Overview

Feature fusion studies how to combine representations—such as image and radar signals, multiple encoder layers, or learned embeddings and handcrafted descriptors—so models can use complementary evidence. Common approaches include concatenation, cross-attention, gated or reliability-aware weighting, and multi-scale or frequency-domain integration, applied in tasks from detection and segmentation to retrieval and medical prediction. Results show that fusion can improve accuracy and robustness, especially when inputs provide distinct useful cues. A central challenge is deciding when and how to fuse: misaligned or redundant features can add complexity without benefit, and the best strategy depends on the task, data, and reliability of each representation.

Latest papers

CardsList

126 papers13 papers in the last four weeks, down 28% on the four weeks before. 0.1% of all new papers.

  1. KEMP-PIP: A Feature-Fusion Based Approach for Pro-inflammatory Peptide Prediction

    Feb 22, 2026Soumik Deb Niloy, Md. Fahmid-Ul-Alam Juboraj, Swakkhar ShatabdaMolecular Property PredictionFeature Fusion

  2. ASPEN: Spectral-Temporal Fusion for Cross-Subject Brain Decoding

    Feb 18, 2026Megan Lee, Seung Ha Hwang, Inhyeok Choi +3Cross-Subject EEG DecodingBrain-Computer Interfaces

  3. GaussianCaR: Gaussian Splatting for Efficient Camera-Radar Fusion

    Feb 9, 2026Santiago Montiel-Marín, Miguel Antunes-García, Fabio Sánchez-García +3Camera-Radar FusionAutonomous Driving Perception

  4. MGDFIS: Multi-scale Global-detail Feature Integration Strategy for Small Object Detection

    Jun 15, 2025Yuxiang Wang, Xuecheng Bai, Chuanzhi Xu +2Remote Sensing Small Object DetectionSmall Object Detection