Learning deep character recognition code -Hinton classic
no vote
Application background Character recognition can be applied in many fields, such as reading, translation, literature retrieval, letters and parcels sorting, manuscript editing and proofreading, a large number of statistical statements and cards the collection and analysis, a statistical summary of the bank check processing, commercial invoice, commodity code identification, goods warehouse management, and water, electricity, gas, rent, insurance and other personal expenses of collection of business of a large number of credit card of the automatic processing and a typist in an office work of local automation and so on. As well as document retrieval, all kinds of documents recognition, convenient user to enter information quickly, improve the work efficiency of all walks of life. Key Technology Data can be converted to low-dimensional codes by training a multilayer neural High-dimensionalWith a small central layer to reconstruct high-dimensional i