Author name cluster

Pengkun Wang

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.

15 papers

1 author row

AAAI Conference 2026 Conference Paper

Learning Dynamics as Feedback: An Adaptive Entropy Flow Dynamics Framework for Long-tailed Human Action Recognition

Yuan Dong
Zhe Zhao
Liheng Yu
Di Wu
Pengkun Wang

Deep human action recognition models trained on real-world data are often challenged by long-tailed distributions, where performance on rare classes is severely degraded. Current solutions typically apply static or heuristic interventions that are disconnected from the model's evolving internal state. To overcome this limitation, we reconceptualize long-tailed human action recognition as a closed-loop, self-regulating system, inspired by ecological theory. We further introduce an Adaptive Ecological Entropy Dynamics (AEED) framework, which is built upon three synergistic components. First, AEED perceives the learning state through entropy flow, providing a robust and directional signal of learning progress. Second, this signal drives an adaptation mechanism, which dynamically adjusts class-specific loss weights to allocate more learning resources to underperforming classes. Finally, AEED facilitates intelligent knowledge transfer via Confidence-Guided Symbiosis (CS-Mix). Extensive experiments demonstrate that AEED achieves state-of-the-art performance on challenging skeleton-based action recognition benchmarks, including NTU-60-LT and Kinetics-400-LT.

PDF Details DOI

AAAI Conference 2026 Conference Paper

Rethinking Crystal Symmetry Prediction: A Decoupled Perspective

Liheng Yu
Zhe Zhao
Xucong Wang
Di Wu
Pengkun Wang

Efficiently and accurately determining the symmetry is a crucial step in the structural analysis of crystalline materials. Existing methods usually mindlessly apply deep learning models while ignoring the underlying chemical rules. More importantly, experiments show that they face a serious sub-property confusion SPC problem. To address the above challenges, from a decoupled perspective, we introduce the XRDecoupler framework, a problem-solving arsenal specifically designed to tackle the SPC problem. Imitating the thinking process of chemists, we innovatively incorporate multidimensional crystal symmetry information as superclass guidance to ensure that the model's prediction process aligns with chemical intuition. We further design a hierarchical PXRD pattern learning model and a multi-objective optimization approach to achieve high-quality representation and balanced optimization. Comprehensive evaluations on three mainstream databases (e.g., CCDC, CoREMOF, and InorganicData) demonstrate that XRDecoupler excels in performance, interpretability, and generalization.

PDF Details DOI

AAAI Conference 2026 Conference Paper

U2B: Scale-unbiased Representation Converter for Graph Classification with Imbalanced and Balanced Scale Distributions

Guanjun Wang
Jianhao Zhang
Jiaming Ma
Sheng Huang
Pengkun Wang
Zhengyang Zhou
Binwu Wang
Yang Wang

Graph classification is a critical task in analyzing graph data, with applications across various domains. While graph neural networks (GNNs) have achieved remarkable results, their ability to generalize across graphs of varying scales remains a challenge. Conventional models often perform well on large-scale graphs but struggle with distributions that are skewed towards small scales. Conversely, models tailored to address scale imbalances frequently prioritize small-scale graphs, leading to diminished performance in more balanced scenarios. To overcome these limitations, we introduce a Unbalanced-Balanced Representation Converter (U2B), which exhibits no explicit bias toward graph scales. U2B employs a two-step workflow: a distillation phase to extract base features from both node-level and graph-level representations, followed by a refinement phase to generate unbiased representations for improved balance. In the distillation phase, a static constraint guides node-level adjustments, improving the representation of nodes in small graphs. Simultaneously, a dynamic constraint in the graph-level process mitigates biases toward features from large graphs. To ensure harmony between the representations, a consistency alignment loss is introduced, aligning node-level and graph-level features to create more cohesive and balanced graph representations. Extensive experiments on multiple datasets show that U2B achieves competitive performance.

PDF Details DOI

IJCAI Conference 2025 Conference Paper

Causal Learning Meet Covariates: Empowering Lightweight and Effective Nationwide Air Quality Forecasting

Jiaming Ma
Zhiqing Cui
Binwu Wang
Pengkun Wang
Zhengyang Zhou
Zhe Zhao
Yang Wang

Air quality prediction plays a crucial role in the development of smart cities, garnering significant attention from both academia and industry. Current air quality prediction models encounter two major limitations: their high computational complexity limits scalability to nationwide datasets, and they often regard weather covariates as optional auxiliary information. In reality, weather covariates can have a substantial impact on air quality indices (AQI), exhibiting a significant causal association. In this paper, we first present a nationwide air quality dataset to address the lack of open-source, large-scale datasets in this field. Then we propose a causal learning model, CauAir, for air quality prediction that harnesses the powerful representation capabilities of the Transformer to explicitly model the causal association between weather covariates and AQI. To address the high complexity of traditional Transformers, we design CachLormer, which features two key innovations: a simplified architecture with redundant components removed, and a cache-attention mechanism that employs learnable embeddings for perceiving causal association between AQI and weather covariates in a coarsegrained perspective. We use information theory to illustrate the superiority of the proposed model. Finally, experimental results on three datasets with 28 as the baseline demonstrate that our model achieves competitive performance, while maintaining high training efficiency and low memory consumption. The source code is available at CauAir Official Repository.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

Deciphering the Extremes: A Novel Approach for Pathological Long-tailed Recognition in Scientific Discovery

Zhe Zhao
Haibin Wen
Xianfu Liu
Rui Mao
Pengkun Wang
Liheng Yu
Linjiang Chen
Bo An

Scientific discovery across diverse fields increasingly grapples with datasets exhibiting pathological long-tailed distributions: a few common phenomena overshadow a multitude of rare yet scientifically critical instances. Unlike standard benchmarks, these scientific datasets often feature extreme imbalance coupled with a modest number of classes and limited overall sample volume, rendering existing long-tailed recognition (LTR) techniques ineffective. Such methods, biased by majority classes or prone to overfitting on scarce tail data, frequently fail to identify the very instances—novel materials, rare disease biomarkers, faint astronomical signals—that drive scientific breakthroughs. This paper introduces a novel, end-to-end framework explicitly designed to address pathological long-tailed recognition in scientific contexts. Our approach synergizes a Balanced Supervised Contrastive Learning (B-SCL) mechanism, which enhances the representation of tail classes by dynamically re-weighting their contributions, with a Smooth Objective Regularization (SOR) strategy that manages the inherent tension between tail-class focus and overall classification performance. We introduce and analyze the real-world ZincFluor chemical dataset ($\mathcal{T}=137. 54$) and synthetic benchmarks with controllable extreme imbalances (CIFAR-LT variants). Extensive evaluations demonstrate our method's superior ability to decipher these extremes. Notably, on ZincFluor, our approach achieves a Tail Top-2 accuracy of $66. 84\%$, significantly outperforming existing techniques. On CIFAR-10-LT with an imbalance ratio of $1000$ ($\mathcal{T}=100$), our method achieves a tail-class accuracy of $38. 99\%$, substantially leading the next best. These results underscore our framework's potential to unlock novel insights from complex, imbalanced scientific datasets, thereby accelerating discovery.

PDF Details

NeurIPS Conference 2025 Conference Paper

Less but More: Linear Adaptive Graph Learning Empowering Spatiotemporal Forecasting

Jiaming Ma
Binwu Wang
Guanjun Wang
Kuo Yang
Zhengyang Zhou
Pengkun Wang
Xu Wang
Yang Wang

The effectiveness of Spatiotemporal Graph Neural Networks (STGNNs) critically hinges on the quality of the underlying graph topology. While end-to-end adaptive graph learning methods have demonstrated promising results in capturing latent spatiotemporal dependencies, they often suffer from high computational complexity and limited expressive capacity. In this paper, we propose MAGE for efficient spatiotemporal forecasting. We first conduct a theoretical analysis demonstrating that the ReLU activation function employed in existing methods amplifies edge-level noise during graph topology learning, thereby compromising the fidelity of the learned graph structures. To enhance model expressiveness, we introduce a sparse yet balanced mixture-of-experts strategy, where each expert perceives the unique underlying graph through kernel-based functions and operates with linear complexity relative to the number of nodes. The sparsity mechanism ensures that each node interacts exclusively with compatible experts, while the balancing mechanism promotes uniform activation across all experts, enabling diverse and adaptive graph representations. Furthermore, we theoretically establish that a single graph convolution using the learned graph in MAGE is mathematically equivalent to multiple convolutional steps under conventional graphs. We evaluate MAGE against advanced baselines on multiple real-world spatiotemporal datasets. MAGE achieves competitive performance while maintaining strong computational efficiency.

PDF Details

NeurIPS Conference 2025 Conference Paper

MoFo: Empowering Long-term Time Series Forecasting with Periodic Pattern Modeling

Jiaming Ma
Binwu Wang
Qihe Huang
Guanjun Wang
Pengkun Wang
Zhengyang Zhou
Yang Wang

The stable periodic patterns present in the time series data serve as the foundation for long-term forecasting. However, existing models suffer from limitations such as continuous and chaotic input partitioning, as well as weak inductive biases, which restrict their ability to capture such recurring structures. In this paper, we propose MoFo, which interprets periodicity as both the correlation of period-aligned time steps and the trend of period-offset time steps. We first design period-structured patches—2D tensors generated through discrete sampling—where each row contains only period-aligned time steps, enabling direct modeling of periodic correlations. Period-offset time steps within a period are aligned in columns. To capture trends across these offset time steps, we introduce a period-aware modulator. This modulator introduces an adaptive strong inductive bias through a regulated relaxation function, encouraging the model to generate attention coefficients that align with periodic trends. This function is end-to-end trainable, enabling the model to adaptively capture the distinct periodic patterns across diverse datasets. Extensive empirical results on widely used benchmark datasets demonstrate that MoFo achieves competitive performance while maintaining high memory efficiency and fast training speed.

PDF Details

AAAI Conference 2025 Conference Paper

Multi-label Self Knowledge Distillation

Xucong Wang
Pengkun Wang
Shurui Zhang
Miao Fang
Yang Wang

Self-Knowledge Distillation (SKD) leverages the student's own knowledge to create a virtual teacher for distillation when the pre-trained bulky teacher is not available. Whilst existing SKD approaches demonstrate gorgeous efficiency in single-label learning, to directly apply them to multi-label learning would suffer from dramatic degradation due to the following inherent imbalance: \textit{targets with unified labels but multifarious visual scales are crammed into one image, resulting in biased learning of major targets and disequilibrium of precision-recall}. To address this issue, this paper proposes a novel SKD method for multi-label learning named Multi-label Self-knowledge Distillation (MSKD), incorporating three Spatial Decoupling mechanisms (i.e. Locality-SD (L-SD), Reconstruction-SD (R-SD), and Step-SD (S-SD)). L-SD exploits relational dark knowledge from regional outputs to amplify the model's perception of visual details. R-SD reconstructs global semantics by integrating regional outputs from local patches and leverages it to guide the model. S-SD aligns outputs of the same input at different steps, aiming to find a synthetical optimizing direction and avoid the overconfidence. In addition, MSKD combines our tailored loss named MBD for balanced distillation. Exhaustive experiments demonstrate that MSKD not only outperforms previous approaches but also effectively mitigates biased learning and equips the model with more robustness.

PDF Details DOI

AAAI Conference 2025 Conference Paper

STEM-LTS: Integrating Semantic-Temporal Dynamics in LLM-driven Time Series Analysis

Zhe Zhao
Pengkun Wang
Haibin Wen
Shuang Wang
Liheng Yu
Yang Wang

Time series forecasting plays a crucial role in domains such as finance, healthcare, and climate science. However, as modern time series data become increasingly complex, featuring high dimensionality, intricate spatiotemporal dependencies, and multi-scale evolutionary patterns, traditional analytical methods and existing predictive models face significant challenges. Although Large Language Models (LLMs) excel in capturing long-range dependencies, they still struggle with multi-scale dynamics and seasonal patterns. Moreover, while LLMs' semantic representation capabilities are rich, they often lack explicit alignment with the numerical patterns and temporal structures of time series data, leading to limitations in predictive accuracy and interpretability. To address these challenges, this paper proposes a novel framework, STEM-LTS (Semantic-TEmporal Modeling for Large-scale Time Series). STEM-LTS enhances the ability to capture complex spatiotemporal dependencies by integrating time series decomposition techniques with LLM-based modeling. The semantic-temporal alignment mechanism within the framework significantly improves LLMs' ability to interpret and forecast time series data. Additionally, we develop an adaptive multi-task learning strategy to optimize the model's performance across multiple dimensions. Through extensive experiments on various real-world datasets, we demonstrate that STEM-LTS achieves significant improvements in prediction accuracy, robustness to noise, and interpretability. Our work not only advances LLM-based time series analysis but also offers new perspectives on handling complex temporal data.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

TS-MOF: Two-Stage Multi-Objective Fine-tuning for Long-Tailed Recognition

Zhe Zhao
Zhiheng Gong
Pengkun Wang
Haibin Wen
Cankun Guo
Bo Xue
Xi Lin
Zhenkun Wang

Long-Tailed Recognition (LTR) presents a significant challenge due to extreme class imbalance, where existing methods often struggle to balance performance across head and tail classes. Directly applying multi-objective optimization (MOO) to leverage multiple LTR strategies can be complex and unstable. To address this, we propose TS-MOF (Two-Stage Multi-Objective Fine-tuning), a novel framework that strategically decouples feature learning from classifier adaptation. After standard pre-training, TS-MOF freezes the feature backbone and focuses on an efficient multi-objective fine-tuning of specialized classifier heads. The core of TS-MOF's second stage lies in two innovations: Refined Performance Level Agreement for adaptive task weighting based on real-time per-class performance, and Robust Deterministic Projective Conflict Gradient for stable gradient conflict resolution and constructive fusion. This approach enables effective synergy between diverse LTR strategies, leading to significant and balanced performance improvements. Extensive experiments on CIFAR100-LT, ImageNet-LT, and iNaturalist 2018 demonstrate that TS-MOF achieves state-of-the-art results, particularly enhancing tail class accuracy (e. g. , +3. 3\% on CIFAR100-LT IR=100 tail) while improving head class performance, all within a remarkably short fine-tuning period of 20 epochs.

PDF Details

AAAI Conference 2024 Conference Paper

A Twist for Graph Classification: Optimizing Causal Information Flow in Graph Neural Networks

Zhe Zhao
Pengkun Wang
Haibin Wen
Yudong Zhang
Zhengyang Zhou
Yang Wang

Graph neural networks (GNNs) have achieved state-of-the-art results on many graph representation learning tasks by exploiting statistical correlations. However, numerous observations have shown that such correlations may not reflect the true causal mechanisms underlying the data and thus may hamper the ability of the model to generalize beyond the observed distribution. To address this problem, we propose an Information-based Causal Learning (ICL) framework that combines information theory and causality to analyze and improve graph representation learning to transform information relevance to causal dependence. Specifically, we first introduce a multi-objective mutual information optimization objective derived from information-theoretic analysis and causal learning principles to simultaneously extract invariant and interpretable causal information and reduce reliance on non-causal information in correlations. To optimize this multi-objective objective, we enable a causal disentanglement layer that effectively decouples the causal and non-causal information in the graph representations. Moreover, due to the intractability of mutual information estimation, we derive variational bounds that enable us to transform the above objective into a tractable loss function. To balance the multiple information objectives and avoid optimization conflicts, we leverage multi-objective gradient descent to achieve a stable and efficient transformation from informational correlation to causal dependency. Our approach provides important insights into modulating the information flow in GNNs to enhance their reliability and generalization. Extensive experiments demonstrate that our approach significantly improves the robustness and interpretability of GNNs across different distribution shifts. Visual analysis demonstrates how our method converts informative dependencies in representations into causal dependencies.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

Breaking Long-Tailed Learning Bottlenecks: A Controllable Paradigm with Hypernetwork-Generated Diverse Experts

Zhe Zhao
Haibin Wen
Zikang Wang
Pengkun Wang
Fanfu Wang
Song Lai
Qingfu Zhang
Yang Wang

Traditional long-tailed learning methods often perform poorly when dealing with inconsistencies between training and test data distributions, and they cannot flexibly adapt to different user preferences for trade-offs between head and tail classes. To address this issue, we propose a novel long-tailed learning paradigm that aims to tackle distribution shift in real-world scenarios and accommodate different user preferences for the trade-off between head and tail classes. We generate a set of diverse expert models via hypernetworks to cover all possible distribution scenarios, and optimize the model ensemble to adapt to any test distribution. Crucially, in any distribution scenario, we can flexibly output a dedicated model solution that matches the user's preference. Extensive experiments demonstrate that our method not only achieves higher performance ceilings but also effectively overcomes distribution shift while allowing controllable adjustments according to user preferences. We provide new insights and a paradigm for the long-tailed learning problem, greatly expanding its applicability in practical scenarios. The code can be found here: https: //github. com/DataLab-atom/PRL.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

LLM-AutoDA: Large Language Model-Driven Automatic Data Augmentation for Long-tailed Problems

Pengkun Wang
Zhe Zhao
Haibin Wen
Fanfu Wang
Binwu Wang
Qingfu Zhang
Yang Wang

The long-tailed distribution is the underlying nature of real-world data, and it presents unprecedented challenges for training deep learning models. Existing long-tailed learning paradigms based on re-balancing or data augmentation have partially alleviated the long-tailed problem. However, they still have limitations, such as relying on manually designed augmentation strategies, having a limited search space, and using fixed augmentation strategies. To address these limitations, this paper proposes a novel LLM-based long-tailed data augmentation framework called LLM-AutoDA, which leverages large-scale pretrained models to automatically search for the optimal augmentation strategies suitable for long-tailed data distributions. In addition, it applies this strategy to the original imbalanced data to create an augmented dataset and fine-tune the underlying long-tailed learning model. The performance improvement on the validation set serves as a reward signal to update the generation model, enabling the generation of more effective augmentation strategies in the next iteration. We conducted extensive experiments on multiple mainstream long-tailed learning benchmarks. The results show that LLM-AutoDA outperforms state-of-the-art data augmentation methods and other re-balancing methods significantly.

PDF Details DOI

IJCAI Conference 2024 Conference Paper

Make Bricks with a Little Straw: Large-Scale Spatio-Temporal Graph Learning with Restricted GPU-Memory Capacity

Binwu Wang
Pengkun Wang
Zhengyang Zhou
Zhe Zhao
Wei Xu
Yang Wang

Traffic prediction plays a key role in various smart city applications, which can help traffic managers make traffic plans in advance, assist online ride-hailing companies in deploying vehicles reasonably, and provide early warning of congestion for safety authorities. While increasingly complex models achieve impressive prediction performance, there are concerns about the effectiveness of these models in handling large-scale road networks. Especially for researchers who don't have access to powerful GPU devices, the expensive memory burden limits the usefulness of these models. In this paper, we take the first step of learning on the large-scale spatio-temporal graph and propose a divide-and-conquer training strategy for Large Spatio-Temporal Graph Learning, namely LarSTL. The core idea behind this strategy is to divide the large graph into multiple subgraphs, which are treated as task streams to sequentially train the model to conquer each subgraph one by one. We introduce a novel perspective based on the continuous learning paradigm to achieve this goal. In order to overcome forgetting the knowledge learned from previous subgraphs, an experience-replay strategy consolidates the learned knowledge by replaying nodes sampled from previous subgraphs. Moreover, we configure specific feature adaptors for each subgraph to extract personalized features, and it is also beneficial to consolidate the learned knowledge from the perspective of parameters. We conduct experiments using multiple large-scale traffic network datasets on a V100 GPU with only 16GB memory, and the results demonstrate that our LarSTL can achieve competitive performance and high efficiency.

PDF Details DOI

AAAI Conference 2024 Conference Paper

Towards Dynamic Spatial-Temporal Graph Learning: A Decoupled Perspective

Binwu Wang
Pengkun Wang
Yudong Zhang
Xu Wang
Zhengyang Zhou
Lei Bai
Yang Wang

With the progress of urban transportation systems, a significant amount of high-quality traffic data is continuously collected through streaming manners, which has propelled the prosperity of the field of spatial-temporal graph prediction. In this paper, rather than solely focusing on designing powerful models for static graphs, we shift our focus to spatial-temporal graph prediction in the dynamic scenario, which involves a continuously expanding and evolving underlying graph. To address inherent challenges, a decoupled learning framework (DLF) is proposed in this paper, which consists of a spatial-temporal graph learning network (DSTG) with a specialized decoupling training strategy. Incorporating inductive biases of time-series structures, DSTG can interpret time dependencies into latent trend and seasonal terms. To enable prompt adaptation to the evolving distribution of the dynamic graph, our decoupling training strategy is devised to iteratively update these two types of patterns. Specifically, for learning seasonal patterns, we conduct thorough training for the model using a long time series (e.g., three months of data). To enhance the learning ability of the model, we also introduce the masked auto-encoding mechanism. During this period, we frequently update trend patterns to expand new information from dynamic graphs. Considering both effectiveness and efficiency, we develop a subnet sampling strategy to select a few representative nodes for fine-tuning the weights of the model. These sampled nodes cover unseen patterns and previously learned patterns. Experiments on dynamic spatial-temporal graph datasets further demonstrate the competitive performance, superior efficiency, and strong scalability of the proposed framework.

PDF Details DOI