Author name cluster
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.
EAAI Journal 2026 Journal Article
YNIMG Journal 2026 Journal Article
BACKGROUND: Intermittent theta burst stimulation (iTBS) targeting the dorsolateral prefrontal cortex (DLPFC) shows promise for enhancing cognitive performance. However, behavioral findings remain inconsistent, likely due to methodological limitations in prior research and the poorly understood role of individual variability in neural responsiveness. Only one prior study has utilized individualized targeting. The effects of iTBS on resting-state functional networks remain unexplored. METHODS: We employed fMRI during an N-back task to identify individualized left DLPFC stimulation targets, based on the 2-back > 1-back contrast. A total of 56 healthy participants were randomly assigned to receive a single session of either active or sham iTBS (28 per group). Resting-state and task-based fMRI (2-back and 3-back) were acquired before and after stimulation. RESULTS: Compared to sham, iTBS increased neural activity in the middle cingulate cortex (MCC) and calcarine cortex during the 3-back task. Moreover, iTBS increased FC between the stimulation target and insula whilst reduced FCs within the default mode network (DMN) and between the DMN and frontoparietal network (FPN). Notably, in the iTBS group, greater MCC activation and enhanced target-insula FC were associated with faster 3-back reaction time (RT), whereas greater DMN FC reductions correlated with improved 3-back accuracy. Behaviorally, individuals with slower baseline 3-back RT in the iTBS group exhibited faster post-stimulation RT - an effect absent in the sham. CONCLUSIONS: These findings suggest that individualized iTBS modulates neural activity and distributed functional networks to support cognitive improvement - particularly in individuals with lower baseline ability - and highlight its potential for personalized cognitive interventions.
EAAI Journal 2026 Journal Article
Permanent magnet synchronous motors (PMSMs) play a critical role in renewable energy, transportation, and industrial drives due to their high efficiency and reliability. Whereas thermal-induced demagnetization under high-temperature conditions can lead to irreversible performance loss, maintenance in current systems is predominantly reactive. To address this limitation, this study proposes a physics-informed framework that integrates an extended Arrhenius degradation model with a feedforward neural network (FNN). The framework explicitly captures the nonlinear coupling between temperature and exposure time, enabling prediction of the demagnetization trend at given thermal condition. Uncertainty quantification is achieved via Monte Carlo parameter perturbations, from which 95% confidence intervals (CI) are constructed to improve prediction reliability. A probabilistic safety-boundary method then classifies material health states and determines the maximum safe operating time. Experimental validation demonstrates that the proposed approach reduces prediction error, enabling more effective maintenance planning. The framework thus offers a robust tool for assessing and mitigating demagnetization risks in high-temperature environments.
AAAI Conference 2026 Conference Paper
Brain decoding currently faces significant challenges in individual differences, modality alignment, and high-dimensional embeddings. To address individual differences, researchers often use source subject data, which leads to issues such as privacy leakage and heavy data storage burdens. In modality alignment, current works focus on aligning the softmax probability distribution but neglect the alignment of marginal probability distributions, resulting in modality misalignment. Additionally, images and text are aligned separately with fMRI without considering the complex interplay between images and text, leading to poor image reconstruction. Finally, the enormous dimensionality of CLIP embeddings causes significant computational costs. Although the dimensionality of CLIP embeddings can be reduced by ignoring the number of patches obtained from images and the number of tokens acquired from text, this comes at the cost of a significant drop in model performance, creating a dilemma. To overcome these limitations, we propose a source-free domain adaptation-based brain decoding framework. Firstly, we apply source-free domain adaptation, which only acquires the source model without accessing source data during target model adaptation, to brain decoding to address cross-subject variations, privacy concerns, and the heavy burden of data storage. Secondly, we employ maximum mean discrepancy (MMD) to align the marginal probability distributions between embeddings of different modalities. Moreover, to accommodate the complex interplay between image and text, we concatenate the embeddings of image and text and then use singular value decomposition (SVD) to obtain a new embedding. What’s more, to achieve better image generation quality, we employ the Wasserstein distance (WD) to align the probability distributions of new embeddings. Finally, in the target model adaptation phase of source-free domain adaptation, we employ low-rank adaptation (LoRA) to reduce the high expense of tuning the target model. Sufficient experiments demonstrate our work outperforms state-of-the-art methods for brain decoding tasks.
IJCAI Conference 2025 Conference Paper
Recent advancements in instruction tuning for large language models (LLMs) suggest that a small, high-quality dataset can significantly equip LLMs with instruction-following capabilities, outperforming large datasets often burdened by quality and redundancy issues. However, the challenge lies in automatically identifying valuable subsets from large datasets to boost both the effectiveness and efficiency of instruction tuning. In this paper, we first establish data selection criteria based on three distinct aspects of data value: diversity, difficulty, and dependability, and then propose the D3 method comprising two key steps of scoring and selection. Specifically, in the scoring step, we define the diversity function to measure sample distinctiveness and introduce the uncertainty-based prediction difficulty to evaluate sample difficulty by mitigating the interference of context-oriented generation diversity. Additionally, we integrate an external LLM for dependability assessment. In the selection step, we formulate the D3 weighted coreset objective, which jointly optimizes three aspects of data value to solve for the most valuable subset. The two steps of D3 can iterate multiple rounds, incorporating feedback to refine the selection focus adaptively. Experiments on both public datasets and the real-world Taobao Live application demonstrate the effectiveness of D3 in endowing LLMs with competitive or even superior instruction-following capabilities using less than 10% of the entire dataset.
NeurIPS Conference 2025 Conference Paper
Equivariant Graph Neural Networks (EGNNs) have demonstrated significant success in modeling microscale systems, including those in chemistry, biology and materials science. However, EGNNs face substantial computational challenges due to the high cost of constructing edge features via spherical tensor products, making them almost impractical for large-scale systems. To address this limitation, we introduce E2Former, an equivariant and efficient transformer architecture that incorporates a Wigner $6j$ convolution (Wigner $6j$ Conv). By shifting the computational burden from edges to nodes, Wigner $6j$ Conv reduces the complexity from $O(| \mathcal{E}|)$ to $O(| \mathcal{V}|)$ while preserving both the model's expressive power and rotational equivariance. We show that this approach achieves a 7x–30x speedup compared to conventional $\mathrm{SO}(3)$ convolutions. Furthermore, our empirical results demonstrate that the derived E2Former mitigates the computational challenges of existing approaches without compromising the ability to capture detailed geometric information. This development could suggest a promising direction for scalable molecular modeling.
IJCAI Conference 2025 Conference Paper
Predicting the performance of deep learning (DL) models, such as execution time and resource utilization, is crucial for Neural Architecture Search (NAS), DL cluster schedulers, and other technologies that advance deep learning. The representation of a model is the foundation for its performance prediction. However, existing methods cannot comprehensively represent diverse model configurations, resulting in unsatisfactory accuracy. To address this, we represent a model as a graph that includes the topology, along with node, edge, and global features, all of which are crucial for effectively capturing the performance of the model. Based on this representation, we propose PerfSeer, a novel predictor that uses a Graph Neural Network (GNN)-based performance prediction model, SeerNet. SeerNet fully leverages the topology and various features, while incorporating optimizations such as Synergistic Max-Mean aggregation (SynMM) and Global-Node Perspective Boost (GNPB) to more effectively capture the critical performance information, enabling it to predict the performance of models accurately. Furthermore, SeerNet can be extended to SeerNet-Multi by using Project Conflicting Gradients (PCGrad), enabling efficient simultaneous prediction of multiple performance metrics without significantly affecting accuracy. We constructed a dataset containing performance metrics for 53k+ model configurations, including execution time, memory usage, and Streaming Multiprocessor (SM) utilization during both training and inference. The evaluation results show that PerfSeer outperforms nn-Meter, Brp-NAS, and DIPPM.
EAAI Journal 2024 Journal Article
The weakly supervised semantic segmentation (WSSS) training based on image-level labels in convolutional neural network (CNN) is usually divided into two stages: multi-label classification and semantic segmentation. However, most of the existing work focuses on the improvement of the multi-label classification network stage, and little effort has been done to improve the performance of the downstream segmentation networks. In addition, CNN-based local convolution lacks in modeling extensive dependencies among categories. Therefore, in this paper, we propose a graph reasoning method to improve both the upstream and the downstream stages of the multi-label classification network and the semantic segmentation networks. In the multi-label classification network, we utilize external knowledge combined with a graph convolutional network (GCN) to perform global reasoning on the dependencies of each category. In the segmentation network, the Graph Reasoning Mapping Module (GRM) is proposed to explore the knowledge acquired from text corpora and facilitate contextual reasoning in various categories of image regions. The proposed GRM module is able to enhance the feature representation of local convolutions on the high-level semantics of the segmentation network, and adaptively learn the semantic consistency of each sample. We achieve state-of-the-art WSSS performance on PASCAL VOC 2012 and MS-COCO 2014 datasets with only image-level supervision. Extensive experiments on multi-label classification networks and semantic segmentation networks demonstrate the effectiveness of our proposed graph reasoning method on WSSS. Our code is available at: https: //github. com/JIA-ZHANG666/GRM_layer.
AAAI Conference 2024 Conference Paper
Most noise cleaning methods adopt one of the correction and filtering modes to build robust models. However, their effectiveness, applicability, and hyper-parameter insensitivity have not been carefully studied. We compare the two cleaning modes via a rebuilt error bound in noisy environments. At the dataset level, Theorem 5 implies that correction is more effective than filtering when the cleaned datasets have close noise rates. At the sample level, Theorem 6 indicates that confident label noises (large noise probabilities) are more suitable to be corrected, and unconfident noises (medium noise probabilities) should be filtered. Besides, an imperfect hyper-parameter may have fewer negative impacts on filtering than correction. Unlike existing methods with a single cleaning mode, the proposed Fusion cleaning framework of Correction and Filtering (FCF) combines the advantages of different modes to deal with diverse suspicious labels. Experimental results demonstrate that our FCF method can achieve state-of-the-art performance on benchmark datasets.
YNIMG Journal 2023 Journal Article
The process of complex cognition, which includes language processing, is dynamic in nature and involves various network modes or cognitive modes. This dynamic process can be manifested by a set of brain states and transitions between them. Previous neuroimaging studies have shed light on how bilingual brains support native language (L1) and second language (L2) through a shared network. However, the mechanism through which this shared brain network enables L1 and L2 processing remains unknown. This study examined this issue by testing the hypothesis that L1 and L2 processing is associated with distinct brain state dynamics in terms of brain state integration and transition flexibility. A group of late Chinese-English bilinguals was scanned using functional magnetic resonance imaging (fMRI) while listening to eight short narratives in Chinese (L1) and English (L2). Brain state dynamics were modeled using the leading eigenvector dynamic analysis framework. The results show that L1 processing involves more integrated states and frequent transitions between integrated and segregated states, while L2 processing involves more segregated states and fewer transitions. Our work provides insight into the dynamic process of narrative listening comprehension in late bilinguals and sheds new light on the neural representation of language processing and related disorders.
YNIMG Journal 2023 Journal Article
Additional neural substance for reading in a second language has been reported by prior studies. However, to date, there has been little investigation into whether and how the brain's adaptation to a second language is induced by specific linguistic tasks or is a general effect during reading in a new language. To address this issue, our study investigated Chinese children learning English as a second language by combining cross-sectional and longitudinal Functional Magnetic Resonance Imaging (fMRI) studies. We compared brain activation across four reading tasks, orthographic tasks and phonological tasks in Chinese (the first language, L1) and English (the second language, L2). By comparing the activation pattern across languages, we observed greater activation in the left inferior parietal lobule (LIPL) in English compared to Chinese, suggesting a functional preference of the LIPL to L2. In addition, greater correlation between LIPL-related FC and L2 was mainly observed in the phonological task, indicating that LIPL could be associated with phonological processing. Moreover, a proportion of the children were enrolled in an 8-week phonological-based reading-training program. We observed significant functional plasticity of the LIPL elicited by this training program only in the English phonological task and not in the orthographic task, further substantiating that the additional requirements of the LIPL in L2 are mainly associated with phonological processing. The findings provide new insights into understanding the functional contribution of the LIPL to reading in a second language.
AAMAS Conference 2021 Conference Paper
Multi-agent reinforcement learning (MARL) has been increasingly explored to learn the cooperative policy towards maximizing a certain global reward. Many existing studies take advantage of graph neural networks (GNN) in MARL to propagate critical collaborative information over the interaction graph, built upon inter-connected agents. Nevertheless, the vanilla GNN approach yields substantial defects in dealing with complex real-world scenarios since the generic message passing mechanism is ineffective between heterogeneous vertices and, moreover, simple message aggregation functions are incapable of accurately modeling the combinational interactions from multiple neighbors. While adopting complex GNN models with more informative message passing and aggregation mechanisms can obviously benefit heterogeneous vertex representations and cooperative policy learning, it could, on the other hand, increase the training difficulty of MARL and demand more intense and direct reward signals compared to the original global reward. To address these challenges, we propose a new cooperative learning framework with pre-trained heterogeneous observation representations. Particularly, we employ an encoder-decoder based graph attention to learn the intricate interactions and heterogeneous representations that can be more easily leveraged by MARL. Moreover, we design a pre-training with local actor-critic algorithm to ease the difficulty in cooperative policy learning. Extensive experiments over real-world scenarios demonstrate that our new approach can significantly outperform existing MARL baselines as well as operational research solutions that are widely-used in industry. Proc. of the 20th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2021), U. Endriss, A. Nowé, F. Dignum, A. Lomuscio (eds.), May 3–7, 2021, Online. © 2021 International Foundation for Autonomous Agents and Multiagent Systems (www. ifaamas. org). All rights reserved.
YNIMG Journal 2021 Journal Article
It is widely accepted that impairment in visual perception impedes children's reading development, and further studies have demonstrated significant enhancement in reading fluency after visual perceptual training. However, the mechanism of the neural linkage between visual perception and reading is unclear. The purpose of this study was to examine the intrinsic functional relationship between visual perception (indexed by the texture discrimination task,TDT) and reading ability (character reading and reading fluency) in Chinese children with developmental dyslexia (DD) and those with typical development (TD). The resting-state functional connectivity (RSFC) between the primary visual cortex (V1, BA17) and the entire brain was analyzed. In addition, how RSFC maps are associated with TDT performance and reading ability in the DD and TD groups was examined. The results demonstrated that the strength of the RSFC between V1 and the left middle frontal gyrus (LMFG, BA9/BA46) was significantly correlated with both the threshold (SOA) of the TDT and reading fluency in TD children but not in DD children. Moreover, LMFG-V1 resting-state connectivity played a mediating role in the association of visual texture discrimination and reading fluency, but not in character reading, in TD children. In contrast, this mediation was absent in DD children, albeit their strengths of RSFC between V1 and the left middle frontal gyrus (LMFG) were comparable to those for the TD group. These findings indicate that typically developing children use the linkage of the RSFC between the V1 and LMFG for visual perception skills, which in turn promote fluent reading; in contrast, children with dyslexia, who had higher TDT thresholds than TD children, could not take advantage of their frontal-occipital connectivity to improve reading fluency abilities. These findings suggest that visual perception plays an important role in reading skills and that children with developmental dyslexia lack the ability to use their frontal-occipital connectivity to link visual perception with reading fluency.
AAAI Conference 2020 Conference Paper
Model compression has become necessary when applying neural networks (NN) into many real application tasks that can accept slightly-reduced model accuracy but with strict tolerance to model complexity. Recently, Knowledge Distillation, which distills the knowledge from well-trained and highly complex teacher model into a compact student model, has been widely used for model compression. However, under the strict requirement on the resource cost, it is quite challenging to make student model achieve comparable performance with the teacher one, essentially due to the drasticallyreduced expressiveness ability of the compact student model. Inspired by the nature of the expressiveness ability in NN, we propose to use multi-segment activation, which can significantly improve the expressiveness ability with very little cost, in the compact student model. Specifically, we propose a highly efficient multi-segment activation, called Light Multisegment Activation (LMA), which can rapidly produce multiple linear regions with very few parameters by leveraging the statistical information. With using LMA, the compact student model is capable of achieving much better performance effectively and efficiently, than the ReLU-equipped one with same model complexity. Furthermore, the proposed method is compatible with other model compression techniques, such as quantization, which means they can be used jointly for better compression performance. Experiments on state-of-the-art NN architectures over the real-world tasks demonstrate the effectiveness and extensibility of the LMA.
IJCAI Conference 2020 Conference Paper
Information theoretical based methods have attracted a great attention in recent years, and gained promising results to deal with multi-label data with high dimensionality. However, most of the existing methods are either directly transformed from heuristic single-label feature selection methods or inefficient in exploiting labeling information. Thus, they may not be able to get an optimal feature selection result shared by multiple labels. In this paper, we propose a general global optimization framework, in which feature relevance, label relevance (i. e. , label correlation), and feature redundancy are taken into account, thus facilitating multi-label feature selection. Moreover, the proposed method has an excellent mechanism for utilizing inherent properties of multi-label learning. Specially, we provide a formulation to extend the proposed method with label-specific features. Empirical studies on twenty multi-label data sets reveal the effectiveness and efficiency of the proposed method. Our implementation of the proposed method is available online at: https: //jiazhang-ml. pub/GRRO-master. zip.
AAMAS Conference 2019 Conference Paper
Resource balancing within complex transportation networks is one of the most important problems in real logistics domain. Traditional solutions on these problems leverage combinatorial optimization with demand and supply forecasting. However, the high complexity of transportation routes, severe uncertainty of future demand and supply, together with non-convex business constraints make it extremely challenging in the traditional resource management field. In this paper, we propose a novel sophisticated multi-agent reinforcement learning approach to address these challenges. In particular, inspired by the externalities especially the interactions among resource agents, we introduce an innovative cooperative mechanism for state and reward design resulting in more effective and efficient transportation. Extensive experiments on a simulated ocean transportation service demonstrate that our new approach can stimulate cooperation among agents and lead to much better performance. Compared with traditional solutions based on combinatorial optimization, our approach can give rise to a significant improvement in terms of both performance and stability.
AAAI Conference 2017 Conference Paper
In this work, we study the guaranteed delivery model which is widely used in online advertising. In the guaranteed delivery scenario, ad exposures (which are also called impressions in some works) to users are guaranteed by contracts signed in advance between advertisers and publishers. A crucial problem for the advertising platform is how to fully utilize the valuable user traffic to generate as much as possible revenue. Different from previous works which usually minimize the penalty of unsatisfied contracts and some other cost (e. g. representativeness), we propose the novel consumption minimization model, in which the primary objective is to minimize the user traffic consumed to satisfy all contracts. Under this model, we develop a near optimal method to deliver ads for users. The main advantage of our method lies in that it consumes nearly as least as possible user traffic to satisfy all contracts, therefore more contracts can be accepted to produce more revenue. It also enables the publishers to estimate how much user traffic is redundant or short so that they can sell or buy this part of traffic in bulk in the exchange market. Furthermore, it is robust with regard to priori knowledge of user type distribution. Finally, the simulation shows that our method outperforms the traditional state-of-the-art methods.
AAAI Conference 2017 Conference Paper
Selling reserved instances (or virtual machines) is a basic service in cloud computing. In this paper, we consider a more flexible pricing model for instance reservation, in which a customer can propose the time length and number of resources of her request, while in today’s industry, customers can only choose from several predefined reservation packages. Under this model, we design randomized mechanisms for customers coming online to optimize social welfare and providers’ revenue. We first consider a simple case, where the requests from the customers do not vary too much in terms of both length and value density. We design a randomized mechanism that achieves a competitive ratio 1 42 for both social welfare and revenue, which is a improvement as there is usually no revenue guarantee in previous works such as (Azar et al. 2015; Wang et al. 2015). This ratio can be improved up to 1 11 when we impose a realistic constraint on the maximum number of resources used by each request. On the hardness side, we show an upper bound 1 3 on competitive ratio for any randomized mechanism. We then extend our mechanism to the general case and achieve a competitive ratio 1 42 log k log T for both social welfare and revenue, where T is the ratio of the maximum request length to the minimum request length and k is the ratio of the maximum request value density to the minimum request value density. This result outperforms the previous upper bound 1 CkT for deterministic mechanisms (Wang et al. 2015). We also prove an upper bound 2 log 8kT for any randomized mechanism. All the mechanisms we provide are in a greedy style. They are truthful and easy to be integrated into practical cloud systems.
TCS Journal 2016 Journal Article
Cooperative games provide a framework for fair and stable profit allocation in multi-agent systems. Core, least-core and nucleolus are such solution concepts that characterize stability of cooperation. In this paper, we study the algorithmic issues of the least-core and nucleolus of threshold cardinality matching games (TCMG). A TCMG is defined on a graph G = ( V, E ) and a threshold T, in which the player set is V and the profit of a coalition S ⊆ V is 1 if the size of a maximum matching in G [ S ] meets or exceeds T, and 0 otherwise. We first show that for a TCMG, the problems of computing least-core value, finding and verifying least-core payoff are all polynomial-time solvable. We also provide a general characterization of the least-core for a large class of TCMG (cf. Theorem 2). Next, based on Gallai–Edmonds Decomposition in matching theory, we establish a concise formulation of the nucleolus for a special case of TCMG (when the threshold T equals 1). For arbitrary T, we prove that the nucleolus of TCMG can be obtained in polynomial time for bipartite graphs and graphs with a perfect matching.
EAAI Journal 2015 Journal Article
Nowadays, many researches focus on the selection of the team members, and aim to form a team that can solve a common task collaboratively or achieve a certain goal cooperatively. The leader or manager needs to maximize the benefit of the team by selecting the appropriate team members from the available candidates. However, the personal opinions of team members have not been considered in the team formation process. In this paper, in order to consider the mutual decision between the leader and the team members, a negotiation based team formation problem is defined firstly, in which the balance of the benefit requirements of the leader and the team members is addressed. Then, a negotiation model between the leader and the team members is proposed to implement the team formation process. According to the conflicts in the process, an incentive mechanism based on the payoff is adopted. Finally, the agent-based negotiation approach is introduced to construct the simulation environment of the negotiation. The negotiation process is iterated until the benefit requirements of the leader agent and the candidate agents are all satisfied or the deadline of the negotiation is reached. The experiment results indicate that the team can be formed successfully in different team contexts, such as different team sizes, different minimum criteria of team positions and different incentive levels. Comparing with the leader׳s unilateral decision process, the negotiation-based team formation can make the team members acquire more benefits. In addition, the different preferences of the candidates have different effects on the final benefit acquirements, and the different minimum criteria of team positions and incentive levels have different effects on the conflicts resolution.