Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Suggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works103 from public data
- DeepGCNs: Can GCNs Go As Deep As CNNs?1,662
This work presents new ways to successfully train very deep GCNs by borrowing concepts from CNNs, specifically residual/dense connections and dilated convolutions, and adapting them to GCN architectures, and building a very deep 56-layer GCN.
- Can GCNs Go as Deep as CNNs1,662
This work presents new ways to successfully train very deep GCNs by borrowing concepts from CNNs, specifically residual/dense connections and dilated convolutions, and adapting them to GCN architectures, and building a very deep 56-layer GCN.
- CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society420
This paper proposes a novel communicative agent framework named role-playing, using inception prompting to guide chat agents toward task completion while maintaining consistency with human intentions, and showcases how role-playing can be used to generate conversational data for studying the behaviors and capabilities of chat agents.
- PU-GCN: Point Cloud Upsampling using Graph Convolutional Networks278
A novel model called NodeShuffle is proposed, which uses a Graph Convolutional Network (GCN) to better encode local point information from point neighborhoods to improve state-of-the-art upsampling methods.
- 252
- DeepGCNs: Making GCNs Go as Deep as CNNs223
This work transfers concepts such as residual/dense connections and dilated convolutions from CNNs to GCNs in order to successfully train very deep GCNs, achieving very promising performance in part segmentation and semantic segmentation on point clouds and in node classification of protein functions across biological protein-protein interaction (PPI) graphs.
- 222
- 212
- 152
- FLAG: Adversarial Data Augmentation for Graph Neural Networks133
This work proposes a simple but effective solution, FLAG (Free Large-scale Adversarial Augmentation on Graphs), which iteratively augments node features with gradient-based adversarial perturbations during training, and boosts performance at test time.
- Mindstorms in Natural Language-Based Societies of Mind124
This work assemble and experiment with several of NLSOMs, leveraging mindstorms in them to solve some practical AI tasks: visual question answering, image captioning, text-to-image synthesis, 3D generation, egocentric retrieval, embodied AI, and general language-based task solving.
- 124
- Robust Optimization as Data Augmentation for Large-scale Graphs118
This work proposes FLAG (Free Large-scale Adversarial Augmentation on Graphs), which iteratively augments node features with gradient-based adversarial perturbations during training, which universally works in node classification, link prediction, and graph classification tasks.
- Room-Temperature Assembled MXene-Based Aerogels for High Mass-Loading Sodium-Ion Storage93
Room temperature graphene oxide-assisted assembly of 3D Ti3C2Tx MXene aerogels have been realized by introducing interfacial mediators (amino-propyltriethoxysilane, Mn2+, Fe2+, Zn2+, and Co2+).
- Vision-Language Models Do Not Understand Negation92
A data-centric approach wherein CLIP models are finetune on large-scale synthetic datasets containing millions of negated captions is explored, which can result in a 10% increase in recall on negated queries and a 28% boost in accuracy on multiple-choice questions with negated captions.
- Achieving High‐Performance 3D K+‐Pre‐intercalated Ti3C2Tx MXene for Potassium‐Ion Hybrid Capacitors via Regulating Electrolyte Solvation Structure86
It is found that the optimized solvation structure of the concentrated ether-based electrolyte leads to the formation of a thin and inorganic-rich solid electrolyte interphase (SEI) on the K + -pre-intercalated Ti 3 C 2 T x electrode, which is beneficial for interfacial stability and reaction kinetics.
- 84
- Diffusion-Based Scene Graph to Image Generation with Masked Contrastive Pre-Training79
The resulting method, called SGDiff, allows for the semantic manipulation of generated images by modifying scene graph nodes and connections, and outperforms state-of-the-art methods, as measured by both the Inception Score and Fr\'echet Inception Distance metrics.
- Can Large Language Model Agents Simulate Human Trust Behavior?76
This study finds that GPT-4 agents manifest high behavioral alignment with humans in terms of trust behavior, indicating the feasibility of simulating human trust behavior with LLM agents, and probes the biases of agent trust and differences in agent trust towards other LLM agents and humans.
- 65
- A Scalable Communication Protocol for Networks of Large Language Models55
This work introduces Agora, a meta protocol that leverages existing communication standards to make LLM-powered agents solve complex problems efficiently and observes the emergence of self-organising, fully automated protocols that achieve complex goals without human intervention.
- 53
- ScoreFlow: Mastering LLM Agent Workflows via Score-based Preference Optimization48
This work introduces ScoreFlow, a simple yet high-performance framework that leverages efficient gradient-based optimization in a continuous space and incorporates Score-DPO, a novel variant of the direct preference optimization method that accounts for quantitative feedback.
- OIL: Observational Imitation Learning46
Observational Imitation Learning is proposed, a novel imitation learning variant that supports online training and automatic selection of optimal behavior by observing multiple imperfect teachers and outperforms conventional imitation learning (IL) and reinforcement learning (RL) baselines and even humans in simulation.
- 43
- Incorporating External Knowledge to Answer Open-Domain Visual Questions with Dynamic Memory Networks41
- EMOS: Embodiment-aware Heterogeneous Multi-robot Operating System with LLM Agents40
The experimental results indicate that the robot's resume and the hierarchical design of the multi-agent system are essential for the effective operation of the heterogeneous multi-robot system within this intricate problem context.
- 40
- 40
- UPainting: Unified Text-to-Image Diffusion Generation with Cross-modal Guidance34
UPainting is proposed, which combines the power of large-scale Transformer language model in understanding language and image-text matching model in capturing cross-modal semantics and style, and greatly outperforms other models in terms of caption similarity and image fidelity in both simple and complex scenes.
- 32
- 32
- 27
- 26
- Surface Chemistry and Mesopore Dual Regulation by Sulfur‐Promised High Volumetric Capacity of Ti3C2Tx Films for Sodium‐Ion Storage26
Ti3 C2 Tx (MXene)-based films are prepared by using sulfur (S) as the mediator to modulate the surface chemistry and microstructure, generating S-doped mesoporous Ti3 C 2 Tx films with high flexibility, which exhibits a high volumetric capacity.
- 25
- Optimizing silicon doping levels for enhanced osteogenic and angiogenic properties of 3D-printed biphasic calcium phosphate scaffolds: An in vitro screening and in vivo validation study24
This study investigated the effects of silicon doping on both the physicochemical and biological properties of BCP, with a specific focus on osteogenic and angiogenic potentials, and confirmed that BCP scaffolds doped with 4 mol.% silicon effectively promoted vascularization and new bone formation, highlighting their potential for clinical bone defect repair.
- Multi-Domain Generalized Graph Meta Learning20
This paper introduces an empirical graph generalization method that uses empirical vectors to form a unified expression of non-Euclidean graph data and proposes a multi- domain graphs transformation approach to transform the learning tasks from multiple source-domain graphs with inequivalent feature spaces into a common domain, where graph meta learning is conducted to learn generalized knowledge.
- 19
- The Heterophilic Snowflake Hypothesis: Training and Empowering GNNs for Heterophilic Graphs17
The prevailing concept of "one node one receptive field" to the heterophilic graph is transferred, or the first time, and a proxy label predictor is constructed, enabling each node to possess a latent prediction distribution, which assists connected nodes in determining whether they should aggregate their associated neighbors.
- 16
- 14
- Learning a Controller Fusion Network by Online Trajectory Filtering for Vision-Based UAV Racing14
This paper proposes learning an optimized controller using a DNN that fuses multiple controllers and develops a network that beats all baselines in extensive experiments and approaches the performance of a professional human pilot.
- 13
- 3D isotropic MXene films enabling minimized polarization for enhanced sodium-ion storage performance12
- The Snowflake Hypothesis: Training Deep GNN with One Node One Receptive field12
The Snowflake Hypothesis is introduced -- a novel paradigm underpinning the concept of ``one node, one receptive field'' that can serve as a universal operator for a range of tasks, and it displays tremendous potential on deep GNNs.
- 10
- 10
- LC-NAS: Latency Constrained Neural Architecture Search for Point Cloud Networks10
This paper introduces a new NAS framework, dubbed LC-NAS, that searches for point cloud architectures constrained to a target latency, and implements a novel latency constraint formulation for the trade-off between accuracy and latency in the architecture search.
- 10
- 10
- Multistage Nanoparticle Delivery System—A New Approach to Cancer Therapeutics10
This review expounds on the main direction of multistage delivery considering the ways in which large particles are triggered to release small particles and how to change the particle size to enhance penetration.
- Teaching UAVs to Race With Observational Imitation Learning10
A modular network architecture that decouples perception from control, and is trained using Observational Imitation Learning (OIL), a novel imitation learning variant that supports online training and automatic selection of optimal behavior from observing multiple teachers is proposed.
- When NAS Meets Trees: An Efficient Algorithm for Neural Architecture Search9
A new NAS method called TNAS (NAS with trees), which improves search efficiency by exploring only a small number of architectures while also achieving a higher search accuracy, is proposed.
- 8
- Generating Editable Head Avatars with 3D Gaussian GANs8
This work proposes a novel approach that enhances the editability and animation control of 3D head avatars by incorporating 3D Gaussian Splatting (3DGS) as an explicit 3D representation, and introduces the Editable Gaussian Head (EG-Head) model, allowing precise expression control and flexible texture editing for accurate animation while preserving identity.
- The Snowflake Hypothesis: Training and Powering GNN with One Node One Receptive Field8
The Snowflake Hypothesis -- a novel paradigm underpinning the concept of "one node, one receptive field'' is introduced, proposing a corresponding uniqueness in the receptive fields of nodes in the GNNs.
- 8
- 8
- 8
- 7
- 6
- 6
- 6
- 6
- 6
- 6
- Learning Scene Flow in 3D Point Clouds with Noisy Pseudo Labels4
A novel scene flow method that captures 3D motions from point clouds without relying on ground-truth scene flow annotations and outperforms state-of-the-art self-supervised approaches, but also outperforms some supervised approaches that use accurate ground- Truth flows.
- A novel improved fuzzy support vector machine based stock price trend forecast model4
Improved the traditional support vector machine fuzzy prediction algorithm is proposed to improve the new model precision in stock price forecast.
- 3
- 3
- 3D Face Modeling via Weakly-Supervised Disentanglement Network Joint Identity-Consistency Prior3
A Weakly-Supervised Disentanglement Framework is introduced to facilitate the training of controllable 3D face models without an overly stringent labeling requirement, and incorporates a label-free second-order loss function for the expression factor to regulate deformation space and eliminate extraneous information, resulting in enhanced disentanglement.
- 3
- 2
- A dual-mode colorimetric and electrochemical strategy for sensitive detection of PRRSV using imidazole-copper nanozymes2
A dual-mode highly sensitive detection platform based on imidazole-copper (I-Cu) nanozymes integrating colorimetric and electrochemical detection for dual-mode detection of PRRSV and has potential application value in animal disease early warning and biosafety monitoring.
- 2
- 2
- 2
- CLOP: Video-and-Language Pre-Training with Knowledge Regularizations2
This work proposes a Cross-modaL knOwledge-enhanced Pre-training (CLOP) method with Knowledge Regularizations, demonstrating the value of incorporating knowledge regularizations into video-and-language pre-training.
- 2
- Commonsense Learning: An Indispensable Path towards Human-centric Multimedia2
The basic concept of commonsense knowledge and commonsense reasoning is introduced, commonsense resources and benchmarks are summarized, an overview on recent commonsense learning and reasoning methods are given, and several popular applications of commonsens knowledge in real-world scenarios are discussed.
- Distilling Tool Knowledge into Language Models via Back-Translated Traces1
This work proposes a new paradigm for distilling tool knowledge into LLMs purely through natural language, and shows that fine-tuning a small open-source model on these synthesized traces enables it to internalize both tool knowledge and structured reasoning patterns, yielding gains on competition-level math benchmarks without requiring tool access at inference.
- 1
- 1
- 1
- 1
- 1
- 1
- –
- DHMGAT: A Dynamic and Hierarchical Multi-Head Graph Attention Network for Fault Location in Distribution Networks–
This paper proposes a Dynamic and Hierarchical Multi-Head Graph Attention Network (DHMGAT) that overcomes the limitations of static graph assumptions and substantially outperforming conventional graph neural networks.
- –
- –
- –
- –
- –
- –
- –
- –
- Quality-aware face alignment using high-resolution spatial dependencies–
A transformer-based data distillation semi-supervised face alignment algorithm to improve algorithm robustness under partial occlusion and intensive training strategy to use more unlabeled data without the need for manual operation to further improve the performance of transformer thermal map detection networks is proposed.
- UnrealNAS: Can We Search Neural Architectures with Unreal Data?–
This work constructs three types of unreal datasets using randomly labeled real images; generated images and labels; and generated Gaussian noise with random labels to explore if NAS needs real data, suggesting the feasibility of performing NAS with unreal data.
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.