paperlist.ai

AI conference papers

AI conference paper library

Search papers from 30+ top conferences, including NeurIPS, ICLR, ICML, CVPR, ACL and AAAI. Filter by keyword, then open any paper for its abstract, authors and links.

IJCAI 2026

PersuHMM: Iterative Learning the Hierarchical Meta-Strategy Memory for Persuasive Dialogue

Yanyue Zhang, Zihao Wang, Xin Zhang, Deyu Zhou

Persuasive dialogue aims to alter people's attitudes or behaviors through conversation. While LLMs can generate emotional responses, they tend to use a uniform approach across varied scenarios, limiting their persuasive effectiveness. Diverse demands need varied persuasion strategies, which challenge both LLMs and human efforts. To address this, this paper incorporates the idea of meta-learning and proposes PersuHMM, an iterative learning approach for constructing the Hierarchical Meta-strategy Memory to generate Persuasive dialogue. Through the interaction between Meta‑layer and Task‑layer, the approach continuously accumulates automated meta‑strategies from generated outcomes and data labels, forming a persuasive meta‑strategy memory that balances efficiency and generalizability. To facilitate model retrieval and use, the meta-strategy memory is hierarchically organized via clustering for a clearer and more logical storage structure. Experimental results show that the hierarchical meta-strategy memory enhances both the persuasiveness and generalizability of the persuasion model, and even enables cross-model performance improvement.

IJCAI 2026

Exploring the System 1 Thinking Capability of Large Reasoning Models

Wenyuan Zhang, Shuaiyi Nie, Xinghua Zhang, Zefeng Zhang, Tingwen Liu

This paper explores the system 1 thinking capability of Large Reasoning Models (LRMs), the intuitive ability to respond efficiently with minimal token usage. While existing LRMs rely on long-chain reasoning and excel at complex tasks, their system 1 thinking ability remains largely underexplored. This capability is essential as it reflects models' difficulty awareness and reasoning efficiency, both critical for real-world applications. We propose S1-Bench, a multi-domain, multilingual benchmark comprising model-simple system 1 questions. Our investigation of 28 LRMs reveals under-accuracy and inefficiency on system 1 problems. We find existing efficient reasoning methods either generalize poorly to simple questions or sacrifice performance for efficiency. Further exploration uncovers LRMs' early difficulty awareness accompanied by lower confidence, and shows that problem difficulty is implicitly encoded in hidden states. Code and the extended version of the paper with appendices are available at https://github.com/WYRipple/S1_Bench.

IJCAI 2026

ScholarSum: Student-Teacher Abstractive Summarization via Knowledge Graph Reasoning and Reflective Refinement

Bohou Zhang, Xiaoyu Tao, Mingyue Cheng, Huijie Liu, Qi Liu

Abstractive summarization plays a crucial role in enabling efficient understanding of scientific literature, yet it inherently demands both linguistic fluency and factual faithfulness. Existing approaches often fail to reconcile these two requirements. Extractive methods rely on rigid sentence splicing that disrupts macro-level logical coherence, while large language model (LLM)-based generative approaches, despite mastering linguistic fluency, exhibit limited factual consistency. In this work, we propose ScholarSum, a hierarchical reflective graph-based framework that emulates a student–teacher writing process for fluent and faithful scientific summarization. ScholarSum first organizes the document into a hierarchical knowledge graph by segmenting it into semantically coherent units, whose multi-layered community structure captures global logic and macro-level themes. Guided by this global structure, the student generates an initial draft, which is subsequently refined through fine-grained evidence retrieval. To ensure factual consistency, a teacher-like reviewer then iteratively examines the draft, identifies unsupported content, and prompts targeted re-retrieval and rewriting until the summary meets rigorous quality standards. Extensive experiments demonstrate that ScholarSum significantly outperforms previous baselines in terms of both completeness and faithfulness. Our code is available at https://github.com/Xiaoyu-Tao/ScholarSum.

IJCAI 2026

Debate with Myself: Zero-Shot Event Causality Identification with Adversarial Evidence Integration via Large Language Models

Zefan Zeng, Yuehang Si, Xingchen Hu, Qing Cheng, Jiakun Liu, Zhong Liu

Event Causality Identification (ECI) is a crucial task in knowledge discovery that extracts structured causal relationships between annotated event mentions from unstructured text. However, existing approaches typically rely on extensive labeled data, which is scarce for specialized domains and topics. Although Large Language Models (LLMs) show strong promise for few-shot and zero-shot information extraction, they are prone to “causal hallucination,” generating unreliable and spurious causal links. To address these limitations, we propose LLM-SD (Large Language Model Self-Debate), a novel framework that formulates ECI as a structured debate among multiple identical instances of a single LLM. Causality is determined through the integration of adversarial evidence. The framework employs LLMs in distinct roles: an affirmative team argues for the existence of causality, a negative team argues against it, and an adjudication committee evaluates the evidence for determination. An evidence strength grading rule guides the quantification and integration of adversarial evidence. The automatic and LLM-driven verification finally produce a reasoned verdict for event causality. LLM-SD reduces spurious causal links resulting from causal hallucination in LLMs and identifies more long-distance causalities by promoting a balanced evaluation of arguments. Extensive experiments on three benchmark datasets demonstrate that LLM-SD achieves state-of-the-art performance in a zero-shot setting.

IJCAI 2026

Learning to Route LLMs from Implicit Cost-Performance Preferences via Meta-Learning

Jiahao Zeng, Ming Tang, Ningning Ding

Large language models (LLMs) present a trade-off between performance and cost, where more powerful models incur greater expense. LLM routing aims to mitigate expenses while maintaining performance by sending queries to the most suitable model. However, existing methods cannot perform well for different user cost-performance preferences. To address this gap, we introduce a novel perceptive LLM routing paradigm for personalized and user-centric cost-performance optimization, which efficiently learns users' implicit preferences through little interaction. To handle the challenge of heterogeneous user needs, we formulate preference profiles as a set of distinct tasks in contextual bandit and propose MetaRouter, a meta-learning framework designed for preference-aware LLM routing. Experimental results show that MetaRouter outperforms strong baselines on both in-distribution and out-of-distribution tasks. Furthermore, it exhibits high efficiency in learning user preferences, robustness to changes in the routable LLMs, and scalability to multi-model routing.

IJCAI 2026

SRJudge: Empowering Large Language Models with Selective Reasoning for Fine-Grained Knowledge Concept Tagging

Zhiwei Yang, Jiahua Yang, Huiru Lin, Xing Chen, Quanlong Guan

Knowledge concept tagging aims to assign specific concept or topic labels to educational content, which is essential for both educators and learners in traditional and online teaching practices. Recent work has explored large language models (LLMs) for this task, achieving promising performance. However, LLMs still struggle to select the correct concept from a large-scale candidate set due to the high dimensionality of the decision space. In this paper, we propose a novel three-stage Select-Reason-Judge (SRJudge) framework, which empowers LLMs with selective reasoning capability for fine-grained knowledge concept tagging. Specifically, the Selector in Stage 1 first narrows the candidate concepts to a top-K shortlist by fine-tuning a small language model (SLM), e.g., BERT, since the top-K predictions hit the correct concept in most cases, thereby reducing the decision space of correct candidates. Next, the Stage 2 Reasoner employs a lightweight LLM for refined reasoning over the shortlisted candidates. It further integrates an improved reinforcement learning strategy with a dynamic task-specific reward function and a pruning mechanism to better align with human reasoning preferences. Finally, a larger LLM acts as a judger that evaluates the overall rationality of the reasoning process and its explanations to determine the final output. In addition, we construct two high-quality datasets for further validation, i.e., the biology dataset S_Bio and the physics dataset S_Phy. Experimental results demonstrate that our method consistently outperforms state-of-the-art baselines across benchmark datasets, verifying its effectiveness and superiority. Resources are available at: https://github.com/Nicozwy/SRJudge.

IJCAI 2026

SeMi-LoRA: Enhancing Low-Rank Adaptation via Separation and Mixing

Zhenfei Yang, Beiming Yu, Peiqin Lin, Yongkang Liu, Deyi Xiong

Low-Rank Adaptation (LoRA) has become a standard paradigm for parameter-efficient fine-tuning (PEFT). However, its low-rank constraint can limit its capacity to express complex weight updates, leaving a performance gap compared with full fine-tuning. Existing extensions improve expressivity, but they often sacrifice parameter mergeability or rely on sparse, block-diagonal updates that restrict global information flow. We propose SeMi-LoRA (Separation and Mixing LoRA), a novel framework that enables complete high-rank updates while preserving mergeability. SeMi-LoRA decomposes adaptation into channel-wise compression followed by two-stage latent mixing, namely Rank Mix and Channel Mix. Our structural analysis shows that, under a fixed base rank, the rank capacity of the update increases with the number of channels and supports complete updates rather than partial sparse ones. Extensive experiments on commonsense reasoning, mathematical reasoning, natural language understanding, and dialogue generation benchmarks demonstrate that SeMi-LoRA consistently outperforms strong PEFT baselines with favorable parameter efficiency.

IJCAI 2026

Mitigating Tool Overuse for LLMs via Active Knowledge Boundary Probing

Zhaoyu Yang, Wenjun Ke, Yuanyao Li, Yuchen Liu, Junqi Xu, Peng Wang, Hengyuan Xu

Tool-augmented methods aim to enhance the reasoning capabilities of large language models (LLMs) by invoking external tools, which can be broadly categorized into training-free and training-based methods. Training-free methods can directly instruct LLMs to invoke external tools, but they exhibit limited generalization in dynamic environments. In contrast, training-based methods use a teacher model to generate tool-use trajectories and fine-tune a student model to distill the trajectories that can generalize across tasks. However, they still face two major challenges: (1) Knowledge boundary mismatch: teacher trajectories may contain steps that the student cannot execute without tools due to their mismatch of knowledge boundaries. (2) Tool overuse propagation: the student inherits the misaligned tool-use strategy from teacher model. To address these challenges, we propose RADAR, an automated framework for knowledge boundary discovery and tool overuse mitigation. First, RADAR performs semantic anchoring to select representative seeds. Second, simulated annealing probing is employed to actively explore the student model’s tool-use boundary and acquire sparse, verified student-aware labels. Third, it globally propagates these labels and rewrites conflicts, yielding deployment-aligned decisions with fewer unnecessary tool calls. We conduct experiments on multiple benchmarks, compared to SOTA method, RADAR reduces average tool-use by 52.7% while improving accuracy by 3.2% across three models.

IJCAI 2026

ChemKGL: Bridging Knowledge Graphs and Large Language Models for Chemical Multi-Step Reaction Pathway Inference

Fan Yang, Feiyang Xu, Kun Zhang, Huadong Liang, Pengyang Shao, Xin Li, Le Wu

Large language models have shown promising potential in chemistry, with prior work exploring molecular recognition, classification, and property prediction. Despite the achieved progress, LLMs are still far from satisfactory when dealing with complex chemical multi-step reaction pathway inference task due to the lack of domain knowledge and limited ability to maintain consistent multi-step reasoning. To address these challenges, we propose ChemKGL, a novel multi-step reasoning framework enhanced by knowledge graph retrieval. We reformulate the task as a knowledge graph retrieval problem to better control LLM generation, and leverage graph technologies to integrate textual descriptions, reaction conditions, and materials. Based on this reformulation, we design a novel multi-retrieval strategy to improve the performance of reaction pathway inference, which includes forward retrieval, reverse retrieval, and historical retrieval. An entropy-based pruning mechanism is further introduced to alleviate cold-start issues caused by retrieval failure. In addition, we construct a comprehensive chemical knowledge graph and corresponding dataset to support model training and analysis. Finally, we employ multiple evaluation methods to assess our proposed ChemKGL, and the experimental results demonstrate its superiority. Our code and dataset can be obtained at https://github.com/Double-Sail/ChemKGL.

IJCAI 2026

System 1&2 Synergy via Dynamic Model Interpolation

Chenxu Yang, Qingyi Si, Chong Tian, Xiyu Liu, Dingyu Yao, Chuanyu Qin, Zheng Lin, Weiping Wang, Jiaqi Wang

Training a unified language model that adapts between intuitive System 1 and deliberative System 2 remains challenging due to interference between their cognitive modes. Recent studies have thus pursued making System 2 models more efficient. However, these approaches focused on output control, limiting what models produce. We argue that this paradigm is misaligned: output length is merely a symptom of the model's cognitive configuration, not the root cause. In this work, we shift the focus to capability control, which modulates how models think rather than what they produce. To realize this, we leverage existing Instruct and Thinking checkpoints through dynamic parameter interpolation, without additional training. Our pilot study establishes that linear interpolation yields a convex, monotonic Pareto frontier, underpinned by representation continuity and structural connectivity. Building on this, we propose DAMI (DynAmic Model Interpolation), a framework that estimates a query-specific Reasoning Intensity lambda(q) to configure cognitive depth. For training-based estimation, we develop a preference learning method encoding accuracy and efficiency criteria. For zero-shot deployment, we introduce a confidence-based method leveraging inter-model cognitive discrepancy. Experiments on five mathematical reasoning benchmarks demonstrate that DAMI achieves higher accuracy than the Thinking model while remaining efficient, effectively combining the efficiency of System 1 with the reasoning depth of System 2.

IJCAI 2026

Profiling the Voice: Speaker-Specific Phoneme Fingerprinting for Speech Deepfake Detection

Jun Xue, Tong Zhang, Zhuolin Yi, Yihuan Huang, Yi Chai, Yiyang Zhang, Yanzhen Ren

The rapid advancement of generative AI has made audio deepfakes increasingly indistinguishable from authentic human vocals, posing significant threats to persons-of-interest (POI) such as public figures. Current detection systems primarily rely on generic, black-box models that fail to capture speaker-specific idiosyncratic traits and lack interpretability. In this paper, we propose Phoneme-based Voice Profiling (PVP), a novel personalized defense framework. By shifting the detection paradigm from macro-utterance analysis to micro-phonetic modeling, PVP captures the unique acoustic distributions underlying a POI’s habitual articulatory patterns. Specifically, our framework models speaker-specific phonetic realizations using lightweight Gaussian Mixture Models (GMMs) estimated solely from bona fide reference speech. This design enables data-efficient profiling and robust generalization to previously unseen spoofing attacks without requiring heavy spoof-specific training. Furthermore, we introduce the first large-scale Chinese POI deepfake dataset to benchmark speaker-specific detection. Experimental results demonstrate that PVP significantly outperforms state-of-the-art generic detectors in POI spoofing scenarios, achieving substantial EER reductions while providing fine-grained, phoneme-level interpretability for forensic analysis. Code and data are available at: https://github.com/JunXue-tech/PVP

IJCAI 2026

MathCritique: Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision

Zhiheng Xi, Dingwen Yang, Jixuan Huang, Jiafu Tang, Xin Guo, Guanyu Li, Yiwen Ding, Wei He, Boyang Hong, Shihan Dou, WenYu Zhan, Xiao Wang, Xiaowei Shi, Yitao Zhai, Rongxiang Weng, Jingang Wang, Rui Zheng, Tao Ji, Tao Gui, Zuxuan Wu, Qi Zhang, Xipeng Qiu, Xuanjing Huang, Yu-Gang Jiang

Training critique models to provide useful feedback for actor models is an effective approach in scalable oversight, especially for complex tasks like math reasoning. However, current research lacks suitable datasets for effectively training critique models and integrating them in a principled way at both test time and training time. To bridge this gap, we first propose AutoMathCritique, an automated and scalable framework for collecting critique data. Using it, we create MathCritique-76k, a dataset of $76,321$ instances paired with step-level feedback, which is then used to train critique models for generating natural-language feedback. We show that critique models consistently improve the actor’s performance—particularly on difficult queries at test time—with larger gains as inference compute scales. Building on the findings, we propose a critique-in-the-loop self-improvement method that incorporates critique-based supervision into the actor’s self-training process. Extensive experiments demonstrate that this method improves the actor’s exploration efficiency and solution diversity, especially on challenging queries, leading to a stronger actor model. Our code and datasets are at https://mathcritique.github.io/.

IJCAI 2026

Improving Code Translation with Syntax-Guided and Semantic-aware Preference Optimization

Yuhan Wu, Huan Zhang, Wei Cheng, Chen Shen, Jingyue Yang, Wei Hu

LLMs have shown immense potential for code translation, yet they often struggle to ensure both syntactic correctness and semantic consistency. While preference-based learning offers a promising alignment strategy, it is hindered by unreliable semantic rewards derived from sparse test cases or restrictive reference translations. We argue that a robust semantic reward for code translation must be derived directly from the source code. In this paper, we propose CTO to improve code translation with syntax-guided and semantic-aware preference optimization. Through contrastive learning, we train a cross-lingual semantic model to directly assess functional equivalence between source and translated code. By formulating code translation as a multi-objective optimization problem, this robust semantic signal is seamlessly unified with compiler-based syntactic feedback within the direct preference optimization framework. Extensive experiments on C++, Java, and Python translations demonstrate that CTO significantly outperforms existing baselines and alternative preference optimization strategies.

IJCAI 2026

ARMOR: Adaptive Curriculum Meta-Learning for Noise-Robust RAG Reasoning

Yan Wang, Yuxin Zhang, Shenyu Zhang, Yongrui Chen, Sheng Bi, Guilin Qi

Retrieval-Augmented Generation (RAG) systems have demonstrated remarkable effectiveness in mitigating hallucinations by incorporating external knowledge. However, the retrieval process inevitably introduces noise, posing significant challenges to RAG robustness. Fundamentally, noise robustness is a composite competency encompassing noise identification, retrieval filtering, and robust reasoning. Unfortunately, existing methods often fail to capture this holistic nature, typically restricted to single-dimensional enhancements or hampered by noise-induced optimization interference. To address these limitations, we decompose RAG robustness into three sub-tasks: knowledge boundary perception, retrieval verification, and robust reasoning. Accordingly, we propose ARMOR, an Adaptive Curriculum Meta-Learning for Noise-Robust RAG Reasoning. Inspired by the teaching process, ARMOR introduces an adaptive task scheduler as a "Teacher" to curate training plans according to the absolute learning progress of the model ("student"). Distinct from GRPO, ARMOR utilizes inter-group advantage estimation during learning to effectively mitigate mode collapse caused by limited output spaces. Experiments on three noise-injected complex reasoning datasets (2WikiMultiHopQA, MusiQue, and HotPotQA) indicate that ARMOR achieves synergistic improvements in retrieval noise robustness, significantly outperforming baselines in output quality.

IJCAI 2026

Performance-Driven Demonstration Selection for In-Context Learning

Wenqiang Wang, Mingbo Yang, Aiping Zhang, Yan Xiao, Peng Chen, Jianjie Huang, Xiaochun Cao

In-context learning (ICL) enables large language models (LLMs) to adapt to new tasks with considerable performance gains, yet its effectiveness is highly sensitive to the choice of demonstrations. Most existing selection methods rely on heuristic or proxy signals, such as similarity, diversity, or uncertainty, and select demonstrations independently, which may misalign with downstream performance and overlook set-level composition effects. Therefore, we propose Performance-Driven Demonstration Selection (PDDS), which directly aligns demonstration selection with ICL performance. PDDS formulates selection as predicting the target LLM’s downstream task performance for a given query–in-context pair, replacing proxy heuristics with a performance-aware objective. Unlike prior approaches that rank demonstrations independently before composing the prompt, PDDS evaluates the entire in-context set, capturing inter-demonstration interactions and composition effects. PDDS trains an end-to-end scorer using supervision from the target LLM’s actual task outcomes. At inference time, it selects high-scoring in-context sets without additional target LLM calls or validation-time feedback. Across 6 NLP tasks, 8 datasets, and 8 target LLMs ranging from 1B to 70B parameters, PDDS achieves state-of-the-art results and generalizes well across LLMs and datasets. PDDS remains effective with as few as 30 training instances and can be integrated as a plug-and-play component to enhance existing ICL methods.

IJCAI 2026

PDAR-RSITR: A Progressive Decoupling-Aggregation-Refinement Framework for Remote Sensing Image-Text Retrieval

Shuhuai Wang, Songwei Pei, Bingfeng Liu, Duo Chai, Yuanzhou Huang, Jia Liu, Qian Li, Shangguang Wang

Remote Sensing Image-Text Retrieval (RSITR) aims to achieve precise retrieval between remote sensing images and textual descriptions. However, existing methods neglect the multi-dimensional cognitive attributes inherent in remote sensing data and struggle to handle them simultaneously, leading to suboptimal retrieval performance. In this paper, we propose a novel Progressive Decoupling-Aggregation-Refinement framework for RSITR (PDAR-RSITR) to comprehensively capture multi-dimensional cognitive attributes. Specifically, we adapt Multi-dimensional Cognitive Decoupling to learn features across cognitive attributes of different dimensions via multi-process clustering. Subsequently, we utilize Salient Cognitive Aggregation to select and aggregate the most salient attributes via dynamic routing. Furthermore, we propose Expert Collaborative Refinement to enhance critical cross-modal relationships via three complementary expert perspectives. Extensive experimental results demonstrate that PDAR-RSITR significantly outperforms existing state-of-the-art methods across multiple metrics.

IJCAI 2026

Mitigating Entity Type Confusion in Cross-Domain NER via Multidimensional Quantification and Reasoning Enhancement

Jingyu Wang, Shijie Wu, Fusheng Jin

Cross-domain Named Entity Recognition (CD-NER) aims to transfer the rich knowledge in the source domain to the target domain. Recent studies adopting decomposition or generation paradigms have achieved significant performance improvements, demonstrating high accuracy in entity span detection. However, during entity type classification, models severely suffer from entity type confusion, the erroneous tendency that models classify entities of one type in the text as another similar but incorrect type. To address this issue, we first propose a Multidimensional Confusion Quantification Model (MCQM) that quantifies a model's confusion extent between entity types from three dimensions: source-target hierarchy analysis, semantic similarity analysis, and explicit data evaluation. Moreover, we propose the Progressive Bidirectional Reasoning Chain (PBRC). PBRC leverages the source-target hierarchy and confusion analysis from the MCQM to prompt the LLM to generate two-stage reasoning information. The two-stage reasoning information is utilized to augment the knowledge of the model, significantly mitigating entity type confusion and improving the model's generalization performance. Experimental results demonstrate that our method achieves new state-of-the-art results on all domains of the CrossNER dataset.

IJCAI 2026

Semantic Similarity Is Not Legal Correctness: Evaluating RAG Systems in Brazilian Civil Procedure

Eryclis Silva, Madelyn Sanfilippo

Evaluating retrieval-augmented generation (RAG) systems in legal domains is challenging due to the nuanced nature of legal reasoning and scarcity of domain-specific benchmarks. We investigate semantic similarity and legal correctness in Brazilian legal question answering, developing a synthetic dataset of 3,012 evaluation instances from the Brazilian Civil Procedure Code spanning seven query types, validated through human expert assessment. Our evaluation framework combines BERTScore with domain-adapted LLM-as-Judge (GPT-4o-mini), validated against expert legal assessment. Analysis reveals 50.8% disagreement between semantic similarity and legal correctness, with correlation varying substantially by query type (ρ = 0.464 to ρ = 0.732). Explicit article citation emerges as the strongest quality predictor (Cohen’s d = 1.099), with cited responses achieving 145% higher legal correctness scores. Concrete examples demonstrate bidirectional divergence: responses may achieve high semantic similarity yet cite incorrect articles, or provide legally perfect answers with minimal lexical overlap. Findings demonstrate that semantic metrics alone are insufficient for legal RAG evaluation, with disagreement patterns structured by query type.

IJCAI 2026

BehaviorBench: A Psychologically Grounded Benchmark for Evaluating Personality in Large Language Models Through Realistic Behaviors

Taowen Pu, Hexi Wang, Zeyang Liu, Dongsheng Guo, Chuan Zhao

Current approaches to evaluating personality in large language models (LLMs) typically prompt them to self-report on psychological questionnaires such as the Big Five Inventory. However, these methods assess introspective labels rather than observable behavior, despite the fact that LLMs are deployed to act in realistic contexts, not to reflect on their own traits. To bridge this gap, we introduce BehaviorBench, a new benchmark for evaluating LLM personality through concrete behaviors in everyday scenarios. Grounded in established personality psychology, BehaviorBench links each Big Five trait to validated behavioral manifestations and embeds them in contextually plausible situations that naturally elicit trait-relevant actions. We evaluate a range of models and personality shaping strategies using BehaviorBench and find a substantial mismatch between self-reported personalities and actual behaviors. By grounding evaluation in observable behavior rather than introspection, our work reveals critical gaps in current LLM personality modeling and control mechanisms. Our code and data are available at https://github.com/butra1n/BehaviorBench

IJCAI 2026

DoMoE: Domain-Aware Semantic Expert Prediction for Efficient MoE Inference Under Expert Offloading

Yao Mu, Fahao Chen, Wenbin Zhu, Mengying Zhao, Zhaoyan Shen, Dongxiao Yu

Mixture-of-Experts (MoE) large language models improve inference efficiency through sparse expert activation, but deployment on resource-constrained devices remains challenging due to the large expert parameter footprint. Expert offloading mitigates this issue by loading experts on demand, yet its effectiveness critically depends on accurate and efficient expert prediction: inaccurate predictions incur redundant expert transfers, while overly expensive predictors negate latency benefits. Existing trajectory-based methods suffer from low accuracy, whereas semantic-based approaches incur prohibitive similarity-matching overhead. We propose DoMoE, a domain-aware MoE inference system that exploits domain locality in inference workloads. DoMoE organizes routing information into domain-specific expert routing tables, restricts semantic matching to domain-relevant tokens and explicitly balances prediction accuracy against prediction overhead. Experiments show that DoMoE achieves a 1.32 X average throughput improvement and a 1.22 X increase in expert hit ratio across multiple MoE models and workloads, enabling efficient and accurate expert routing for practical inference.

Keep browsing

More papers from top AI conferences

30+ top conferences, by conference, year, and topic.