (DisGAN) which controls the variation degrees of generated samples in the hyperplane space. Specifically, we instantiate the idea of DisGAN by combining two ways. The first way is vertical distance GAN (VerDisGAN) where the inter-domain generation is conditioned on the vertical distances. The second way is horizontal distance GAN (HorDisGAN) where the intra-domain generation is conditioned on the horizontal distances. Furthermore, VerDisGAN can produce the class-specific regions by mapping the source images to the hyperplane. Experimental results show that DisGAN consistently outperforms the GAN-based augmentation methods with explainable binary classification. The proposed method can apply to different classification architectures and has potential to extend to multi-class classification.

cs.CV↗

arXiv2023

gcDLSeg: Integrating Graph-cut into Deep Learning for Binary Semantic Segmentation

Binary semantic segmentation in computer vision is a fundamental problem. As a model-based segmentation method, the graph-cut approach was one of the most successful binary segmentation methods thanks to its global optimality guarantee of the solutions and its practical polynomial-time complexity. Recently, many deep learning (DL) based methods have been developed for this task and yielded remarkable performance, resulting in a paradigm shift in this field. To combine the strengths of both approaches, we propose in this study to integrate the graph-cut approach into a deep learning network for end-to-end learning. Unfortunately, backward propagation through the graph-cut module in the DL network is challenging due to the combinatorial nature of the graph-cut algorithm. To tackle this challenge, we propose a novel residual graph-cut loss and a quasi-residual connection, enabling the backward propagation of the gradients of the residual graph-cut loss for effective feature learning guided by the graph-cut segmentation model. In the inference phase, globally optimal segmentation is achieved with respect to the graph-cut energy defined on the optimized image features learned from DL networks. Experiments on the public AZH chronic wound data set and the pancreas cancer data set from the medical segmentation decathlon (MSD) demonstrated promising segmentation accuracy, and improved robustness against adversarial attacks.

cs.CV↗

arXiv2023

Agent-based models of social behaviour and communication in evacuations: A systematic review

Most modern agent-based evacuation models involve interactions between evacuees. However, the assumed reasons for interactions and portrayal of them may be overly simple. Research from social psychology suggests that people interact and communicate with one another when evacuating and evacuee response is impacted by the way information is communicated. Thus, we conducted a systematic review of agent-based evacuation models to identify 1) how social interactions and communication approaches between agents are simulated, and 2) what key variables related to evacuation are addressed in these models. We searched Web of Science and ScienceDirect to identify articles that simulated information exchange between agents during evacuations, and social behaviour during evacuations. From the final 70 included articles, we categorised eight types of social interaction that increased in social complexity from collision avoidance to social influence based on strength of social connections with other agents. In the 17 models which simulated communication, we categorised four ways that agents communicate information: spatially through information trails or radii around agents, via social networks and via external communication. Finally, the variables either manipulated or measured in the models were categorised into the following groups: environmental condition, personal attributes of the agents, procedure, and source of information. We discuss promising directions for agent-based evacuation models to capture the effects of communication and group dynamics on evacuee behaviour. Moreover, we demonstrate how communication and group dynamics may impact the variables commonly used in agent-based evacuation models.

cs.MA↗

arXiv2023

Development of 15kA/cm$^2$ Fabrication Process for Superconducting Integrated Digital Circuits

A new fabrication process for superconducting integrated digital circuits is reported. We have developed the "SIMIT Nb04" fabrication technique for superconducting integrated circuits with Nb-based Josephson junctions based on the validated "SIMIT Nb03" process and Chemical Mechanical Planarization (CMP) technology. Seven Nb superconducting layers and one Mo resistor layer are included in the "SIMIT Nb04" process with 19 mask levels. The device structure is composed of active layers including junctions at the bottom, two passive transmission line (PTL) layers in the middle and a DC power layer at the top. The circuit fabrication started with the fabrication of Mo resistors with a target sheet resistance Rsh of 3 $Ω$, followed by the deposition of Nb/Al-AlO$_x$/Nb trilayer Josephson-junction with a target critical current density Jc at 15 kA/cm$^2$. To increase the Al-AlO$_x$ barrier layer etching's repeatability, an additional barrier protection layer was applied. To accomplish high-quality planarization, we created a planarization procedure coupled with dummy filling. To assess the process dependability and controllability, a set of process control monitors (PCMs) for monitoring fabrication and design parameters was designed and monitored. The successful manufacturing and testing of a few small-scale circuits, like our standard library cells, further attests to the viability of our fabrication process for superconducting integrated circuits.

physics.app-ph↗

arXiv2022

A deep learning network with differentiable dynamic programming for retina OCT surface segmentation

Multiple-surface segmentation in Optical Coherence Tomography (OCT) images is a challenge problem, further complicated by the frequent presence of weak image boundaries. Recently, many deep learning (DL) based methods have been developed for this task and yield remarkable performance. Unfortunately, due to the scarcity of training data in medical imaging, it is challenging for DL networks to learn the global structure of the target surfaces, including surface smoothness. To bridge this gap, this study proposes to seamlessly unify a U-Net for feature learning with a constrained differentiable dynamic programming module to achieve an end-to-end learning for retina OCT surface segmentation to explicitly enforce surface smoothness. It effectively utilizes the feedback from the downstream model optimization module to guide feature learning, yielding a better enforcement of global structures of the target surfaces. Experiments on Duke AMD (age-related macular degeneration) and JHU MS (multiple sclerosis) OCT datasets for retinal layer segmentation demonstrated very promising segmentation accuracy.

eess.IV↗

arXiv2022

Discontinuity Computing using Physics-Informed Neural Network

Simulating discontinuities is a long standing problem especially for shock waves with strong nonlinear feather. Despite being a promising method, the recently developed physics-informed neural network (PINN) is still weak for calculating discontinuities compared with traditional shock-capturing methods. In this paper, we intend to improve the shock-capturing ability of the PINN. The primary strategy of this work is to weaken the expression of the network near discontinuities by adding a gradient-weight into the governing equations locally at each residual point. This strategy allows the network to focus on training smooth parts of the solutions. Then, automatically affected by the compressible property near shock waves, a sharp discontinuity appears with wrong inside shock transition-points compressed into well-trained smooth regions as passive particles. We study the solutions of one-dimensional Burgers equation and one- and two-dimensional Euler equations. Compared with the traditional high-order WENO-Z method in numerical examples, the proposed method can substantially improve discontinuity computing.

physics.flu-dyn↗