Feng Sun

AAAI Conference 2026 Conference Paper

SegMem-RAG: Adaptive Memory for Retrieval-Augmented Generation in Open-Ended Knowledge Environments

• Xuanbo Fan
• Tianqi Zhao
• Yi Cheng
• Chi Xiu
• Jiaxin Guo
• Boci Peng
• Bingjing Xu
• Jessica Zhang

Retrieval-Augmented Generation (RAG) improves the factual accuracy of large language models by grounding responses in external content. However, most RAG systems assume access to static and well-organized corpora with fixed retrieval logic. In practice, real-world sources are heterogeneous and unlabeled, including user-uploaded documents, manuals, and datasets. Effective access in such settings requires adaptive and self-directed retrieval behavior. We present SegMem‑RAG, a memory-augmented RAG framework that learns to route queries across multiple unlabeled corpora based on experience. It incrementally updates a structured memory and uses self-reflection to guide retrieval over time without supervision. Experimental results demonstrate that SegMem‑RAG significantly outperforms recent baselines in generation quality on multi-corpus QA tasks.

JBHI Journal 2025 Journal Article

EDSRNet: An Enhanced Decoder Semantic Recovery Network for 2D Medical Image Segmentation

• Feng Sun
• Ying Zhou
• Longxiangfeng Hu
• Yongyan Li
• Dan Zhao
• Yufeng Chen
• Yu He

In recent years, with the advancement of medical imaging technology, medical image segmentation has played a key role in assisting diagnosis and treatment planning. Current deep learning-based medical image segmentation methods mainly adopt encoder-decoder architecture design and have received wide attention. However, these methods still have some limitations, including: (1) Existing methods are often influenced by the significant semantic information gap when supplementing features for the decoder. (2) Existing methods do not simultaneously consider global and local information interaction during decoding, resulting in ineffective semantic recovery. Therefore, this paper proposes a novel Enhanced Decoder Semantic Recovery Network to address these challenges. Firstly, the Multi-Level Semantic Fusion (MLSF) module is introduced, which effectively fuses low-level features of the original image, encoder features, high-level features of the deepest network layer, and decoder features, and assigns weights based on semantic gaps. Secondly, the Multiscale Spatial Attention (MSSA) and Cross Convolution Channel Attention (CCCA) modules are employed to obtain richer feature information. Finally, the Global-Local Semantic Recovery (GLSR) module is designed to achieve better semantic recovery. Experiments on public datasets such as BUSI, CVC-ClinicDB, and Kvasir-SEG demonstrate that the proposed model improves IoU compared to suboptimal algorithms by 0. 81%, 0. 85% and 1. 98%, respectively, significantly enhancing the performance of 2D medical image segmentation. This method provides effective technical support for further development in the field of medical image.

AAAI Conference 2025 Conference Paper

MTL-LoRA: Low-Rank Adaptation for Multi-Task Learning

• Yaming Yang
• Dilxat Muhtar
• Yelong Shen
• Yuefeng Zhan
• Jianfeng Liu
• Yujing Wang
• Hao Sun
• Weiwei Deng

Parameter-efficient fine-tuning (PEFT) has been widely employed for domain adaptation, with LoRA being one of the most prominent methods due to its simplicity and effectiveness. However, in multi-task learning (MTL) scenarios, LoRA tends to obscure the distinction between tasks by projecting sparse high-dimensional features from different tasks into the same dense low-dimensional intrinsic space. This leads to task interference and suboptimal performance for LoRA and its variants. To tackle this challenge, we propose MTL-LoRA, which retains the advantages of low-rank adaptation while significantly enhancing MTL capabilities. MTL-LoRA augments LoRA by incorporating additional task-adaptive parameters that differentiate task-specific information and capture shared knowledge across various tasks within low-dimensional spaces. This approach enables pretrained models to jointly adapt to different target domains with a limited number of trainable parameters. Comprehensive experimental results, including evaluations on public academic benchmarks for natural language understanding, commonsense reasoning, and image-text understanding, as well as real-world industrial text Ads relevance datasets, demonstrate that MTL-LoRA outperforms LoRA and its various variants with comparable or even fewer learnable parameters in MTL setting.

EAAI Journal 2024 Journal Article

A traffic-weather generative adversarial network for traffic flow prediction for road networks under bad weather

• Wensong Zhang
• Ronghan Yao
• Ying Yuan
• Xiaojing Du
• Libing Wang
• Feng Sun

AAAI Conference 2024 Conference Paper

Text Diffusion with Reinforced Conditioning

• Yuxuan Liu
• Tianchi Yang
• Shaohan Huang
• Zihan Zhang
• Haizhen Huang
• Furu Wei
• Weiwei Deng
• Feng Sun

Diffusion models have demonstrated exceptional capability in generating high-quality images, videos, and audio. Due to their adaptiveness in iterative refinement, they provide a strong potential for achieving better non-autoregressive sequence generation. However, existing text diffusion models still fall short in their performance due to a challenge in handling the discreteness of language. This paper thoroughly analyzes text diffusion models and uncovers two significant limitations: degradation of self-conditioning during training and misalignment between training and sampling. Motivated by our findings, we propose a novel Text Diffusion model called TReC, which mitigates the degradation with Reinforced Conditioning and the misalignment by Time-Aware Variance Scaling. Our extensive experiments demonstrate the competitiveness of TReC against autoregressive, non-autoregressive, and diffusion baselines. Moreover, qualitative analysis shows its advanced ability to fully utilize the diffusion process in refining samples.