Author name cluster

Yunpu Ma

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

12 papers

2 author rows

AAAI Conference 2026 Conference Paper

ASCD: Attention-Steerable Contrastive Decoding for Reducing Hallucination in MLLM

Yujun Wang
Aniri
Jinhe Bi
Soren Pirk
Yunpu Ma

Multimodal large language models (MLLMs) frequently hallucinate by over-committing to spurious visual cues. Prior remedies–Visual and Instruction Contrastive Decoding (VCD, ICD)–mitigate this issue, yet the mechanism remains opaque. We first empirically show that their improvements systematically coincide with redistributions of cross-modal attention. Building on this insight, we propose Attention-Steerable Contrastive Decoding (ASCD), which directly steers the attention scores during decoding. ASCD combines (i) positive steering, which amplifies automatically mined text-centric heads–stable within a model and robust across domains–with (ii) negative steering, which dampens on-the-fly identified critical visual tokens. The method incurs negligible runtime/memory overhead and requires no additional training. Across five MLLM backbones and three decoding schemes, ASCD reduces hallucination on POPE, CHAIR, and MMHal-Bench by up to 38.2% while improving accuracy on standard VQA benchmarks, including MMMU, MM-VET, ScienceQA, TextVQA, and GQA. These results position attention steering as a simple, model-agnostic, and principled route to safer, more faithful multimodal generation.

PDF Details DOI

AAAI Conference 2026 Conference Paper

DAWN: Distributed LLM Multi-Agent Workflow Synthesis

Guancheng Wan
Mo Zhou
Ziyi Wang
Xiaoran Shang
Eric Hanchen Jiang
Guibin Zhang
Jinhe Bi
Yunpu Ma

Large language models (LLMs) have recently empowered multi-agent systems (MAS) to achieve remarkable advances in collaborative reasoning and complex task automation. The effectiveness of these systems fundamentally depends on the design of adaptive communication graphs—the underlying workflows that coordinate agent interactions. However, in real-world scenarios, strict privacy constraints often silo data across organizations, and client distributions are highly non-IID, posing major challenges for synthesizing such workflows. In this work, we are the first to systematically study distributed multi-agent workflow synthesis under these privacy and heterogeneity constraints, and we introduce the Difficulty-Based Skew (DBS) benchmark to emulate such challenging environments. Drawing inspiration from federated graph learning (FGL)—which has primarily focused on classification over static graphs—we identify a critical gap: existing FGL methods do not address the generative design of communication topologies. We reveal two fundamental obstacles to generative workflow synthesis in this setting: (i) workflow specialization conflict, where agents optimized for different task distributions generate incompatible communication patterns that resist meaningful aggregation, and (ii) structural communication shift, where locally optimal agent interaction graphs fail to compose into globally coherent multi-agent workflows. To address these challenges, we propose DAWN, a federated framework that integrates two key innovations: Parametric Resonance, which robustly aggregates heterogeneous local updates via layer-wise SVD-based denoising and alignment, and Structural Gravity, which regularizes local workflow generation by penalizing the Fusion Gromov-Wasserstein distance to a set of prototype communication graphs, ensuring global structural coherence without stifling local adaptation. Experiments on the DBS benchmark show that DAWN surpasses baselines in global task success and reduces inter-client graph divergence, laying a solid foundation for privacy-preserving, adaptive MAS workflow design in heterogeneous settings.

PDF Details DOI

AAAI Conference 2026 Conference Paper

OpenDriveVLA: Towards End-to-end Autonomous Driving with Large Vision Language Action Model

Xingcheng Zhou
Xuyuan Han
Feng Yang
Yunpu Ma
Volker Tresp
Alois Knoll

We present OpenDriveVLA, a Vision-Language Action (VLA) model designed for end-to-end autonomous driving, built upon open-source large language models. OpenDriveVLA generates spatially-grounded driving actions by leveraging multimodal inputs, including both 2D and 3D instance-aware visual representations, ego vehicle states, and language commands. To bridge the modality gap between driving visual representations and language embeddings, we introduce a hierarchical vision-language alignment process, projecting both 2D and 3D structured visual tokens into a unified semantic space. Furthermore, we incorporate structured agent–environment–ego interaction modeling into the autoregressive decoding process, enabling the model to capture fine-grained spatial dependencies and behavior-aware dynamics critical for reliable trajectory planning. Extensive experiments on the nuScenes dataset demonstrate that OpenDriveVLA achieves state-of-the-art results across open-loop trajectory planning and driving-related question-answering tasks. Qualitative analyses further illustrate its superior capability to follow high-level driving commands and robustly generate trajectories under challenging scenarios, highlighting its potential for next-generation end-to-end autonomous driving.

PDF Details DOI

TMLR Journal 2025 Journal Article

DyGMamba: Efficiently Modeling Long-Term Temporal Dependency on Continuous-Time Dynamic Graphs with State Space Models

Zifeng Ding
Yifeng Li
Yuan He
Antonio Norelli
Jingcheng Wu
Volker Tresp
Michael M. Bronstein
Yunpu Ma

Learning useful representations for continuous-time dynamic graphs (CTDGs) is challenging, due to the concurrent need to span long node interaction histories and grasp nuanced temporal details. In particular, two problems emerge: (1) Encoding longer histories requires more computational resources, making it crucial for CTDG models to maintain low computational complexity to ensure efficiency; (2) Meanwhile, more powerful models are needed to identify and select the most critical temporal information within the extended context provided by longer histories. To address these problems, we propose a CTDG representation learning model named DyGMamba, originating from the popular Mamba state space model (SSM). DyGMamba first leverages a node-level SSM to encode the sequence of historical node interactions. Another time-level SSM is then employed to exploit the temporal patterns hidden in the historical graph, where its output is used to dynamically select the critical information from the interaction history. We validate DyGMamba experimentally on the dynamic link prediction task. The results show that our model achieves state-of-the-art in most cases. DyGMamba also maintains high efficiency in terms of computational resources, making it possible to capture long temporal dependencies with a limited computation budget.

PDF Details

AAAI Conference 2025 Conference Paper

WebPilot: A Versatile and Autonomous Multi-Agent System for Web Task Execution with Strategic Exploration

Yao Zhang
Zijian Ma
Yunpu Ma
Zhen Han
Yu Wu
Volker Tresp

LLM-based autonomous agents often fail to execute complex web tasks that require dynamic interaction, largely due to the inherent uncertainty and complexity of these environments. Existing LLM-based web agents typically rely on rigid, expert-designed policies specific to certain states and actions, lacking the flexibility and generalizability needed to adapt to unseen tasks. In contrast, humans excel by exploring unknowns, continuously adapting strategies based on new observations, and resolving ambiguities through exploration. To emulate human-like adaptability, web agents need strategic exploration and complex decision-making. Monte Carlo Tree Search (MCTS) is well-suited for this, but classical MCTS struggles with vast action spaces, unpredictable state transitions, and incomplete information in web tasks. In light of this, we develop WebPilot, a multi-agent system with a dual optimization strategy that improves MCTS to better handle complex web environments. Specifically, the Global Optimization phase involves generating a high-level plan by breaking down tasks into manageable subtasks, continuously refining this plan through reflective analysis of new observations and previous subtask attempts, thereby focusing the search process and mitigating challenges posed by vast action spaces in classical MCTS. Subsequently, the Local Optimization phase executes each subtask using a tailored MCTS designed for complex environments, effectively addressing uncertainties and managing incomplete information by iteratively refining decisions based on new observations. Experimental results on WebArena and MiniWoB++ demonstrate the effectiveness of WebPilot. Notably, on WebArena, WebPilot achieves SOTA performance with GPT-4, achieving a 93% relative increase in success rate over the concurrent tree search-based method. WebPilot advances autonomous agents, enabling more reliable decision-making in practical environments.

PDF Details DOI

NeurIPS Conference 2023 Conference Paper

Benchmarking Robustness of Adaptation Methods on Pre-trained Vision-Language Models

Shuo Chen
Jindong Gu
Zhen Han
Yunpu Ma
Philip Torr
Volker Tresp

Various adaptation methods, such as LoRA, prompts, and adapters, have been proposed to enhance the performance of pre-trained vision-language models in specific domains. As test samples in real-world applications usually differ from adaptation data, the robustness of these adaptation methods against distribution shifts are essential. In this study, we assess the robustness of 11 widely-used adaptation methods across 4 vision-language datasets under multimodal corruptions. Concretely, we introduce 7 benchmark datasets, including 96 visual and 87 textual corruptions, to investigate the robustness of different adaptation methods, the impact of available adaptation examples, and the influence of trainable parameter size during adaptation. Our analysis reveals that: 1) Adaptation methods are more sensitive to text corruptions than visual corruptions. 2) Full fine-tuning does not consistently provide the highest robustness; instead, adapters can achieve better robustness with comparable clean performance. 3) Contrary to expectations, our findings indicate that increasing the number of adaptation data and parameters does not guarantee enhanced robustness; instead, it results in even lower robustness. We hope this study could benefit future research in the development of robust multimodal adaptation methods. The benchmark, code, and dataset used in this study can be accessed at https: //adarobustness. github. io.

PDF Details

AAAI Conference 2022 Conference Paper

TLogic: Temporal Logical Rules for Explainable Link Forecasting on Temporal Knowledge Graphs

Yushan Liu
Yunpu Ma
Marcel Hildebrandt
Mitchell Joblin
Volker Tresp

Conventional static knowledge graphs model entities in relational data as nodes, connected by edges of specific relation types. However, information and knowledge evolve continuously, and temporal dynamics emerge, which are expected to influence future situations. In temporal knowledge graphs, time information is integrated into the graph by equipping each edge with a timestamp or a time range. Embeddingbased methods have been introduced for link prediction on temporal knowledge graphs, but they mostly lack explainability and comprehensible reasoning chains. Particularly, they are usually not designed to deal with link forecasting – event prediction involving future timestamps. We address the task of link forecasting on temporal knowledge graphs and introduce TLogic, an explainable framework that is based on temporal logical rules extracted via temporal random walks. We compare TLogic with state-of-the-art baselines on three benchmark datasets and show better overall performance while our method also provides explanations that preserve time consistency. Furthermore, in contrast to most state-ofthe-art embedding-based methods, TLogic works well in the inductive setting where already learned rules are transferred to related datasets with a common vocabulary.

PDF Details

ICLR Conference 2021 Conference Paper

Explainable Subgraph Reasoning for Forecasting on Temporal Knowledge Graphs

Zhen Han 0003
Peng Chen
Yunpu Ma
Volker Tresp

Modeling time-evolving knowledge graphs (KGs) has recently gained increasing interest. Here, graph representation learning has become the dominant paradigm for link prediction on temporal KGs. However, the embedding-based approaches largely operate in a black-box fashion, lacking the ability to interpret their predictions. This paper provides a link forecasting framework that reasons over query-relevant subgraphs of temporal KGs and jointly models the structural dependencies and the temporal dynamics. Especially, we propose a temporal relational attention mechanism and a novel reverse representation update scheme to guide the extraction of an enclosing subgraph around the query. The subgraph is expanded by an iterative sampling of temporal neighbors and by attention propagation. Our approach provides human-understandable evidence explaining the forecast. We evaluate our model on four benchmark temporal knowledge graphs for the link forecasting task. While being more explainable, our model obtains a relative improvement of up to 20 $\%$ on Hits@1 compared to the previous best temporal KG forecasting method. We also conduct a survey with 53 respondents, and the results show that the evidence extracted by the model for link forecasting is aligned with human understanding.

Details

AAAI Conference 2020 Conference Paper

Reasoning on Knowledge Graphs with Debate Dynamics

Marcel Hildebrandt
Jorge Andres Quintero Serna
Yunpu Ma
Martin Ringsquandl
Mitchell Joblin
Volker Tresp

We propose a novel method for automatic reasoning on knowledge graphs based on debate dynamics. The main idea is to frame the task of triple classiﬁcation as a debate game between two reinforcement learning agents which extract arguments – paths in the knowledge graph – with the goal to promote the fact being true (thesis) or the fact being false (antithesis), respectively. Based on these arguments, a binary classiﬁer, called the judge, decides whether the fact is true or false. The two agents can be considered as sparse, adversarial feature generators that present interpretable evidence for either the thesis or the antithesis. In contrast to other black-box methods, the arguments allow users to get an understanding of the decision of the judge. Since the focus of this work is to create an explainable method that maintains a competitive predictive accuracy, we benchmark our method on the triple classiﬁcation and link prediction task. Thereby, we ﬁnd that our method outperforms several baselines on the benchmark datasets FB15k-237, WN18RR, and Hetionet. We also conduct a survey and ﬁnd that the extracted arguments are informative for users.

PDF Details

UAI Conference 2018 Conference Paper

Holistic Representations for Memorization and Inference

Yunpu Ma
Marcel Hildebrandt
Volker Tresp
Stephan Baier

In this paper we introduce a novel holographic memory model for the distributed storage of complex association patterns and apply it to knowledge graphs. In a knowledge graph, a labelled link connects a subject node with an object node, jointly forming a subject-predicateobjects triple. In the presented work, nodes and links have initial random representations, plus holistic representations derived from the initial representations of nodes and links in their local neighbourhoods. A memory trace is represented in the same vector space as the holistic representations themselves. To reduce the interference between stored information, it is required that the initial random vectors should be pairwise quasi-orthogonal. We show that pairwise quasi-orthogonality can be improved by drawing vectors from heavy-tailed distributions, e. g. , a Cauchy distribution, and, thus, memory capacity of holistic representations can significantly be improved. Furthermore, we show that, in combination with a simple neural network, the presented holistic representation approach is superior to other methods for link predictions on knowledge graphs.

Details

IJCAI Conference 2018 Conference Paper

Improving Information Extraction from Images with Learned Semantic Models

Stephan Baier
Yunpu Ma
Volker Tresp

Many applications require an understanding of an image that goes beyond the simple detection and classification of its objects. In particular, a great deal of semantic information is carried in the relationships between objects. We have previously shown, that the combination of a visual model and a statistical semantic prior model can improve on the task of mapping images to their associated scene description. In this paper, we review the model and compare it to a novel conditional multi-way model for visual relationship detection, which does not include an explicitly trained visual prior model. We also discuss potential relationships between the proposed methods and memory models of the human brain.

PDF Details

NeSy Conference 2017 Conference Paper

Learning with Knowledge Graphs

Volker Tresp
Yunpu Ma
Stephan Baier

In recent years a number of large-scale triple-oriented knowledge graphs have been generated. They are being used in research and in applications to support search, text understanding and question answering. Knowledge graphs pose new challenges for machine learning, and research groups have developed novel statistical models that can be used to compress knowledge graphs, to derive implicit facts, and to detect errors in the knowledge graph. In this paper we decribe the concept of triple-oriented knowledge graphs and corresponding learning approaches. We also discuss episodic knowledge graphs which are able to represent temporal data; learning with episodic data can be the basis for decision support systems, e.g. in a clinical context. Finally we discuss how knowledge graphs can support perception, by mapping subsymbolic sensory inputs, such as images, to semantic triples. A particular feature of our approach would be that perception, episodic memory and semantic memory are highly interconnected and that, in a cognitive interpretation, all rely on the same brain structures.

Details