Handwriting Recognition with Novelty

This paper introduces an agent-centric approach to handle novelty in the visual recognition domain of handwriting (HWR). An ideal transcription agent would rival or surpass human perception, being able recognize known and new characters image, detect any stylistic changes that may occur within across documents. A key confound is presence novelty, which has continued stymie even best machine learning-based algorithms for these tasks. In handwritten documents, can be a change writer, character attributes, writing overall document appearance, among other things. Instead looking at each aspect independently, we suggest integrated process novelties simultaneously better strategy. formalizes with describes baseline agent, evaluation protocol benchmark data, provides experimentation set state-of-the-art. Results show feasibility approach, but more work needed human-levels reading ability, giving HWR community formal basis build upon as they solve this challenging problem.