Researchers have developed a novel framework for online handwritten character recognition that addresses the challenge of rotational deformations. The approach utilizes the Sliding Window Path Signature (SW-PS) to extract rotation-invariant local structural features and employs Linear Recurrent Units (LRU) as the classifier. This combination of SW-PS and LRU demonstrates superior performance in both convergence speed and accuracy compared to existing models on datasets of digits, English letters, and Chinese radicals. AI
IMPACT This research could improve the accuracy and robustness of character recognition systems, particularly in applications where input data may be subject to rotation.
RANK_REASON The cluster contains an academic paper detailing a new method for handwritten character recognition. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- CASIA-OLHWDB1.1
- Danyu Yang
- Linear Recurrent Units
- recurrent neural network
- Sliding Window Path Signature
- State Space Model
- SW-PS+LRU
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
