A Hybrid Classifier for Handwriting Recognition on Multi-domain Financial Bills Based on DCNN and SVM

Abstract

With the rapid growth of the global economy, the automatic recognition of financial bills becomes the primary way to reduce the burden of the traditional manual approach for bill recognition and classification. However, most automatic recognition methods cannot effectively recognize the handwritten characters on financial bills, especially when the bills come from different financial companies. To solve the problem, this paper fully explores the bill system in banks and the operations of bill number recognition, and then develops a hybrid classifier based on deep convolutional neural network (DCNN) and support vector machine (SVM), with the aim to recognize the handwritten numbers on financial bills in different domains. The DCNN with different channels was adopted to effectively mine the local handwritten numbers on financial bills from varied sources. Then, the extracted information was fed to the SVM to realize accurate classification of numbers. Our method makes full use of the distribution difference between information in different fields, and adapts to different fields based on the parameter sharing mechanism. Experimental results show that our method can recognize the handwritten numbers on financial bills more accurately (>3%) than benchmark methods.

Typical artificial intelligence algorithms and real-world applications related to handwritten number classifier

“…The handwritten number classifier can be used in the field of finance and taxation [9]. To clarify, with the economic recovery after the COVID-19 pandemic, more and more financial and tax statements, checks, and invoices are waiting to be processed every day.…”

Research on Default Prediction for Credit Card Users Based on XGBoost-LSTM Model

“…where b, U, and W represent the deviation, input weight, and cycle weight of the LSTM unit, respectively. On the right side of formula (14), the first half is the cell state information controlled by the forget gate, and the second half is the input information controlled by the input gate [16].…”

“…e LSTM [15] algorithm performs well in time-series data mining and has had many applications in the financial field [16][17][18][19], such as customer service marketing, risk control, and trading strategy. In the field of antifraud, for example, deep learning technology automatically recognizes fraudulent transactions from massive amounts of transaction data, realizes successful interception, and blocks fraudulent transactions, thereby improving system effectiveness, reducing the rate of false alarms, and reducing compliance risks [20].…”