arXiv subjects

Jun Luo

Publications and source records attributed to Jun Luo.

At least 73 records · Page 4Linked to original sources

arXiv2026

SmoothCLAP: Soft-Target Enhanced Contrastive Language\--Audio Pretraining for Affective Computing

The ambiguity of human emotions poses several challenges for machine learning models, as they often overlap and lack clear delineating boundaries. Contrastive language-audio pretraining (CLAP) has emerged as a key technique for generalisable emotion recognition. However, as conventional CLAP enforces a strict one-to-one alignment between paired audio-text samples, it overlooks intra-modal similarity and treats all non-matching pairs as equally negative. This conflicts with the fuzzy boundaries between different emotions. To address this limitation, we propose SmoothCLAP, which introduces softened targets derived from intra-modal similarity and paralinguistic features. By combining these softened targets with conventional contrastive supervision, SmoothCLAP learns embeddings that respect graded emotional relationships, while retaining the same inference pipeline as CLAP. Experiments on eight affective computing tasks across English and German demonstrate that SmoothCLAP is consistently achieving superior performance. Our results highlight that leveraging soft supervision is a promising strategy for building emotion-aware audio-text models.

cs.SD↗

arXiv2026

Branch, or Layer? Zeroth-Order Optimization for Continual Learning of Vision-Language Models

Vision-Language Continual Learning (VLCL) has attracted significant research attention for its robust capabilities, and the adoption of Parameter-Efficient Fine-Tuning (PEFT) strategies is enabling these models to achieve competitive performance with substantially reduced resource consumption. However, dominated First-Order (FO) optimization is prone to trap models in suboptimal local minima, especially in limited exploration subspace within PEFT. To overcome this challenge, this paper pioneers a systematic exploration of adopting Zeroth-Order (ZO) optimization for PEFT-based VLCL. We first identify the incompatibility of naive full-ZO adoption in VLCL due to optimization process instability. We then investigate the application of ZO optimization from a modality branch-wise to a fine-grained layer-wise across various training units to identify an optimal strategy. Besides, a key theoretical insight reveals that vision modality exhibit higher variance than language counterparts in VLCL during the ZO optimization process, and we propose a modality-aware ZO strategy, which adopts gradient sign normalization in ZO and constrains vision modality perturbation to further improve performance. Benefiting from the adoption of ZO optimization, PEFT-based VLCL fulfills better ability to escape local minima during the optimization process, extensive experiments on four benchmarks demonstrate that our method achieves state-of-the-art results.

cs.CV↗

arXiv2026

A Classification of Fractal Squares

Let $λ_K:\bbR^2\rightarrow\{0,1,\ldots\}\cup\{\infty\}$ be the lambda function of a planar comapctum $K$, as defined in MR4488162. It is known that a planar continuum is locally connected if and only if its lambda function vanishes everywhere, or equivalently, $λ_K(K)=\{0\}$. In this article we show that every fractal square $K$ satisfies $λ_K(K)\subset\{0,1\}$ and find criterions to classify when $λ_K(K)$ equals $\{0\}$, $\{1\}$ or $\{0,1\}$. Here for any integer $N\ge2$ and any set $\Dc=\left\{(i,j): 0\le i,j\le N-1\right\}$ with cardinality $\ge2$, if we set $K^{(0)}=[0,1]^2$ and $\displaystyle K^{(n)}=\left\{\frac{x+d}{N}: x\in K^{(n-1)}, d\in\Dc\right\}(n\ge1)$ then $K=\bigcap_nK^{(n)}$ is called a fractal square.

math.GN↗

arXiv2025

To Define the Core Entropy for All Polynomials Having a Connected Julia Set

For all polynomials $f$ with ${\rm deg}(f)\ge2$ that have a connected filled Julia set $K$, we introduce a new quantity $h_{\rm GCE}(f)$, such that $h_{\rm GCE}\left(f^n\right)=n\cdot h_{\rm GCE}(f)$ for all $n\ge1$ and $h_{\rm GCE}(f)=h_{\rm GCE}(g)$ for $J$-equivalent $f$ and $g$. When the coefficients and the critical points of $f$ are real, $h_{\rm GCE}(f)=h(K\cap\mathbb{R},f)$. When $f$ is post-critically finite, $h_{\rm GCE}(f)$ equals the core entropy $h(\mathcal{H}(f),f)$, where $\mathcal{H}(f)$ is the Hubbard tree. For $f_c(z)=z^2+c$ with $c$ varying in the Mandelbrot set $\mathcal{M}$, the entropy map $c\mapsto h_{\rm GCE}(f_c)$ is not continuous. However, its lower envelope $h_{\rm core}:\mathcal{M}\rightarrow\mathbb{R}$ given by $h_{\rm core}(c)=\inf\left\{t:\ \exists\ c_n\ne c\ \text{with}\ c_n\rightarrow c\ \text{and}\ t=\lim\limits_{n\rightarrow\infty}h_{\rm GCE}\left(f_{c_n}\right)\right\}$ is continuous over $\mathcal{M}$ and has three properties. First, every $h_{\rm core}^{-1}([0,s])$ with $s\ge0$ is connected. In particular, $h_{\rm core}^{-1}(0)$ coincides with the central molecule. Second, $h_{\rm core}(c)=h(\mathbb{R},f_c)$ for $c\in[-2,\frac14]$. Third, $h_{\rm core}(c)=h(\mathcal{H}(f_c),f_c)$ for post-critically finite $f_c$.

math.DS↗

arXiv2025

DeepGuard: Defending Deep Joint Source-Channel Coding Against Eavesdropping at Physical-Layer

Deep joint source-channel coding (DeepJSCC) has emerged as a promising paradigm for efficient and robust information transmission. However, its intrinsic characteristics also pose new security challenges, notably an increased vulnerability to eavesdropping attacks. Existing studies on defending against eavesdropping attacks in DeepJSCC, while demonstrating certain effectiveness, often incur considerable computational overhead or introduce performance trade-offs that may adversely affect legitimate users. In this paper, we present DeepGuard, to the best of our knowledge, the first physical-layer defense framework for DeepJSCC against eavesdropping attacks, validated through over-the-air experiments using software-defined radios (SDRs). Considering that existing eavesdropping attacks against DeepJSCC are limited to simulation under ideal channels, we take a step further by identifying and implementing four representative types of attacks under various configurations in orthogonal frequency-division multiplexing systems. These attacks are evaluated over-the-air under diverse scenarios, allowing us to comprehensively characterize the real-world threat landscape. To mitigate these threats, DeepGuard introduces a novel preamble perturbation mechanism that modifies the preamble shared only between legitimate transceivers. To realize it, we first conduct a theoretical analysis of the perturbation's impact on the signals intercepted by the eavesdropper. Building upon this, we develop an end-to-end perturbation optimization algorithm that significantly degrades eavesdropping performance while preserving reliable communication for legitimate users. We prototype DeepGuard using SDRs and conduct extensive over-the-air experiments in practical scenarios. Extensive experiments demonstrate that DeepGuard effectively mitigates eavesdropping threats.

eess.SP↗

arXiv2025

Where to Start Alignment? Diffusion Large Language Model May Demand a Distinct Position

Diffusion Large Language Models (dLLMs) have recently emerged as a competitive non-autoregressive paradigm due to their unique training and inference approach. However, there is currently a lack of safety study on this novel architecture. In this paper, we present the first analysis of dLLMs' safety performance and propose a novel safety alignment method tailored to their unique generation characteristics. Specifically, we identify a critical asymmetry between the defender and attacker in terms of security. For the defender, we reveal that the middle tokens of the response, rather than the initial ones, are more critical to the overall safety of dLLM outputs; this seems to suggest that aligning middle tokens can be more beneficial to the defender. The attacker, on the contrary, may have limited power to manipulate middle tokens, as we find dLLMs have a strong tendency towards a sequential generation order in practice, forcing the attack to meet this distribution and diverting it from influencing the critical middle tokens. Building on this asymmetry, we introduce Middle-tOken Safety Alignment (MOSA), a novel method that directly aligns the model's middle generation with safe refusals exploiting reinforcement learning. We implement MOSA and compare its security performance against eight attack methods on two benchmarks. We also test the utility of MOSA-aligned dLLM on coding, math, and general reasoning. The results strongly prove the superiority of MOSA.

cs.CR↗

arXiv2025

Attack via Overfitting: 10-shot Benign Fine-tuning to Jailbreak LLMs

Despite substantial efforts in safety alignment, recent research indicates that Large Language Models (LLMs) remain highly susceptible to jailbreak attacks. Among these attacks, finetuning-based ones that compromise LLMs' safety alignment via fine-tuning stand out due to its stable jailbreak performance. In particular, a recent study indicates that fine-tuning with as few as 10 harmful question-answer (QA) pairs can lead to successful jailbreaking across various harmful questions. However, such malicious fine-tuning attacks are readily detectable and hence thwarted by moderation models. In this paper, we demonstrate that LLMs can be jailbroken by fine-tuning with only 10 benign QA pairs; our attack exploits the increased sensitivity of LLMs to fine-tuning data after being overfitted. Specifically, our fine-tuning process starts with overfitting an LLM via fine-tuning with benign QA pairs involving identical refusal answers. Further fine-tuning is then performed with standard benign answers, causing the overfitted LLM to forget the refusal attitude and thus provide compliant answers regardless of the harmfulness of a question. We implement our attack on the ten LLMs and compare it with five existing baselines. Experiments demonstrate that our method achieves significant advantages in both attack effectiveness and attack stealth. Our findings expose previously unreported security vulnerabilities in current LLMs and provide a new perspective on understanding how LLMs' security is compromised, even with benign fine-tuning. Our code is available at https://github.com/ZHIXINXIE/tenBenign.

cs.CR↗

arXiv2025

LLM-Driven Stationarity-Aware Expert Demonstrations for Multi-Agent Reinforcement Learning in Mobile Systems

Multi-agent reinforcement learning (MARL) has been increasingly adopted in many real-world applications. While MARL enables decentralized deployment on resource-constrained edge devices, it suffers from severe non-stationarity due to the synchronous updates of agent policies. This non stationarity results in unstable training and poor policy con vergence, especially as the number of agents increases. In this paper, we propose RELED, a scalable MARL framework that integrates large language model (LLM)-driven expert demonstrations with autonomous agent exploration. RELED incorporates a Stationarity-Aware Expert Demonstration module, which leverages theoretical non-stationarity bounds to enhance the quality of LLM-generated expert trajectories, thus providing high reward and training-stable samples for each agent. Moreover, a Hybrid Expert-Agent Policy Optimization module adaptively balances each agent's learning from both expert-generated and agent-generated trajectories, accelerating policy convergence and improving generalization. Extensive experiments with real city networks based on OpenStreetMap demonstrate that RELED achieves superior performance compared to state-of-the-art MARL methods.

cs.LG↗

arXiv2025

Co-Reinforcement Learning for Unified Multimodal Understanding and Generation

This paper presents a pioneering exploration of reinforcement learning (RL) via group relative policy optimization for unified multimodal large language models (ULMs), aimed at simultaneously reinforcing generation and understanding capabilities. Through systematic pilot studies, we uncover the significant potential of ULMs to enable the synergistic co-evolution of dual capabilities within a shared policy optimization framework. Building on this insight, we introduce CoRL, a co-reinforcement learning framework comprising a unified RL stage for joint optimization and a refined RL stage for task-specific enhancement. With the proposed CoRL, our resulting model, ULM-R1, achieves average improvements of 7% on three text-to-image generation datasets and 23% on nine multimodal understanding benchmarks. These results demonstrate the effectiveness of CoRL and highlight the substantial benefit of reinforcement learning in facilitating cross-task synergy and optimization for ULMs. Code is available at https://github.com/mm-vl/ULM-R1.

cs.CV↗

arXiv2025

Joint Optimization of Cooperation Efficiency and Communication Covertness for Target Detection with AUVs

This paper investigates underwater cooperative target detection using autonomous underwater vehicles (AUVs), with a focus on the critical trade-off between cooperation efficiency and communication covertness. To tackle this challenge, we first formulate a joint trajectory and power control optimization problem, and then present an innovative hierarchical action management framework to solve it. According to the hierarchical formulation, at the macro level, the master AUV models the agent selection process as a Markov decision process and deploys the proximal policy optimization algorithm for strategic task allocation. At the micro level, each selected agent's decentralized decision-making is modeled as a partially observable Markov decision process, and a multi-agent proximal policy optimization algorithm is used to dynamically adjust its trajectory and transmission power based on its local observations. Under the centralized training and decentralized execution paradigm, our target detection framework enables adaptive covert cooperation while satisfying both energy and mobility constraints. By comprehensively modeling the considered system, the involved signals and tasks, as well as energy consumption, theoretical insights and practical solutions for the efficient and secure operation of multiple AUVs are provided, offering significant implications for the execution of underwater covert communication tasks.

cs.LG↗

arXiv2025

Development of 3D Pixel Sensors via an 8-inch CMOS-Compatible Process

In the construction of High-Luminosity Large Hadron Collider (HL-LHC) and Future Circular Collider (FCC) experiments, 3D pixel sensors have become indispensable components due to their superior radiation hardness, fast response, and low power consumption. However, there are still significant challenges in the process of 3D sensors manufacturing. In this work, single devices and arrays of 3D sensors based on 30 $μ$m epitaxial silicon wafer have been designed, simulated, fabricated, and tested. This process was developed on the 8-inch CMOS process platform of the Institute of Microelectronics of the Chinese Academy of Sciences (IMECAS). The key processes include Deep Reactive Ion Etching (DRIE) with the Bosch process, in-situ doping, and an innovative back-etching. After testing the 3D pixel sensors, we have summarized the leakage current and capacitance of devices with different sizes with respect to bias voltages. We also found that the fabricated devices were almost all successfully produced, which laid a strong foundation for subsequent large-scale mass production.

physics.ins-det↗

arXiv2025

Traj-Transformer: Diffusion Models with Transformer for GPS Trajectory Generation

The widespread use of GPS devices has driven advances in spatiotemporal data mining, enabling machine learning models to simulate human decision making and generate realistic trajectories, addressing both data collection costs and privacy concerns. Recent studies have shown the promise of diffusion models for high-quality trajectory generation. However, most existing methods rely on convolution based architectures (e.g. UNet) to predict noise during the diffusion process, which often results in notable deviations and the loss of fine-grained street-level details due to limited model capacity. In this paper, we propose Trajectory Transformer, a novel model that employs a transformer backbone for both conditional information embedding and noise prediction. We explore two GPS coordinate embedding strategies, location embedding and longitude-latitude embedding, and analyze model performance at different scales. Experiments on two real-world datasets demonstrate that Trajectory Transformer significantly enhances generation quality and effectively alleviates the deviation issues observed in prior approaches.

cs.LG↗

arXiv2025

More Than Meets the Eye? Uncovering the Reasoning-Planning Disconnect in Training Vision-Language Driving Models

Vision-Language Model (VLM) driving agents promise explainable end-to-end autonomy by first producing natural-language reasoning and then predicting trajectory planning. However, whether planning is causally driven by this reasoning remains a critical but unverified assumption. To investigate this, we build DriveMind, a large-scale driving Visual Question Answering (VQA) corpus with plan-aligned Chain-of-Thought (CoT), automatically generated from nuPlan. Our data generation process converts sensors and annotations into structured inputs and, crucially, separates priors from to-be-reasoned signals, enabling clean information ablations. Using DriveMind, we train representative VLM agents with Supervised Fine-Tuning (SFT) and Group Relative Policy Optimization (GRPO) and evaluate them with nuPlan's metrics. Our results, unfortunately, indicate a consistent causal disconnect in reasoning-planning: removing ego/navigation priors causes large drops in planning scores, whereas removing CoT produces only minor changes. Attention analysis further shows that planning primarily focuses on priors rather than the CoT. Based on this evidence, we propose the Reasoning-Planning Decoupling Hypothesis, positing that the training-yielded reasoning is an ancillary byproduct rather than a causal mediator. To enable efficient diagnosis, we also introduce a novel, training-free probe that measures an agent's reliance on priors by evaluating its planning robustness against minor input perturbations. In summary, we provide the community with a new dataset and a diagnostic tool to evaluate the causal fidelity of future models.

cs.AI↗

arXiv2025

A Survey on the Topology of Fractal Squares

We consider a special type of self-similar sets, called fractal squares, and give a brief review on recent results and unsolved issues with an emphasis on their topological properties.

math.GN↗

arXiv2025

Rehearsal-free and Task-free Online Continual Learning With Contrastive Prompt

The main challenge of continual learning is \textit{catastrophic forgetting}. Because of processing data in one pass, online continual learning (OCL) is one of the most difficult continual learning scenarios. To address catastrophic forgetting in OCL, some existing studies use a rehearsal buffer to store samples and replay them in the later learning process, other studies do not store samples but assume a sequence of learning tasks so that the task identities can be explored. However, storing samples may raise data security or privacy concerns and it is not always possible to identify the boundaries between learning tasks in one pass of data processing. It motivates us to investigate rehearsal-free and task-free OCL (F2OCL). By integrating prompt learning with an NCM classifier, this study has effectively tackled catastrophic forgetting without storing samples and without usage of task boundaries or identities. The extensive experimental results on two benchmarks have demonstrated the effectiveness of the proposed method.

cs.LG↗

arXiv2025

Dagger Behind Smile: Fool LLMs with a Happy Ending Story

The wide adoption of Large Language Models (LLMs) has attracted significant attention from $\textit{jailbreak}$ attacks, where adversarial prompts crafted through optimization or manual design exploit LLMs to generate malicious contents. However, optimization-based attacks have limited efficiency and transferability, while existing manual designs are either easily detectable or demand intricate interactions with LLMs. In this paper, we first point out a novel perspective for jailbreak attacks: LLMs are more responsive to $\textit{positive}$ prompts. Based on this, we deploy Happy Ending Attack (HEA) to wrap up a malicious request in a scenario template involving a positive prompt formed mainly via a $\textit{happy ending}$, it thus fools LLMs into jailbreaking either immediately or at a follow-up malicious request. This has made HEA both efficient and effective, as it requires only up to two turns to fully jailbreak LLMs. Extensive experiments show that our HEA can successfully jailbreak on state-of-the-art LLMs, including GPT-4o, Llama3-70b, Gemini-pro, and achieves 88.79% attack success rate on average. We also provide quantitative explanations for the success of HEA.

cs.CL↗

arXiv2025

Spin-supersolidity induced quantum criticality and magnetocaloric effect in the triangular-lattice antiferromagnet Rb$_2$Co(SeO$_3$)$_2$

We performed high-field magnetization, magnetocaloric effect (MCE), and NMR measurements on the Ising triangular-lattice antiferromagnet Rb$_2$Co(SeO$_3$)$_2$. The observations of the 1/3-magnetization plateau, the split NMR lines, and the thermal activation behaviors of the spin-lattice relaxation rate $1/T_1$ between 2 T and 15.8 T provide unambiguous evidence of a gapped up-up-down (UUD) magnetic ordered phase. For fields between 15.8 T and 18.5 T, the anomaly in the magnetic susceptibility, the slow saturation of the NMR line spectral ratio with temperature, and the power-law temperature dependence of $1/T_1$ suggest the ground state to be a spin supersolid with gapless spin excitations. With further increasing the field, the Grüneisen ratio, extracted from the MCE data, reveals a continuous quantum phase transition at $H_{\rm C}\approx$ 19.5 T and a universal quantum critical scaling with the exponents $νz~\approx~$1. Near $H_{\rm C}$, the large high-temperature MCE signal and the broad peaks in the NMR Knight shift and $1/T_1$, manifest the strong spin fluctuations driven by both magnetic frustration and quantum criticality. These results establish Rb$_2$Co(SeO$_3$)$_2$ as a candidate platform for cryogenic magnetocaloric cooling.

cond-mat.str-el↗

arXiv2025

Optimal Data Reduction under Information-Theoretic Criteria

Selecting an optimal subset of features or instances under an information theoretic criterion has become an effective preprocessing strategy for reducing data complexity while preserving essential information. This study investigates two representative problems within this paradigm: feature selection based on the maximum relevance minimum redundancy criterion, and instance selection grounded in the Kullback Leibler divergence. To address the intrinsic nonconvexities of these problems, we develop polyhedral relaxations that yield exact mixed integer linear programming formulations, thereby enabling globally optimal data reduction. By leveraging modern optimization techniques, we further design efficient algorithmic implementations capable of solving practically sized instances. Extensive numerical experiments on both real world and synthetic datasets demonstrate that our method efficiently solves data reduction problems to global optimality, significantly outperforming existing benchmark approaches.

math.OC↗