cd /news/artificial-intelligence/deo-training-free-direct-embedding-o… · home › topics › artificial-intelligence › article
[ARTICLE · art-173075] src=machinebrief.com ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

DEO: Training-Free Direct Embedding Optimization for Negation-Aware Retrieval

A training-free method called Direct Embedding Optimization (DEO) improves negation- and exclusion-aware retrieval without additional training data or model updates, according to an arXiv paper (2603.09185v2). DEO decomposes queries into positive and negative components and optimizes the query embedding with a contrastive objective; on the NegConstraint benchmark it beats baselines by +0.0738 nDCG@10 and +0.1028 MAP@100, and lifts Recall@5 by +6% over OpenAI CLIP in multimodal retrieval. The authors present DEO as a practical alternative to embedding adaptation or fine-tuning, which add computational cost and deployment complexity for retrieval-augmented generation systems.

recommendation

For negation and exclusion queries, decompose the query into positive and negative components and optimize the query embedding with a contrastive objective at retrieval time instead of fine-tuning; the paper reports +0.0738 nDCG@10 on NegConstraint.

more in playbook: Retrieval and RAG
by read1 min views1 publishedNov 20, 2026

arXiv:2603.09185v2 Announce Type: replace Abstract: Recent advances in Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) have enabled diverse retrieval methods. However, existing retrieval methods often fail to accurately retrieve results for negation and exclusion queries. To address this limitation, prior approaches rely on embedding adaptation or fine-tuning, which introduce additional computational cost and deployment complexity. We propose Direct Embedding Optimization (DEO), a training-free method for negation-aware text and multimodal retrieval. DEO decomposes queries into positive and negative components and optimizes the query embedding with a contrastive objective. Without additional training data or model updates, DEO outperforms baselines on NegConstraint, with gains of +0.0738 nDCG@10 and +0.1028 MAP@100, while improving Recall@5 by +6% over OpenAI CLIP in multimodal retrieval. These results demonstrate the practicality of DEO for negation- and exclusion-aware retrieval in real-world settings.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @direct embedding optimization 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/deo-training-free-di…] indexed:0 read:1min 2026-11-20 · —