arXiv subjects

Kai Wang

Publications and source records attributed to Kai Wang.

At least 325 records · Page 18Linked to original sources

arXiv2025

Potential-Driven Metal Cycling: JADES Census of Gas-Phase Metallicity for galaxies at 1 < z < 7

The gravitational potential is established as a critical determinant of gas-phase metallicity (12+log(O/H)) in low-redshift galaxies, whereas its influence remains unconfirmed at high redshifts. We investigate the correlation between gas-phase metallicity and effective radius ($R_{\rm e}$) for a sample of galaxies with redshifts ranging from 1 to 7, drawn from JADES (JWST Advanced Deep Extragalactic Survey) Data Release 3. We calculate the metallicities using four strong-line methods: ${\rm N2S2Hα}$, ${\rm R23}$, ${\rm N2}$, and ${\rm O3N2}$, respectively. After taking out the evolution of size, we find that the offsets of mass-size relation ($Δ\log R_{\rm e}$) are significantly negatively correlated with the offset of mass-metallicity relation ($Δ\log({\rm O/H})$) for the four metallicity tracers. Regardless of the metallicity tracer used, we obtain Spearman rank $p-$values much less than 0.01, rejecting the null hypothesis that the observed correlation is statistically nonsignificant and attributable to random chance. This is also true for galaxies with $z>3$, with $p-$values less than 0.05 for the four metallicity tracers. We for the first time find evidence of size playing a key role in determining gas-phase metallicity towards cosmic dawn, suggesting that the gravitational potential influences their material-exchange processes with the surrounding environment at very early universe.

astro-ph.GA↗

arXiv2025

A symmetric primal-dual algorithmic framework for saddle point problems

In this paper, we propose a new primal-dual algorithmic framework for a class of convex-concave saddle point problems frequently arising from image processing and machine learning. Our algorithmic framework updates the primal variable between the twice calculations of the dual variable, thereby appearing a symmetric iterative scheme, which is accordingly called the symmetric primal-dual algorithm (SPIDA). It is noteworthy that the subproblems of our SPIDA are equipped with Bregman proximal regularization terms, which make SPIDA versatile in the sense that it enjoys an algorithmic framework to understand the iterative schemes of some existing algorithms, such as the classical augmented Lagrangian method (ALM), linearized ALM, and Jacobian splitting algorithms for linearly constrained optimization problems. Besides, our algorithmic framework allows us to derive some customized versions so that SPIDA works as efficiently as possible for structured optimization problems. Theoretically, under some mild conditions, we prove the global convergence of SPIDA and estimate the linear convergence rate under a generalized error bound condition defined by Bregman distance. Finally, a series of numerical experiments on the basis pursuit, robust principal component analysis, and image restoration demonstrate that our SPIDA works well on synthetic and real-world datasets.

math.OC↗

arXiv2025

One-Way Ticket:Time-Independent Unified Encoder for Distilling Text-to-Image Diffusion Models

Text-to-Image (T2I) diffusion models have made remarkable advancements in generative modeling; however, they face a trade-off between inference speed and image quality, posing challenges for efficient deployment. Existing distilled T2I models can generate high-fidelity images with fewer sampling steps, but often struggle with diversity and quality, especially in one-step models. From our analysis, we observe redundant computations in the UNet encoders. Our findings suggest that, for T2I diffusion models, decoders are more adept at capturing richer and more explicit semantic information, while encoders can be effectively shared across decoders from diverse time steps. Based on these observations, we introduce the first Time-independent Unified Encoder TiUE for the student model UNet architecture, which is a loop-free image generation approach for distilling T2I diffusion models. Using a one-pass scheme, TiUE shares encoder features across multiple decoder time steps, enabling parallel sampling and significantly reducing inference time complexity. In addition, we incorporate a KL divergence term to regularize noise prediction, which enhances the perceptual realism and diversity of the generated images. Experimental results demonstrate that TiUE outperforms state-of-the-art methods, including LCM, SD-Turbo, and SwiftBrushv2, producing more diverse and realistic results while maintaining the computational efficiency.

cs.CV↗

arXiv2025

Delayed-KD: Delayed Knowledge Distillation based CTC for Low-Latency Streaming ASR

CTC-based streaming ASR has gained significant attention in real-world applications but faces two main challenges: accuracy degradation in small chunks and token emission latency. To mitigate these challenges, we propose Delayed-KD, which applies delayed knowledge distillation on CTC posterior probabilities from a non-streaming to a streaming model. Specifically, with a tiny chunk size, we introduce a Temporal Alignment Buffer (TAB) that defines a relative delay range compared to the non-streaming teacher model to align CTC outputs and mitigate non-blank token mismatches. Additionally, TAB enables fine-grained control over token emission delay. Experiments on 178-hour AISHELL-1 and 10,000-hour WenetSpeech Mandarin datasets show consistent superiority of Delayed-KD. Impressively, Delayed-KD at 40 ms latency achieves a lower character error rate (CER) of 5.42% on AISHELL-1, comparable to the competitive U2++ model running at 320 ms latency.

cs.SD↗

arXiv2025

Solving Euler equations with Multiple Discontinuities via Separation-Transfer Physics-Informed Neural Networks

Despite the remarkable progress of physics-informed neural networks (PINNs) in scientific computing, they continue to face challenges when solving hydrodynamic problems with multiple discontinuities. In this work, we propose Separation-Transfer Physics Informed Neural Networks (ST-PINNs) to address such problems. By sequentially resolving discontinuities from strong to weak and leveraging transfer learning during training, ST-PINNs significantly reduce the problem complexity and enhance solution accuracy. To the best of our knowledge, this is the first study to apply a PINNs-based approach to the two-dimensional unsteady planar shock refraction problem, offering new insights into the application of PINNs to complex shock-interface interactions. Numerical experiments demonstrate that ST-PINNs more accurately capture sharp discontinuities and substantially reduce solution errors in hydrodynamic problems involving multiple discontinuities.

physics.flu-dyn↗

arXiv2025

SLearnLLM: A Self-Learning Framework for Efficient Domain-Specific Adaptation of Large Language Models

When using supervised fine-tuning (SFT) to adapt large language models (LLMs) to specific domains, a significant challenge arises: should we use the entire SFT dataset for fine-tuning? Common practice often involves fine-tuning directly on the entire dataset due to limited information on the LLM's past training data. However, if the SFT dataset largely overlaps with the model's existing knowledge, the performance gains are minimal, leading to wasted computational resources. Identifying the unknown knowledge within the SFT dataset and using it to fine-tune the model could substantially improve the training efficiency. To address this challenge, we propose a self-learning framework for LLMs inspired by human learning pattern. This framework takes a fine-tuning (SFT) dataset in a specific domain as input. First, the LLMs answer the questions in the SFT dataset. The LLMs then objectively grade the responses and filter out the incorrectly answered QA pairs. Finally, we fine-tune the LLMs based on this filtered QA set. Experimental results in the fields of agriculture and medicine demonstrate that our method substantially reduces training time while achieving comparable improvements to those attained with full dataset fine-tuning. By concentrating on the unknown knowledge within the SFT dataset, our approach enhances the efficiency of fine-tuning LLMs.

cs.CL↗

arXiv2025

REPA Works Until It Doesn't: Early-Stopped, Holistic Alignment Supercharges Diffusion Training

Diffusion Transformers (DiTs) deliver state-of-the-art image quality, yet their training remains notoriously slow. A recent remedy -- representation alignment (REPA) that matches DiT hidden features to those of a non-generative teacher (e.g. DINO) -- dramatically accelerates the early epochs but plateaus or even degrades performance later. We trace this failure to a capacity mismatch: once the generative student begins modelling the joint data distribution, the teacher's lower-dimensional embeddings and attention patterns become a straitjacket rather than a guide. We then introduce HASTE (Holistic Alignment with Stage-wise Termination for Efficient training), a two-phase schedule that keeps the help and drops the hindrance. Phase I applies a holistic alignment loss that simultaneously distills attention maps (relational priors) and feature projections (semantic anchors) from the teacher into mid-level layers of the DiT, yielding rapid convergence. Phase II then performs one-shot termination that deactivates the alignment loss, once a simple trigger such as a fixed iteration is hit, freeing the DiT to focus on denoising and exploit its generative capacity. HASTE speeds up training of diverse DiTs without architecture changes. On ImageNet 256X256, it reaches the vanilla SiT-XL/2 baseline FID in 50 epochs and matches REPA's best FID in 500 epochs, amounting to a 28X reduction in optimization steps. HASTE also improves text-to-image DiTs on MS-COCO, demonstrating to be a simple yet principled recipe for efficient diffusion training across various tasks. Our code is available at https://github.com/NUS-HPC-AI-Lab/HASTE .

cs.CV↗

arXiv2025

Silicon photonics LMA amplifiers: High power, high gain, low noise and tunable polarization sensitivity

High-power amplifiers are of great importance in many optical systems deployed in optical sensing, ranging, medical surgery, material processing and more. Likewise, high-gain, low-noise amplifiers with low polarization dependence are critical components of long-range optical communication systems. Integrated photonic solutions show great potential in challenging application fields thanks to their drastic reduction in size, weight and cost, but this comes at the expense of low optical power due to reduced energy storage capacity in small devices. Recently, the large mode area (LMA) technology, which is well known for dramatically increasing the output power of fiber amplifiers by orders of magnitude, has been brought to the chip-level. With a large optical mode in the gain medium, the energy storage capacity and saturation power are increased significantly, allowing for high-power amplification with watt-level output power directly from the chip. In this work we demonstrate that a single integrated LMA amplifier is capable of both high-power amplification up to 800 mW with output saturation powers > 115 mW as well as high small-signal net gain up to 30 dB and low-noise amplification with noise figures < 4 dB. The LMA design further allows for a tuning of the polarization dependent gain (PDG) by adjusting the pump power and pump polarization, making it possible to completely nullify the polarization-sensitivity for any given signal power from the μW to mW-level. The power and noise performance achieved surpasses the performance level of many integrated amplifiers and fiber-based amplifiers. We believe that the tunability of the PDG combined with high gain and low noise figure can play a disruptive role for next-generation integrated amplifiers in telecommunication networks.

physics.optics↗

arXiv2025

Quantifying the Capability Boundary of DeepSeek Models: An Application-Driven Performance Analysis

DeepSeek-R1, known for its low training cost and exceptional reasoning capabilities, has achieved state-of-the-art performance on various benchmarks. However, detailed evaluations for DeepSeek Series models from the perspective of real-world applications are lacking, making it challenging for users to select the most suitable DeepSeek models for their specific needs. To address this gap, we presents the first comprehensive evaluation of the DeepSeek and its related models (including DeepSeek-V3, DeepSeek-R1, DeepSeek-R1-Distill-Qwen series, DeepSeek-R1-Distill-Llama series, their corresponding 4-bit quantized models, and the reasoning model QwQ-32B) using our enhanced A-Eval benchmark, A-Eval-2.0. Our systematic analysis reveals several key insights: (1) Given identical model architectures and training data, larger parameter models demonstrate superior performance, aligning with the scaling law. However, smaller models may achieve enhanced capabilities when employing optimized training strategies and higher-quality data; (2) Reasoning-enhanced model show significant performance gains in logical reasoning tasks but may underperform in text understanding and generation tasks; (3) As the data difficulty increases, distillation or reasoning enhancements yield higher performance gains for the models. Interestingly, reasoning enhancements can even have a negative impact on simpler problems; (4) Quantization impacts different capabilities unevenly, with significant drop on logical reasoning and minimal impact on text generation. Based on these results and findings, we design an model selection handbook enabling users to select the most cost-effective models without efforts.

cs.AI↗

arXiv2025

Safety Evaluation and Enhancement of DeepSeek Models in Chinese Contexts

DeepSeek-R1, renowned for its exceptional reasoning capabilities and open-source strategy, is significantly influencing the global artificial intelligence landscape. However, it exhibits notable safety shortcomings. Recent research conducted by Robust Intelligence, a subsidiary of Cisco, in collaboration with the University of Pennsylvania, revealed that DeepSeek-R1 achieves a 100\% attack success rate when processing harmful prompts. Furthermore, multiple security firms and research institutions have identified critical security vulnerabilities within the model. Although China Unicom has uncovered safety vulnerabilities of R1 in Chinese contexts, the safety capabilities of the remaining distilled models in the R1 series have not yet been comprehensively evaluated. To address this gap, this study utilizes the comprehensive Chinese safety benchmark CHiSafetyBench to conduct an in-depth safety evaluation of the DeepSeek-R1 series distilled models. The objective is to assess the safety capabilities of these models in Chinese contexts both before and after distillation, and to further elucidate the adverse effects of distillation on model safety. Building on these findings, we implement targeted safety enhancements for the entire DeepSeek-R1 model series. Evaluation results indicate that the enhanced models achieve significant improvements in safety while maintaining reasoning capabilities without notable degradation. We open-source the safety-enhanced models at https://github.com/UnicomAI/DeepSeek-R1-Safe to serve as a valuable resource for future research and optimization of DeepSeek models.

cs.CL↗

arXiv2025

Towards Graph Foundation Models: Training on Knowledge Graphs Enables Transferability to General Graphs

Inspired by the success of large language models, there is a trend toward developing graph foundation models to conduct diverse downstream tasks in various domains. However, current models often require extra fine-tuning to apply their learned structural and semantic representations to new graphs, which limits their versatility. Recent breakthroughs in zero-shot inductive reasoning on knowledge graphs (KGs), offer us a new perspective on extending KG reasoning to general graph applications. In this paper, we introduce SCR, a unified graph reasoning framework designed to train on knowledge graphs and effectively generalize across a wide range of graph tasks and domains. We begin by designing the task-specific KG structures to establish a unified topology for different task formats. Then we propose semantic-conditioned message passing, a novel mechanism addressing the inherent semantic isolation in traditional KG reasoning, by jointly modeling structural and semantic invariance patterns in graph representations. To demonstrate the effectiveness, we evaluate the inductive reasoning capability of SCR using 38 diverse graph datasets, covering node-level, link-level, and graph-level tasks across multiple domains. Our results show substantial performance gains over existing foundation models and supervised baselines, highlighting the efficacy and adaptability of our approach.

cs.LG↗

arXiv2025

Unsupervised Learning for Class Distribution Mismatch

Class distribution mismatch (CDM) refers to the discrepancy between class distributions in training data and target tasks. Previous methods address this by designing classifiers to categorize classes known during training, while grouping unknown or new classes into an "other" category. However, they focus on semi-supervised scenarios and heavily rely on labeled data, limiting their applicability and performance. To address this, we propose Unsupervised Learning for Class Distribution Mismatch (UCDM), which constructs positive-negative pairs from unlabeled data for classifier training. Our approach randomly samples images and uses a diffusion model to add or erase semantic classes, synthesizing diverse training pairs. Additionally, we introduce a confidence-based labeling mechanism that iteratively assigns pseudo-labels to valuable real-world data and incorporates them into the training process. Extensive experiments on three datasets demonstrate UCDM's superiority over previous semi-supervised methods. Specifically, with a 60% mismatch proportion on Tiny-ImageNet dataset, our approach, without relying on labeled data, surpasses OpenMatch (with 40 labels per class) by 35.1%, 63.7%, and 72.5% in classifying known, unknown, and new classes.

cs.CV↗

arXiv2025

Safety Evaluation of DeepSeek Models in Chinese Contexts

Recently, the DeepSeek series of models, leveraging their exceptional reasoning capabilities and open-source strategy, is reshaping the global AI landscape. Despite these advantages, they exhibit significant safety deficiencies. Research conducted by Robust Intelligence, a subsidiary of Cisco, in collaboration with the University of Pennsylvania, revealed that DeepSeek-R1 has a 100\% attack success rate when processing harmful prompts. Additionally, multiple safety companies and research institutions have confirmed critical safety vulnerabilities in this model. As models demonstrating robust performance in Chinese and English, DeepSeek models require equally crucial safety assessments in both language contexts. However, current research has predominantly focused on safety evaluations in English environments, leaving a gap in comprehensive assessments of their safety performance in Chinese contexts. In response to this gap, this study introduces CHiSafetyBench, a Chinese-specific safety evaluation benchmark. This benchmark systematically evaluates the safety of DeepSeek-R1 and DeepSeek-V3 in Chinese contexts, revealing their performance across safety categories. The experimental results quantify the deficiencies of these two models in Chinese contexts, providing key insights for subsequent improvements. It should be noted that, despite our efforts to establish a comprehensive, objective, and authoritative evaluation benchmark, the selection of test samples, characteristics of data distribution, and the setting of evaluation criteria may inevitably introduce certain biases into the evaluation results. We will continuously optimize the evaluation benchmark and periodically update this report to provide more comprehensive and accurate assessment outcomes. Please refer to the latest version of the paper for the most recent evaluation results and conclusions.

cs.CL↗

arXiv2025

STRGCN: Capturing Asynchronous Spatio-Temporal Dependencies for Irregular Multivariate Time Series Forecasting

Irregular multivariate time series (IMTS) are prevalent in real-world applications across many fields, where varying sensor frequencies and asynchronous measurements pose significant modeling challenges. Existing solutions often rely on a pre-alignment strategy to normalize data, which can distort intrinsic patterns and escalate computational and memory demands. Addressing these limitations, we introduce STRGCN, a Spatio-Temporal Relational Graph Convolutional Network that avoids pre-alignment and directly captures the complex interdependencies in IMTS by representing them as a fully connected graph. Each observation is represented as a node, allowing the model to effectively handle misaligned timestamps by mapping all inter-node relationships, thus faithfully preserving the asynchronous nature of the data. Moreover, we enhance this model with a hierarchical ``Sandwich'' structure that strategically aggregates nodes to optimize graph embeddings, reducing computational overhead while maintaining detailed local and global context. Extensive experiments on four public datasets demonstrate that STRGCN achieves state-of-the-art accuracy, competitive memory usage and training speed.

cs.LG↗

arXiv2025

FilterTS: Comprehensive Frequency Filtering for Multivariate Time Series Forecasting

Multivariate time series forecasting is crucial across various industries, where accurate extraction of complex periodic and trend components can significantly enhance prediction performance. However, existing models often struggle to capture these intricate patterns. To address these challenges, we propose FilterTS, a novel forecasting model that utilizes specialized filtering techniques based on the frequency domain. FilterTS introduces a Dynamic Cross-Variable Filtering Module, a key innovation that dynamically leverages other variables as filters to extract and reinforce shared variable frequency components across variables in multivariate time series. Additionally, a Static Global Filtering Module captures stable frequency components, identified throughout the entire training set. Moreover, the model is built in the frequency domain, converting time-domain convolutions into frequency-domain multiplicative operations to enhance computational efficiency. Extensive experimental results on eight real-world datasets have demonstrated that FilterTS significantly outperforms existing methods in terms of prediction accuracy and computational efficiency.

cs.LG↗

arXiv2025

LipidBERT: A Lipid Language Model Pre-trained on METiS de novo Lipid Library

In this study, we generate and maintain a database of 10 million virtual lipids through METiS's in-house de novo lipid generation algorithms and lipid virtual screening techniques. These virtual lipids serve as a corpus for pre-training, lipid representation learning, and downstream task knowledge transfer, culminating in state-of-the-art LNP property prediction performance. We propose LipidBERT, a BERT-like model pre-trained with the Masked Language Model (MLM) and various secondary tasks. Additionally, we compare the performance of embeddings generated by LipidBERT and PhatGPT, our GPT-like lipid generation model, on downstream tasks. The proposed bilingual LipidBERT model operates in two languages: the language of ionizable lipid pre-training, using in-house dry-lab lipid structures, and the language of LNP fine-tuning, utilizing in-house LNP wet-lab data. This dual capability positions LipidBERT as a key AI-based filter for future screening tasks, including new versions of METiS de novo lipid libraries and, more importantly, candidates for in vivo testing for orgran-targeting LNPs. To the best of our knowledge, this is the first successful demonstration of the capability of a pre-trained language model on virtual lipids and its effectiveness in downstream tasks using web-lab data. This work showcases the clever utilization of METiS's in-house de novo lipid library as well as the power of dry-wet lab integration.

cs.CL↗

arXiv2025

Convergence analysis of nonmonotone proximal gradient methods under local Lipschitz continuity and Kurdyka--Łojasiewicz property

The proximal gradient method is a standard approach for solving composite minimization problems in which the objective function is the sum of a continuously differentiable function and a lower semicontinuous, extended-valued function. The traditional convergence theory for both monotone and nonmonotone variants replies heavily on the assumption of global Lipschitz continuity of the gradient of the smooth part of the objective function. Recent work has shown that monotone proximal gradient methods converge globally only when the local (rather than global) Lipschitz continuity is assumed, provided that the Kurdyka--Łojasiewicz (KL) property holds. However, these results have not been extended to nonmonotone proximal gradient (NPG) methods. In this manuscript, we consider two types of NPG methods: those combined with the average line search and the max line search, respectively. By partitioning indices into two subsets, one of which aims to achieve a sufficient decrease in the functional sequence, we establish global convergence and rate-of-convergence results using the local Lipschitz continuity and the KL property, without requiring boundedness of the iterates. While finalizing this work, we noticed that [18] presented analogous results for the NPG method with average line search, but with a different partitioning strategy. Together, we confidently conclude that the convergence theory of the NPG method is independent on index partitioning choices.

math.OC↗

arXiv2025

A Large Vision-Language Model based Environment Perception System for Visually Impaired People

It is a challenging task for visually impaired people to perceive their surrounding environment due to the complexity of the natural scenes. Their personal and social activities are thus highly limited. This paper introduces a Large Vision-Language Model(LVLM) based environment perception system which helps them to better understand the surrounding environment, by capturing the current scene they face with a wearable device, and then letting them retrieve the analysis results through the device. The visually impaired people could acquire a global description of the scene by long pressing the screen to activate the LVLM output, retrieve the categories of the objects in the scene resulting from a segmentation model by tapping or swiping the screen, and get a detailed description of the objects they are interested in by double-tapping the screen. To help visually impaired people more accurately perceive the world, this paper proposes incorporating the segmentation result of the RGB image as external knowledge into the input of LVLM to reduce the LVLM's hallucination. Technical experiments on POPE, MME and LLaVA-QA90 show that the system could provide a more accurate description of the scene compared to Qwen-VL-Chat, exploratory experiments show that the system helps visually impaired people to perceive the surrounding environment effectively.

cs.CV↗