Research map / Label-efficient and robust learning / Domain adaptation and generalization
MemSAC: Memory Augmented Sample Consistency for Large Scale Domain Adaptation
Tarun Kalluri, Astuti Sharma, Manmohan Chandraker
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 sample consistency loss Lsc to CDAN (ablation) | MemSAC · CUB-Drawings average | accuracy: 60.98 -> 67.95 | Improved |
| add adversarial loss Ladv to Lsc only (ablation) | MemSAC · CUB-Drawings average | accuracy: 52.79 -> 67.95 | Improved |
| add Lsc on top of source-only training without adversarial loss (ablation) | MemSAC · CUB-Drawings average | accuracy: 51.47 -> 52.79 | Improved |
| use temperature 0.07 instead of 1.0 (ablation) | MemSAC · CUB-Drawings average | accuracy: 60.91 -> 67.95 | Improved |
| use temperature 0.07 instead of 0.007 (ablation) | MemSAC · CUB-Drawings average | accuracy: 64.23 -> 67.95 | Improved |
| use cosine similarity instead of inverse Euclidean (ablation) | MemSAC · CUB-Drawings average | accuracy: 64.23 -> 67.95 | Improved |
| use cosine similarity instead of Gaussian (ablation) | MemSAC · CUB-Drawings average | accuracy: 60.47 -> 67.95 | Improved |
| add MemSAC to DANN | DANN · DomainNet-345 R->C and C->R | accuracy: 46.39 -> 49.74 (+3.35%) | Improved |
| add MemSAC to CAN | CAN · DomainNet-345 R->C and C->R | accuracy: 47.82 -> 49.57 (+1.75%) | Improved |
| add MemSAC to CDAN | CDAN · DomainNet-345 R->C and C->R | accuracy: 50.08 -> 54.37 (+4.29%) | Improved |
| add MemSAC with ResNet-101 backbone versus CDAN | CDAN · DomainNet-345 R->C and C->R (ResNet-101) | accuracy: 49.55 -> 54.88 | Improved |
| use a memory bank (size above batch size 32) instead of none (ablation) | MemSAC · CUB-Drawings C->D and D->C | accuracy: no numeric delta in text | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FixBi — FixBi: Bridging Domain Spaces for Unsupervised Domain Adaptation (CVPR 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Instance Level Affinity-Based Transfer for Unsupervised Domain Adaptation (CVPR 2021 · similarity 0.89)
- A Prototype-Oriented Framework for Unsupervised Domain Adaptation (NeurIPS 2021 · similarity 0.89)
- Prototypical Partial Optimal Transport for Universal Domain Adaptation (AAAI 2023 · similarity 0.88)
- Memory-Assisted Sub-Prototype Mining for Universal Domain Adaptation (ICLR 2024 · similarity 0.88)
- Theoretical Performance Guarantees for Partial Domain Adaptation via Partial Optimal Transport (ICML 2025 · similarity 0.87)
- Gradient-Based Sample Selection for Black-Box Universal Domain Adaptation (AAAI 2025 · similarity 0.87)
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.
