Research map / Label-efficient and robust learning / Domain adaptation and generalization
Multi-Source Domain Adaptation for Visual Sentiment Classification
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 MVAE and GAN losses to baseline multi-encoder GAN (ablation) | MSGAN · FI target | accuracy (%): 65.71 -> 67.2 | Helped |
| add MVAE and GAN losses to baseline multi-encoder GAN (ablation) | MSGAN · ArtPhoto target | accuracy (%): 58.02 -> 60.99 | Helped |
| add cycle-consistency loss to MVAE+GAN (ablation) | MSGAN · FI target | accuracy (%): 67.2 -> 69.03 | Helped |
| add cycle-consistency loss to MVAE+GAN (ablation) | MSGAN · ArtPhoto target | accuracy (%): 60.99 -> 62.16 | Helped |
| add emotional semantic consistency loss to MVAE+GAN (ablation) | MSGAN · FI target | accuracy (%): 67.2 -> 68.76 | Helped |
| add emotional semantic consistency loss to MVAE+GAN (ablation) | MSGAN · ArtPhoto target | accuracy (%): 60.99 -> 60.7 | Mixed |
| add ESC loss to MVAE+GAN+CC (ablation) | MSGAN · FI target | accuracy (%): 69.03 -> 70.63 | Helped |
| add ESC loss to MVAE+GAN+CC (ablation) | MSGAN · ArtPhoto target | accuracy (%): 62.16 -> 63.58 | Helped |
| use MSGAN instead of source-combined CycleEmotionGAN | multi-source DA · FI target | accuracy (%): 67.78 -> 70.63 | Helped |
| use MSGAN instead of source-combined CycleEmotionGAN | multi-source DA · Twitter II target | accuracy (%): 72.72 -> 72.72 | No clear effect |
| combine multiple sources into one and apply single-source DA instead of single-best source | single-source UDA · Twitter I target (CycleGAN) | accuracy (%): 61.59 -> 59.84 | Hurt |
| vary weights on KL and NLL terms (ablation) | MSGAN · FI and ArtPhoto | accuracy (%) | Helped |
Closest papers by meaning
- Adversarial Multiple Source Domain Adaptation (NeurIPS 2018 · similarity 0.89)
- Adversarial Training Based Multi-Source Unsupervised Domain Adaptation for Sentiment Analysis (AAAI 2020 · similarity 0.89)
- Improving Domain-Adapted Sentiment Classification by Deep Adversarial Mutual Learning (AAAI 2020 · similarity 0.89)
- CycleEmotionGAN: Emotional Semantic Consistency Preserved CycleGAN for Adapting Image Emotions (AAAI 2019 · similarity 0.87)
- Bridge Then Begin Anew: Generating Target-Relevant Intermediate Model for Source-Free Visual Emotion Adaptation (AAAI 2025 · similarity 0.84)
- Style Adaptation and Uncertainty Estimation for Multi-Source Blended-Target Domain Adaptation (NeurIPS 2024 · 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.