arXiv subjects

Jiale Luo

Publications and source records attributed to Jiale Luo.

5 recordsLinked to original sources

arXiv2026

Rethinking Length-Based Training: Batch Composition and Loss Normalization in Speech Token Language Models

Short-to-long training is a simple curriculum for speech models, but its gains can be difficult to interpret. In speech token language models, length-based training can change the shuffle policy, batch composition, token retention, and token weights under batch-mean loss. We disentangle these factors through matched comparisons. In the tested settings, short-to-long ordering shows no independent benefit when batch composition and token exposure are fixed. First-epoch grouping lowers perplexity for Mimi under batch-mean loss, but this gain is not observed under token-balanced loss. The cross-tokenizer results are consistent with a link between chunk-length variation and token weighting. This work provides a systematic analysis protocol for studying length-based training in variable-length speech models.

cs.CL↗

arXiv2026

CASCADE Against Jailbreaks: Combination Across Stages with Controlled Attack-Defense Evaluation

Defenses against jailbreak attacks on Large Language Models (LLMs) operate at different pipeline stages, such as input modification or output guard, but it remains unclear which defenses to deploy at each stage and how to combine them. Prior empirical studies, fragmented by inconsistent attack-success-rate definitions and experimental settings, have evaluated defenses largely in isolation. Here we present the first systematic study, to our knowledge, of defense combinations both within and across pipeline stages, under a consistent threat model of direct, black-box, single-turn attacks. Our decision framework standardizes evaluation through a principled attack-success-rate formulation with controlled query budgets, together with explicit fairness rules. Across 19 attacks and 15 defenses, we find that no single defense is universally best, but well-chosen combinations achieve substantial safety with minimal utility degradation, yielding practical recommendations for layered defense pipelines.

cs.CR↗

arXiv2026

Locality Matters for Training-Free Audio Token Compression in Audio-Language Models

Audio-language models (ALMs) are increasingly used for audio captioning, question answering, and open-ended audio understanding, but their inference cost remains high when audio inputs are represented as long prefix-token sequences. These audio prefixes consume context budget, increase memory usage, and make deployment harder in resource-constrained or latency-sensitive settings. Existing training-free audio-token reduction methods mainly rely on fixed pooling or score-based pruning. Fixed pooling is content-agnostic, while score-based pruning can preserve isolated salient tokens but discard nearby acoustic context. We propose Local Temporal Bipartite Merging (LTBM), a training-free encoder-space compression method that merges similar nearby audio tokens under an explicit temporal window constraint. Beyond introducing LTBM, we use a controlled Global Merge variant to isolate whether temporal locality itself is a useful inductive bias for audio-token compression. Experiments on AudioCaps, Clotho, and MMAU with Qwen2-Audio show evidence of a task-dependent locality effect: locality-aware merging is more favorable for captioning at several compression settings, especially under stronger compression, while global matching is more competitive for multiple-choice audio understanding. A cross-backbone validation on Audio Flamingo 3 further supports the captioning-side advantage of locality-aware merging under moderate and aggressive compression.

cs.CL↗

arXiv2026

Learn Before Represent: Bridging Generative and Contrastive Learning for Domain-Specific LLM Embeddings

Large Language Models (LLMs) adapted via contrastive learning excel in general representation learning but struggle in vertical domains like chemistry and law, primarily due to a lack of domain-specific knowledge. This work identifies a core bottleneck: the prevailing ``LLM+CL'' paradigm focuses on semantic alignment but cannot perform knowledge acquisition, leading to failures on specialized terminology. To bridge this gap, we propose Learn Before Represent (LBR), a novel two-stage framework. LBR first injects domain knowledge via an Information Bottleneck-Constrained Generative Learning stage, preserving the LLM's causal attention to maximize knowledge acquisition while compressing semantics. It then performs Generative-Refined Contrastive Learning on the compressed representations for alignment. This approach maintains architectural consistency and resolves the objective conflict between generative and contrastive learning. Extensive experiments on medical, chemistry, and code retrieval tasks show that LBR significantly outperforms strong baselines. Our work establishes a new paradigm for building accurate and robust representations in vertical domains.

cs.IR↗

arXiv2024

A Hybrid BPMN-DMN Framework for Secure Inter-organizational Processes and Decisions Collaboration on Permissioned Blockchain

In the rapidly evolving digital business landscape, organizations increasingly need to collaborate across boundaries to achieve complex business objectives, requiring both efficient process coordination and flexible decision-making capabilities. Traditional collaboration approaches face significant challenges in transparency, trust, and decision flexibility, while existing blockchain-based solutions primarily focus on process execution without addressing the integrated decision-making needs of collaborative enterprises. This paper proposes BlockCollab, a novel model-driven framework that seamlessly integrates Business Process Model and Notation (BPMN) with Decision Model and Notation (DMN) to standardize and implement collaborative business processes and decisions on permissioned blockchain platforms. Our approach automatically translates integrated BPMN-DMN models into smart contracts(SCs) compatible with Hyperledger Fabric, enabling privacy-aware multi-organizational process execution through blockchain-based Attribute-Based Access Control (ABAC). The framework introduces three key innovations: (1) a standardized method for modeling collaborative processes and decisions using integrated BPMN-DMN model, (2) an automated SC generator that preserves both process logic and decision rules while maintaining privacy constraints, and (3) a hybrid on-chain/off-chain execution environment that optimizes collaborative workflows through secure data transfer and external system integration. Experimental evaluation across 11 real-world collaboration scenarios demonstrates that our approach achieves 100\% accuracy in process execution. Furthermore, an analysis of various execution processes highlights the strong practical applicability and reliability of our approach. The proposed framework includes an open-source third-party collaboration platform based on blockchain.

cs.SE↗