Research map / Label-efficient and robust learning / Recognition and long-tail
Large-Scale Heterogeneous Feature Embedding
Xiao Huang, Qingquan Song, Fan Yang, Xia Hu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add network G to feature matrix X (FeatWalk vs FeatWalk X) (ablation) | FeatWalk · Flickr, 100% training | micro-average: 0.868 -> 0.893 | Improved |
| add network G to feature matrix X (FeatWalk vs FeatWalk X) (ablation) | FeatWalk · ACM, 100% training | micro-average: 0.667 -> 0.751 | Improved |
| add network G to feature matrix X (FeatWalk vs FeatWalk X) (ablation) | FeatWalk · Yelp, 100% training | micro-average: 0.699 -> 0.701 | No clear effect |
| add second feature matrix X(2) (Spanish) to X(1) (Italian) (ablation) | FeatWalk · Reuters, 100% training | micro-average: 0.744 -> 0.771 | Improved |
| use Feature Walks instead of explicit similarity-matrix random walks (w/o FW) (ablation) | FeatWalk · ACM, Yelp | running time | Improved |
| add features to network embedding (FeatWalk vs DeepWalk) | DeepWalk · ACM, 100% training | micro-average: 0.684 -> 0.751 | Improved |
| FeatWalk vs AANE | AANE · Flickr, 100% training | micro-average: 0.885 -> 0.893 | Improved |
| FeatWalk vs AANE | AANE · Yelp, 100% training | micro-average: 0.714 -> 0.701 | Hurt |
| add multi-threading workers (1 to 2) (ablation) | FeatWalk · all datasets | running time | Improved |
| vary embedding dimension d from 20 to 180 (ablation) | FeatWalk · Flickr | micro-average | No clear effect |
| vary window size 1 to 10 (ablation) | FeatWalk · Flickr | micro-average | No clear effect |
| balance feature and network weight alpha (0.62 vs 0 or 1) (ablation) | FeatWalk · Flickr | micro-average | Improved |
Closest papers by meaning
- VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks (ICLR 2025 · similarity 0.85)
- NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval (ICLR 2025 · similarity 0.85)
- Repurposing Language Models into Embedding Models: Finding the Compute-Optimal Recipe (NeurIPS 2024 · similarity 0.85)
- EmbedLLM: Learning Compact Representations of Large Language Models (ICLR 2025 · similarity 0.84)
- NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models (ICLR 2025 · similarity 0.84)
- Generalization Principles for Inference over Text-Attributed Graphs with Large Language Models (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
