paper-with-me

홈 › Papers

Guiding the Experts: Semantic Priors for Efficient and Focused MoE Routing

2025-05-24 · Chengxi Min, Wei Wang, Yahui Liu, Weixin Ye, Enver Sangineto, Qi Wang, Yao Zhao

Mixture-of-Experts (MoE) models have emerged as a promising direction for scaling vision architectures efficiently. Among them, Soft MoE improves training stability by assigning each token to all experts via continuous dispatch weights. However, current designs overlook the semantic structure which is implicitly encoded in these weights, resulting in suboptimal expert routing. In this paper, we discover that dispatch weights in Soft MoE inherently exhibit segmentation-like patterns but are not explicitly aligned with semantic regions. Motivated by this observation, we propose a foreground-guided enhancement strategy. Specifically, we introduce a spatially aware auxiliary loss that encourages expert activation to align with semantic foreground regions. To further reinforce this supervision, we integrate a lightweight LayerScale mechanism that improves information flow and stabilizes optimization in skip connections. Our method necessitates only minor architectural adjustments and can be seamlessly integrated into prevailing Soft MoE frameworks. Comprehensive experiments on ImageNet-1K and multiple smaller-scale classification benchmarks not only showcase consistent performance enhancements but also reveal more interpretable expert routing mechanisms.

📄 PDF Abstract BibTeX arXiv:2505.18586

Code (1)

0930mcx/guiding-experts 공식 구현 pytorch

Tasks

Mixture-of-Experts

Methods 이 논문이 사용한 방법론

MoE 설명 없음
LayerScale LayerScale is a method used for vision transformer architectures to help improve training dynamics. It adds a…
ALIGN In the ALIGN method, visual and language representations are jointly trained from noisy image alt-text data. The image and text encoders are learned via contrastive loss…
AWARE We propose to theoretically and empirically examine the effect of incorporating weighting schemes into walk-aggregating GNNs. To this end, we propose a simple, interpretable, and…

Similar Papers 제목 키워드 기반

Cosine-Similarity Routing with Semantic Anchors for Interpretable Mixture-of-Experts Language Models

2025-09-12 · Ivan Ternovtsii, Yurii Bilak arxiv

Mixture-of-Experts (MoE) models improve efficiency through sparse activation, but their learned gating functions provide limited insight into routing decisions. This work introduces the Semantic Resonance Architecture (S…

DoReMi: Bridging 3D Domains via Topology-Aware Domain-Representation Mixture of Experts

2025-11-14 · Mingwei Xing, Xinliang Wang, Yifeng Shi arxiv

Constructing a unified 3D scene understanding model has long been hindered by the significant topological discrepancies across different sensor modalities. While applying the Mixture-of-Experts (MoE) architecture is an e…

Scene Understanding

S2MoE: Robust Sparse Mixture of Experts via Stochastic Learning

2025-03-29 · Giang Do, Hung Le, Truyen Tran

Sparse Mixture of Experts (SMoE) enables efficient training of large language models by routing input tokens to a select number of experts. However, training SMoE remains challenging due to the issue of representation co…

Mixture-of-Experts

MoE-LPR: Multilingual Extension of Large Language Models through Mixture-of-Experts with Language Priors Routing

2024-08-21 · Hao Zhou, Zhijun Wang, ShuJian Huang, Xin Huang 외

Large Language Models (LLMs) are often English-centric due to the disproportionate distribution of languages in their pre-training data. Enhancing non-English language capabilities through post-pretraining often results …

Mixture-of-Experts

WPG-MoE: Weak-Prior-Guided Dense Mixture-of-Experts for User-Level Social Media Depression Detection

2026-07-05 · Xian Li, Yuanhe Tian, Yang Yang, Guoqing Wang 외 arxiv

Online social media posts provide scalable signals for early depression screening, and recent studies mainly improve pre-classification evidence through risk-post selection, symptom grounding, and clinically informed fea…