Instruction tuning has been identified as a crucial technique for optimizing large language models (LLMs) to generate human-aligned responses. Nonetheless, gathering diversified and superior-quality instruction data for such tuning presents notable obstacles, especially in privacy-sensitive domains. Federated instruction tuning ( FedIT ) has emerged as a promising solution by consolidating collaborative training across multiple data owners, resulting in a privacy-enhancing learning model. Existing FedIT studies assume that clients have sufficient training data, however, in reality, many clients only have few-shot samples, leading to either overfitting in federated LLM or degraded performance. At the same time, this federated few-shot environment also increases the risk of training data extraction attacks, as the LLM may well memorize the limited training data. To address these issues, this article proposes a novel federated algorithm, PPFedIT , designed to enhance privacy protection and model performance of federated few-shot learning. PPFedIT comprises three vital steps on the client side: (1) synthetic data generation, which utilizes the strong generation capacity of LLMs to generate synthetic data, aiming to diversify and enrich local data; (2) parameter isolation training, which respectively updates the parameters of a shared global LLM on the synthetic data and the parameters of local LLMs on the local data, consequently mitigating the noise impact of the synthetic data; (3) local aggregation then sharing mechanism, which mixes the parameters of the global LLM and those of a local LLM first, before uploading them to a server for aggregation. This effectively mitigates data extraction attacks. Extensive experiments on three open source datasets demonstrate PPFedIT significantly enhances model performance (averaging 8.4%) and reduces the risk of data extraction attacks (approximately 20%) in practical and challenging federated few-shot scenarios.

JointCloud environments enable scalable deployment of large language models (LLMs) across heterogeneous multi-cloud and edge infrastructures, but they also introduce critical challenges in cross-cloud consistency and reliable service orchestration. In multilingual applications, JointCloud deployment of translation models amplifies discrepancies in instruction-following capability, resulting in cross-node behavioral inconsistency, including instruction drift, off-target generation, and unstable output lengths, which poses a significant challenge to service reliability. In this paper, we propose Auto-Constriction Tuning (ACT), a training-time alignment approach for improving controllable generation in JointCloud LLM serving. ACT introduces task-aware trigger tokens to construct implicit constrained templates during fine-tuning, encoding task-level control signals directly into model parameters. This design shifts consistency control from runtime orchestration to training-time alignment, enabling independently deployed model instances to maintain consistent generation behavior without requiring centralized coordination. We evaluate ACT across diverse LLM backbones and multilingual benchmarks, simulating heterogeneous deployment scenarios in JointCloud environments. Experimental results demonstrate that ACT consistently reduces off-target and over/under generation errors while improving cross-model consistency and zero-shot generalization across unseen translation directions. Our findings suggest that token-level constrained tuning provides an effective and scalable alternative to runtime orchestration for achieving reliable LLM services.

Long Document Question Answering (LDQA) remains a critical challenge for LLMs, as lengthy contexts often exceed context window and contain irrelevant distractors. We propose Topic-Chain Compression, a novel prompt compression framework that organizes documents into topic chains. The method first splits the document into sentences, each labeled with a topic derived by a Large Language Model (LLM). Sentences sharing the same topic are linked sequentially, forming coherent topic chains. Relevance of a chain to a given query is assessed via a Cross-Encoder at the sentence level, with chain scores obtained by aggregation. The most relevant chains are retained to form a compressed prompt for final inference. TCC balances structural coherence with query sensitivity, enabling more accurate and efficient long-document reasoning in LLMs. Empirical results demonstrate the effectiveness of the framework compared with baseline methods.

Integrating memory components into large language models (LLMs) can improve the generation quality for long-term conversations. However, existing memory management methods largely overlook the cognition and regulation of the memory process, lacking the capability to dynamically manage and utilize memory on demand. To address this challenge, this paper approaches Meta-Memory for Memory Management (M$^{4}$), a novel paradigm that equips LLMs with the ability for self-monitoring and self-reflective memory management. In long-term conversations, where dialogue history accumulates continuously, the meta-memory capability of M$^{4}$ enables LLMs to autonomously 1) identify what knowledge needs to be memorized; 2) determine how to construct and store memory; 3) monitor the correctness and validity of the acquired information; and 4) decide when to learn more and how to retrieve information to refine their responses. Experimental results on two long-term conversation datasets and two long-term question-answering datasets demonstrate that our M$^{4}$ significantly enhances the memory management capacity of LLMs in long-term information learning, achieving more efficient storage and higher-quality response generation.

Accurate and robust wireless localization is a critical enabler for emerging 5G/6G applications, including autonomous driving, extended reality, and smart manufacturing. Despite its importance, achieving precise localization across diverse environments remains challenging due to the complex nature of wireless signals and their sensitivity to environmental changes. Existing data-driven approaches often suffer from limited generalization capability, requiring extensive labeled data and struggling to adapt to new scenarios. To address these limitations, we propose SigMap, a multimodal foundation model that introduces two key innovations: (1) A cycle-adaptive masking strategy that dynamically adjusts masking patterns based on channel periodicity characteristics to learn robust wireless representations; (2) A novel "map-as-prompt" framework that integrates 3D geographic information through lightweight soft prompts for effective cross-scenario adaptation. Extensive experiments demonstrate that our model achieves state-of-the-art performance across multiple localization tasks while exhibiting strong zero-shot generalization in unseen environments, significantly outperforming both supervised and self-supervised baselines by considerable margins.

Mixture-of-Experts (MoE) models enable scalable performance by activating large parameter sets sparsely, minimizing computational overhead. To mitigate the prohibitive cost of training MoEs from scratch, recent work employs upcycling, reusing a single pre-trained dense model by replicating its feed-forward network (FFN) layers into experts. However, this limits expert diversity, as all experts originate from a single pre-trained dense model. This paper addresses this limitation by constructing powerful MoE models using experts sourced from multiple identically-architected but disparate pre-trained models (e.g., Qwen2.5-Coder and Qwen2). A key challenge lies in the fact that these source models occupy disparate, dissonant regions of the parameter space, making direct upcycling prone to severe performance degradation. To overcome this, we propose Symphony-MoE, a novel two-stage framework designed to harmonize these models into a single, coherent expert mixture. First, we establish this harmony in a training-free manner: we construct a shared backbone via a layer-aware fusion strategy and, crucially, alleviate parameter misalignment among experts using activation-based functional alignment. Subsequently, a stage of post-training coordinates the entire architecture. Experiments demonstrate that our method successfully integrates experts from heterogeneous sources, achieving an MoE model that significantly surpasses baselines in multi-domain tasks and out-of-distribution generalization.

Voice cloning for Text-to-Speech (TTS) aims to generate expressive and personalized speech from text using limited data from a target speaker. Federated Learning (FL) offers a collaborative and privacy-preserving framework for this task, but existing approaches suffer from high communication costs and tend to suppress stylistic heterogeneity, resulting in insufficient personalization. To address these issues, we propose Fed-PISA, which stands for Federated Personalized Identity-Style Adaptation. To minimize communication costs, Fed-PISA introduces a disentangled Low-Rank Adaptation (LoRA) mechanism: the speaker's timbre is retained locally through a private ID-LoRA, while only a lightweight style-LoRA is transmitted to the server, thereby minimizing parameter exchange. To harness heterogeneity, our aggregation method, inspired by collaborative filtering, is introduced to create custom models for each client by learning from stylistically similar peers. Experiments show that Fed-PISA improves style expressivity, naturalness, and speaker similarity, outperforming standard federated baselines with minimal communication costs.

Reinforcement learning from human feedback (RLHF) emerges as a promising paradigm for aligning large language models (LLMs). However, a notable challenge in RLHF is overoptimization, where beyond a certain threshold, the pursuit of higher rewards leads to a decline in human preferences. In this paper, we observe the weakness of KL regularization which is commonly employed in existing RLHF methods to address overoptimization. To mitigate this limitation, we scrutinize the RLHF objective in the offline dataset and propose uncertainty-penalized RLHF (UP-RLHF), which incorporates uncertainty regularization during RL-finetuning. To enhance the uncertainty quantification abilities for reward models, we first propose a diverse low-rank adaptation (LoRA) ensemble by maximizing the nuclear norm of LoRA matrix concatenations. Then we optimize policy models utilizing penalized rewards, determined by both rewards and uncertainties provided by the diverse reward LoRA ensembles. Our experimental results, based on two real human preference datasets, showcase the effectiveness of diverse reward LoRA ensembles in quantifying reward uncertainty. Additionally, uncertainty regularization in UP-RLHF proves to be pivotal in mitigating overoptimization, thereby contributing to the overall performance.

Discovering possible candidate elements and capturing their connections to form the output triplets constitute the core challenge of aspect sentiment triplet extraction (ASTE). However, the information encapsulated by prevailing methods within a solitary sentence may often prove insufficient, particularly in complex scenarios characterized by uncommon aspect, opinion terms or intricate syntax patterns. To mitigate these limitations, we advocate incorporating inter-sentence information retrieval to enrich intra-sentence representations within ASTE. One existing study has proposed a method dubbed Retrieval-Based Aspect Sentiment Triplet Extraction via Label Interpolation (RLI), which retrieves triplets from the corpus to augment the representations of a candidate aspect-opinion pair and further improve sentiment prediction. Nevertheless, obtaining data with standard triplets might be challenging in practice. Therefore, we propose an approach, namely Multi-Task ASTE with the Corpus-Enhanced Graph (MACG), to conduct sentence-level retrieval and extract helpful information from unlabeled similar sentences. Specifically, we design a corpus-level enhanced graph to capture inter-sentence information, alongside a local graph preserving intra-sentence information. A graph neural network is subsequently employed to adaptively learn enhanced representations of the target sentence for ASTE. RLI and MACG collaboratively form a comprehensive methodological framework, which is effective in both scenarios with and without standard triplet labels. Extensive experiments on two benchmarks demonstrate the superiority and flexibility of retrieving inter-sentence information, which underscores their potential to advance ASTE by leveraging neighboring information.

Pretrained vision-language models (VLMs), such as CLIP, have shown remarkable potential in few-shot image classification and led to numerous effective transfer learning strategies. These methods leverage the pretrained knowledge of VLMs to enable effective domain adaptation while mitigating overfitting through parameter-efficient tuning or instance-based consistency constraints. However, such regularizations often neglect the geometric structure of data distribution, which may lead to distortion of the overall semantic representation. To overcome this limitation, we propose a novel fine-tuning method, Manifold-Preserving and Sculpting Tuning (MPS-Tuning). Regarding the data distribution in feature space as a semantic manifold, MPS-Tuning explicitly constrains the intrinsic geometry of this manifold while further sculpting it to enhance class separability. Specifically, MPS-Tuning preserves both macroscopic and microscopic topological structures of the original manifold by aligning Gram matrices of features before and after fine-tuning. Theoretically, this constraint is shown to approximate an upper bound of the Gromov-Wasserstein distance. Furthermore, features from the image and text modalities are paired, and pairwise similarities are optimized to enhance the manifold's class discriminability. Extensive experiments demonstrate that MPS-Tuning significantly improves model performance while effectively preserving the structure of the semantic manifold. The code will be released.

Large Language Models (LLMs) have revolutionized numerous fields, yet their adaptation to specialized tasks in privacy-sensitive domains such as healthcare and finance remains constrained due to the scarcity of accessible training data caused by stringent privacy requirements. Secure Multi-party Computation (MPC)-based privacy-preserving machine learning provides theoretical guarantees for the privacy of model parameters and data. However, its application to LLMs has been predominantly limited to inference, as fine-tuning introduces significant efficiency challenges, particularly in backward propagation, optimizer, and self-attention operations. To address these challenges, we propose SecP-Tuning, the first MPC-based framework designed for efficient, privacy-preserving prompt tuning of LLMs. SecP-Tuning innovatively integrates Forward-only Tuning (FoT) through the ''data owner-server interaction" paradigm, effectively removing the need for privacy-preserving computations in backward propagation and optimization processes. Furthermore, it devises an efficient privacy-preserving Random Feature Attention (RFA), effectively mitigating the computational complexity of softmax-based self-attention and circumventing MPC-incompatible nonlinear operations. Experimental results demonstrate that, compared to full-Parameter Supervised Fine-Tuning (SFT) and gradient-based prompt tuning, SecP-Tuning achieves approximately 12$\times$ and 16$\times$ end-to-end acceleration, as well as 18$\times$ and 20$\times$ reductions in communication overhead, respectively. Moreover, it delivers performance comparable to gradient-based methods across multiple few-shot tasks. Additionally, the ''black-box/API-style" privacy-preserving tuning paradigm of SecP-Tuning effectively avoids memory leakage risks caused by gradient/parameter transmission, thereby striking an optimal balance between efficiency, accuracy, deployability, and privacy.

Despite their potential in standardized graph tasks, Large Language Models (LLMs) remain brittle to real-world shifts in node identifiers and task formulation. While deterministic graph tools are invariant to such shifts, extracting topological structures from noisy text is highly fragile for LLMs, which often overfit to surface patterns. Moreover, mitigating these parsing failures via multi-agent systems incurs prohibitive latency. To address this, we propose GRAIN, a single-agent framework optimized via reinforcement learning. GRAIN models reasoning as a semantic parsing and tool-execution pipeline, guided by a Structure Invariance Reward. By validating extracted intermediate graphs against ground-truth topologies, this reward forces the LLM to learn robust text-to-structure mappings rather than memorizing linguistic artifacts. We also introduce GRIT, a benchmark evaluating sensitivity to such linguistic shifts. GRAIN outperforms multi-agent baselines by 16.45% in accuracy with approximately 24% lower latency. Furthermore, it demonstrates superior structural generalization, halving the out-of-distribution (OOD) gap of SFT models (from 15.77% to 7.80%) and maintaining robustness on large-scale graphs beyond the training distribution.

Rapid advances in AI-generated image (AIGI) technology enable highly realistic synthesis, threatening public information integrity and security. Recent studies have demonstrated that incorporating texture-level artifact features alongside semantic features into multimodal large language models (MLLMs) can enhance their AIGI detection capability. However, our preliminary analyses reveal that artifact features exhibit high intra-feature similarity, leading to an almost uniform attention map after the softmax operation. This phenomenon causes attention dilution, thereby hindering effective fusion between semantic and artifact features. To overcome this limitation, we propose a lightweight fusion adapter, TranX-Adapter, which integrates a Task-aware Optimal-Transport Fusion that leverages the Jensen-Shannon divergence between artifact and semantic prediction probabilities as a cost matrix to transfer artifact information into semantic features, and an X-Fusion that employs cross-attention to transfer semantic information into artifact features. Experiments on standard AIGI detection benchmarks upon several advanced MLLMs, show that our TranX-Adapter brings consistent and significant improvements (up to +6% accuracy). Code can be found in supplementary material.

The vehicle privacy protection plays a vital role in releasing or sharing of traffic videos. License plate, as the identifiable mark of a vehicle, contains the most sensitive information for a vehicle. Therefore, masking the license plates is a common way to protect the privacy of corresponding vehicles. However, in the real world scenarios, it is often hard to locate the small and shifting license plates, and therefore precise and cost-effective privacy protection is quite challenging. To address this problem in surveillance video, we fully explore all available spatio-temporal cues and design bidirectional Kalman filter model in the consecutive frames to locate missing license plates. To verify effectiveness of the proposed benchmark, we build a new License Plates Privacy-preserving Dataset (LPPD) collected from various scenes with diverse privacy and utility annotations. We demonstrate that our proposed method show very promising capability of privacy protection on the real world dataset without sacrificing its utility.

As single-center computing approaches power constraints, decentralized training becomes essential. However, traditional Reinforcement Learning (RL) methods, crucial for enhancing large model post-training, cannot adapt to decentralized distributed training due to the tight coupling between parameter learning and rollout sampling. For this, we propose HeteroRL, a heterogeneous RL architecture that decouples these processes, enabling stable training across geographically distributed nodes connected via the Internet. The core component is Group Expectation Policy Optimization (GEPO), an asynchronous RL algorithm robust to latency caused by network delays or heterogeneity in computational resources. Our study reveals that high latency significantly increases KL divergence, leading to higher variance of importance weights and training instability. GEPO mitigates this issue by using group expectation weighting to exponentially reduce the variance of importance weights, with theoretical guarantees. Experiments show GEPO achieves superior stability—only a 3\% performance drop from online to 1800s latency—and reduces the best-to-last gap by 85\% versus GSPO ($\Delta$=1.8 vs. 12.0) while attaining the highest scores, highlighting its effectiveness in decentralized, resource-heterogeneous environments.

Few-shot class-incremental learning (FSCIL) aims to incrementally learn novel classes from limited examples while preserving knowledge of previously learned classes. Existing methods face a critical dilemma: static architectures rely on a constant parameter space to learn from data that arrive sequentially, making them prone to overfitting to the current session, while dynamic architectures continually expand the parameter space, leading to increased complexity. In this study, we explore the potential of Selective State Space Models (SSMs) for FSCIL. Mamba leverages its input-dependent parameters to dynamically adjust its processing patterns and generate content-aware scan patterns without session-wise projector expansion. This enables it to configure distinct processing for base and novel classes, helping preserve existing knowledge while adapting to new ones. To leverage Mamba's potential for FSCIL, we design two key modules: First, we propose a dual selective SSM projector that generates input-conditioned state-space parameters from intermediate features for dynamic adaptation. The dual design structurally decouples base and novel-class processing, employing a frozen base branch to maintain stable base-class features and a dynamic incremental branch that adaptively learns distinctive feature shifts for novel classes. Second, we develop a classsensitive selective scan mechanism to guide dynamic adaptation of the incremental branch. It reduces the disruption to baseclass representations caused by training on novel data, and meanwhile, encourages the selective scan to perform in distinct patterns between base and novel classes. Extensive experiments on miniImageNet, CIFAR-100, and CUB-200 demonstrate that Mamba-FSCIL achieves state-of-the-art performance. The code is available at https://github.com/xiaojieli0903/Mamba-FSCIL.

Learning-based dynamic fault localization techniques play a crucial role in the field of software engineering. These techniques dynamically execute test cases to meticulously extract useful knowledge from the execution information in the program, with the aim of identifying fault locations by leveraging machine learning, deep learning, and large language models. Currently, there is already a flourishing body of research that is intensely focused on learning-based dynamic fault localization. Research literature can be categorized into two main aspects for learning-based dynamic fault localization: data-based enhancements (i.e., the datasets) and model-based enhancements (i.e., the suspiciousness algorithms). Thus, we conduct an extensive literature review on learning-based dynamic fault localization from the aspects of the data task and the model task. Among them, each task is divided into multiple sub-tasks in a systematic manner to comprehensively discuss the details. In addition, we analyze and summarize the datasets and metrics that have been widely used to evaluate the effectiveness of the proposed techniques in recent years, so that researchers can have an intuitive perception of them. We also discuss the present challenges and the directions for future research.

Instruction tuning has emerged as a predominant method for adapting large language models (LLMs) to downstream tasks, with prevailing approaches predominantly relying on scaling up instruction data to enhance model performance. However, growing evidence suggests that indiscriminate data scaling may yield suboptimal results, as the absence of systematic evaluation criteria often leads to redundant or low-quality samples in instruction datasets. Consequently, in this paper, we propose DEFT, a multi-dimensional data selection framework that assesses instruction data from four perspectives: complexity, quality, knowledge and diversity. For complexity and quality, we develop Evol-Ranking to distill ranking capabilities from teacher models (e.g., gpt-3.5-turbo) to specialized student models. Furthermore, we propose refinement distillation to progressively optimize the student model. For knowledge, we define the average negative log-probability of text on a given LLM as knowledge, providing model-aware measurement. For diversity, we first obtain semantic representation of each sample, then calculate the similarity between samples. Finally, we ensemble all dimensions mentioned above through an ensemble scoring mechanism to select the data for instruction fine-tuning. Extensive experiments performed on MT-Bench and AlpacaEval demonstrate that DEFT performs better or on pair with the state-of-the-art open-source alignment models with only 6,000 SFT training samples.