Paper Title
Journal Entries with Deep Learning Model

Abstract
Deep learning is the most recent approach to achieve Artificial intelligence. Especially neural networks are used for solving many human problems - from repetitive operations to intelligent recognizing in image, sound and text processing. They are used in medicine, car industry, game industry and robotics. Business companies also try to find the way of exploitation of the latest technology despite the fact that it is the long way to the point where machines will be capable to replace the human intelligence. Authors of this paper explore possibilities of semi-supervised learning application in accounting. One of the latest deep learning algorithm is successfully used to reconstruct the journal entry key columns. The model was trained and tested on a real-world dataset so it could become base for developing the wide pallet of accounting and audit applications - as anomaly detection module of Enterprise Resource Planning (ERP) software or as a standalone application. Index Terms - General ledger, journal entry, bookkeeping, accounting, deep learning, variational autoencoder, anomaly detection, accounting control system.