Junchen Fu

🚀 Added a reusable Cursor/Claude Code prompt for IISAN! Paste it into your AI coding assistant to instantly integrate our side adapters into your multimodal model pipeline for efficient fine-tuning. 📷 github.com/GAIR-Lab/IISAN #PEFT #MultimodalLearning #RS #IR #SIGIR #TKDE

Differentiable Semantic ID for Generative Recommendation @ron_junchen_fu et al. enable joint optimization of semantic IDs and generative recommenders through Gumbel noise-based exploration and uncertainty decay strategies to prevent codebook collapse. 📝 arxiv.org/abs/2601.19711

✨ Honored to present at DIR 2025! Shared my PhD works — IISAN, IISAN-Versa, & CROSSAN — advancing efficient fine-tuning of multimodal foundation models for recommender systems. Grateful for the inspiring discussions with the Dutch-Belgian IR community! #DIR2025 #PEFT #IR #RS

You can find the arXiv paper here: arxiv.org/abs/2411.02992. The code implementation will be available at: github.com/GAIR-Lab/IISAN.”

Excited to share that our paper "Efficient and Effective Adaptation of Multimodal Foundation Models in Sequential Recommendation" is accepted at IEEE TKDE! 🎉📷 Joint work with @Solis_xurige @alexk_z @iarapakis @joemonj.

3 papers in 5 days. Authors shall cite each other and consolidate all findings :) #recsys

@AixinSG Haha, what a coincidence — good point! We’ll be sure to include citations to the other two papers in the updated version.

Call for papers of the EReL@MIR Workshop (Efficient Representation Learning for Multimodal IR) at WWW2025: erel-mir.github.io. Submission Website: easychair.org/conferences/?c… Submission Deadline extended to Jan 1, 2025. Contributions about multimodal IR&RecSys are welcome!

The submission link for our EReL@MIR Workshop (Efficient Representation Learning for Multimodal IR, erel-mir.github.io) at WWW2025 (Sydney, Australia) is now open. easychair.org/conferences/?c… Welcome to contribute. Submission DDL: 18 Dec., 2024 Notification Date: 13 Jan., 2025

Glad to organize the 1st EReL@MIR Workshop (Efficient Representation Learning for Multimodal IR) erel-mir.github.io at The Webconf25 (Sydney, Australia) with @joemonj @alexk_z @iarapakis @XinXin_SDU @YueFeng__ @twitoogle @ron_junchen_fu . Welcome everyone to contribute.

See our paper for further details: arxiv.org/abs/2404.02059

Our paper, "IISAN: Efficiently Adapting Multimodal Representation for Sequential Recommendation with Decoupled PEFT," has been accepted at #SIGIR2024 as a full paper!🎉🎉🎉 Thrilled to share this collaboration with @Solis_xurige , @joemonj , @alexk_z, & @iarapakis. 👏👏👏

IISAN merges two modality towers with straightforward intra- and inter-modal side-adapted networks, significantly reducing GPU memory and training time. Additionally, we introduce a simple TPME metric and analysis for PEFTs' efficiency. arxiv.org/abs/2404.02059 #SIGIR2024