Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• Light-ResKAN achieves 99.09% accuracy on the MSTAR dataset for SAR image recognition.result 95%
• Light-ResKAN reduces FLOPs by 82.90× compared to VGG16 when processing MSTAR images resized to 1024×1024.result 95%
• Light-ResKAN reduces parameters by 163.78× compared to VGG16.result 95%
• Light-ResKAN uses Gram Polynomials as activations to capture complex non-linear relationships in SAR data.method 90%
• Light-ResKAN employs a parameter-sharing strategy where each kernel shares parameters per channel to reduce parameters and FLOPs while preserving unique features.method 90%
• Light-ResKAN modifies ResNet by replacing convolutions with KAN convolutions to enable adaptive feature extraction for SAR images.method 90%
• Light-ResKAN achieves 93.01% accuracy on the FUSAR-Ship dataset.result 95%
• Light-ResKAN achieves 97.26% accuracy on the SAR-ACD dataset.result 95%

Competitive landscape

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

Startup potential card