Comparison of Three Structured Prediction Algorithms on a Sequence Prediction Problem
I empirically compared three sequence prediction algorithms on a structured prediction problem, Handwritten OCR. I used existing frameworks for Structured SVM and Conditional Random Fields, and I self-implemented the Auto-context algorithm.
I varied the window size and performed the recognition on the full word level (using the sliding window strategy). I gave a thorough comparison to the methods I adopted in terms of training and testing errors w.r.t. different window sizes. I varied the number of training and testing samples using different splits including 1,000/4,000, 2,500/2,500, and 4,000/1,000.