{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/implicit-language-model-in-lstm-for-ocr","title":"Implicit Language Model in LSTM for OCR","arxiv_id":"1805.09441","date":"2018-05-23","proceeding":null,"authors":["Ekraam Sabir","Stephen Rawls","Prem Natarajan"],"abstract":"Neural networks have become the technique of choice for OCR, but many aspects\nof how and why they deliver superior performance are still unknown. One key\ndifference between current neural network techniques using LSTMs and the\nprevious state-of-the-art HMM systems is that HMM systems have a strong\nindependence assumption. In comparison LSTMs have no explicit constraints on\nthe amount of context that can be considered during decoding. In this paper we\nshow that they learn an implicit LM and attempt to characterize the strength of\nthe LM in terms of equivalent n-gram context. We show that this implicitly\nlearned language model provides a 2.4\\% CER improvement on our synthetic test\nset when compared against a test set of random characters (i.e. not naturally\noccurring sequences), and that the LSTM learns to use up to 5 characters of\ncontext (which is roughly 88 frames in our configuration). We believe that this\nis the first ever attempt at characterizing the strength of the implicit LM in\nLSTM based OCR systems.","url_abs":"http://arxiv.org/abs/1805.09441v1","url_pdf":"http://arxiv.org/pdf/1805.09441v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"implicit-language-model-in-lstm-for-ocr","repo_url":"https://github.com/amirabbasasadi/PersianOCR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"optical-character-recognition","task_name":"Optical Character Recognition (OCR)"},{"task_slug":"model","task_name":"model"}],"methods":[{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}