Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• IMSE achieves state-of-the-art performance on various distribution-shift benchmarks under the TTA setting.result 90%
• IMSE improves accuracy by 3.4 percentage points in CTTA and 2.4 percentage points in Gradual CTTA compared to previous methods.result 90%
• IMSE requires 385 times fewer trainable parameters than previous methods.technical 90%
• Entropy minimization in TTA often induces feature collapse, causing models to rely on domain-specific features rather than class-discriminative features.limitation 85%
• IMSE adapts only the singular values while keeping singular vectors fixed during fine-tuning.method 90%
• IMSE introduces a diversity maximization loss based on expert-input alignment to encourage diverse utilization of spectral experts.method 90%
• IMSE introduces Domain-Aware Spectral Code Retrieval to detect domain shifts and retrieve adapted singular values for rapid adaptation.method 90%
• Fully leveraging rich representations of large pretrained models with minimal parameter updates remains underexplored in TTA.technical 80%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card