Degraded but Not Entirely Ineffective: PE-Based Deformable Graph Neural Networks
本文提出基于位置编码的可变形图神经网络模块PEBDSAM，解决传统GNN过平滑、长依赖压缩等问题，适用于异质图和大规模数据集。
0 citationsRead paper
Selected work
Representative Papers
PRIME: Mitigating Subgroup Optimization Competition in Shared CTR Top Networks with Plug-in Residual Input-Conditioned Mixture of Expert
本文针对CTR模型中子群优化竞争问题，提出PRIME方法，通过基于输入条件的专家混合和低秩残差专家组合，在保持原有预测功能的同时增强了模型性能。
0 citationsRead paper
Multi-Granularity Reasoning for Image Quality Assessment via Attribute-Aware Reinforcement Learning to Rank
This work addresses the limitation of existing image quality assessment (IQA) methods, which typically predict only a single scalar score while neglecting the multidimensional attributes—such as sharpness, color fidelity, noise, and composition—that underpin human perception. To overcome this, the authors propose MG-IQA, a novel framework that jointly predicts overall quality and fine-grained perceptual attributes in a single inference pass. MG-IQA integrates vision-language models with reinforcement learning–based ranking, leveraging attribute-aware prompts, a multidimensional Thurstone reward model, and a cross-domain alignment mechanism to enable interpretable, multi-granularity evaluation without requiring perceptual scale realignment. Experiments demonstrate that MG-IQA consistently outperforms state-of-the-art methods across eight IQA benchmarks, achieving an average 2.1% improvement in SRCC for overall quality prediction and generating explanations highly aligned with human judgments.
0 citationsRead paper
HSI Image Enhancement Classification Based on Knowledge Distillation: A Study on Forgetting
This work addresses catastrophic forgetting in incremental hyperspectral image classification by proposing a knowledge retention method that operates without storing samples from previously seen classes. The approach leverages a teacher model and employs a masking mechanism to perform partial-class knowledge distillation using only the new-class data available in each incremental phase. This design effectively decouples the distillation process and filters out misleading information, thereby preserving relevant knowledge from earlier tasks. Without relying on rehearsal or replay of old samples, the proposed method significantly enhances both classification accuracy and model robustness. Its effectiveness is consistently validated through comprehensive comparative and ablation experiments across multiple benchmarks.
0 citationsRead paper
DeCoT: Decomposing Complex Instructions for Enhanced Text-to-Image Generation with Large Language Models
Current text-to-image (T2I) models struggle to accurately interpret complex, lengthy textual prompts, exhibiting significant semantic gaps in fine-grained detail reconstruction, spatial relationship modeling, and multi-constraint coordination. To address this, we propose DeCoT—a two-stage framework. In Stage I, a large language model (LLM) performs structured decomposition and semantic clarification of the original prompt. In Stage II, hierarchical prompt construction and chain-of-thought–driven semantic enhancement jointly model text embeddings, compositional logic, and constraint conditions. DeCoT is architecture-agnostic and seamlessly integrates with mainstream T2I models without architectural modification. Evaluated on the LongBench-T2I benchmark, DeCoT integrated with Infinity-8B achieves a score of 3.52—substantially outperforming baseline methods. Both multimodal LLM (MLLM)-based and human evaluations confirm consistent improvements across textual fidelity, spatial consistency, and compositional plausibility.
0 citationsRead paper
Recent publications
Latest Papers
Degraded but Not Entirely Ineffective: PE-Based Deformable Graph Neural Networks
本文提出基于位置编码的可变形图神经网络模块PEBDSAM，解决传统GNN过平滑、长依赖压缩等问题，适用于异质图和大规模数据集。
0 citationsRead paper
PRIME: Mitigating Subgroup Optimization Competition in Shared CTR Top Networks with Plug-in Residual Input-Conditioned Mixture of Expert
本文针对CTR模型中子群优化竞争问题，提出PRIME方法，通过基于输入条件的专家混合和低秩残差专家组合，在保持原有预测功能的同时增强了模型性能。
0 citationsRead paper
Multi-Granularity Reasoning for Image Quality Assessment via Attribute-Aware Reinforcement Learning to Rank
This work addresses the limitation of existing image quality assessment (IQA) methods, which typically predict only a single scalar score while neglecting the multidimensional attributes—such as sharpness, color fidelity, noise, and composition—that underpin human perception. To overcome this, the authors propose MG-IQA, a novel framework that jointly predicts overall quality and fine-grained perceptual attributes in a single inference pass. MG-IQA integrates vision-language models with reinforcement learning–based ranking, leveraging attribute-aware prompts, a multidimensional Thurstone reward model, and a cross-domain alignment mechanism to enable interpretable, multi-granularity evaluation without requiring perceptual scale realignment. Experiments demonstrate that MG-IQA consistently outperforms state-of-the-art methods across eight IQA benchmarks, achieving an average 2.1% improvement in SRCC for overall quality prediction and generating explanations highly aligned with human judgments.
0 citationsRead paper
HSI Image Enhancement Classification Based on Knowledge Distillation: A Study on Forgetting
This work addresses catastrophic forgetting in incremental hyperspectral image classification by proposing a knowledge retention method that operates without storing samples from previously seen classes. The approach leverages a teacher model and employs a masking mechanism to perform partial-class knowledge distillation using only the new-class data available in each incremental phase. This design effectively decouples the distillation process and filters out misleading information, thereby preserving relevant knowledge from earlier tasks. Without relying on rehearsal or replay of old samples, the proposed method significantly enhances both classification accuracy and model robustness. Its effectiveness is consistently validated through comprehensive comparative and ablation experiments across multiple benchmarks.
0 citationsRead paper
DeCoT: Decomposing Complex Instructions for Enhanced Text-to-Image Generation with Large Language Models
Current text-to-image (T2I) models struggle to accurately interpret complex, lengthy textual prompts, exhibiting significant semantic gaps in fine-grained detail reconstruction, spatial relationship modeling, and multi-constraint coordination. To address this, we propose DeCoT—a two-stage framework. In Stage I, a large language model (LLM) performs structured decomposition and semantic clarification of the original prompt. In Stage II, hierarchical prompt construction and chain-of-thought–driven semantic enhancement jointly model text embeddings, compositional logic, and constraint conditions. DeCoT is architecture-agnostic and seamlessly integrates with mainstream T2I models without architectural modification. Evaluated on the LongBench-T2I benchmark, DeCoT integrated with Infinity-8B achieves a score of 3.52—substantially outperforming baseline methods. Both multimodal LLM (MLLM)-based and human evaluations confirm consistent improvements across textual fidelity, spatial consistency, and compositional plausibility.
0 citationsRead paper