Multilayer neural networks trained with the back-propagation algorithm constitute the best example of a successful gradient based learning technique. Given an appropriate network architecture, gradient-based learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns, such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional neural networks, which are specifically designed to deal with the variability of 2D shapes, are shown to outperform all other techniques. Real-life document recognition systems are composed of multiple modules including field extraction, segmentation recognition, and language modeling. A new learning paradigm, called graph transformer networks (GTN), allows such multimodule systems to be trained globally using gradient-based methods so as to minimize an overall performance measure. Two systems for online handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of graph transformer networks. A graph transformer network for reading a bank cheque is also described. It uses convolutional neural network character recognizers combined with global training techniques to provide record accuracy on business and personal cheques. It is deployed commercially and reads several million cheques per day.

Gradient-based learning applied to document recognition

Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

The primary goal of pattern recognition is supervised or unsupervised classification. Among the various frameworks in which pattern recognition has been traditionally formulated, the statistical approach has been most intensively studied and used in practice. More recently, neural network techniques and methods imported from statistical learning theory have been receiving increasing attention. The design of a recognition system requires careful attention to the following issues: definition of pattern classes, sensing environment, pattern representation, feature extraction and selection, cluster analysis, classifier design and learning, selection of training and test samples, and performance evaluation. In spite of almost 50 years of research and development in this field, the general problem of recognizing complex patterns with arbitrary orientation, location, and scale remains unsolved. New and emerging applications, such as data mining, web searching, retrieval of multimedia data, face recognition, and cursive handwriting recognition, require robust and efficient pattern recognition techniques. The objective of this review paper is to summarize and compare some of the well-known methods used in various stages of a pattern recognition system and identify research topics and applications which are at the forefront of this exciting and challenging field.

/pdf/statistical-pattern-recognition-a-review-1kz1b0tte4.pdf

Statistical pattern recognition: a review

We develop a common theoretical framework for combining classifiers which use distinct pattern representations and show that many existing schemes can be considered as special cases of compound classification where all the pattern representations are used jointly to make a decision. An experimental comparison of various classifier combination schemes demonstrates that the combination rule developed under the most restrictive assumptions-the sum rule-outperforms other classifier combinations schemes. A sensitivity analysis of the various schemes to estimation errors is carried out to show that this finding can be justified theoretically.

On combining classifiers

From the Publisher:
Pattern recognition has long been studied in relation to many different (and mainly unrelated) applications, such as remote sensing, computer vision, space research, and medical imaging. In this book Professor Ripley brings together two crucial ideas in pattern recognition; statistical methods and machine learning via neural networks. Unifying principles are brought to the fore, and the author gives an overview of the state of the subject. Many examples are included to illustrate real problems in pattern recognition and how to overcome them.This is a self-contained account, ideal both as an introduction for non-specialists readers, and also as a handbook for the more expert reader.

Pattern recognition and neural networks

A method for comparing thinning algorithms involving a series of experiments with human subjects is presented. Several statistical tests are reported to analyze the preference structure exhibited by the data. It is concluded that humans are coherent in comparing thinning algorithm outputs and that reference skeletons can be useful to facilitate the evaluation of thinning algorithms. An evaluation protocol is then proposed and applied to the design and evaluation of a new thinning algorithm for handwritten character recognition. >

Validation of preprocessing algorithms: A methodology and its application to the design of a thinning algorithm for handwritten characters

This paper presents the history of the Persian (Farsi) script, as well as the development of different writing styles for the current Persian script. It also addresses the Arabic alphabet adopted and evolved for writing the Persian language as well as different writing styles. This evolution includes further extensions to the Arabic alphabet and altered shapes of some Arabic letters. The differences between current Arabic and Persian handwritings for automated recognition purposes are discussed. A short review on techniques in Persian script recognition is presented, and the shortcomings and challenges in this area are highlighted.

A review on Persian script and recognition techniques

To support large vocabulary handwriting recognition in standard computer platforms, a fast algorithm for hidden Markov model alignment is necessary. To address this problem, we propose a non-heuristic fast decoding algorithm which is based on hidden Markov model representation of characters. The decoding algorithm breaks up the computation of word likelihoods into two levels: state level and character level. Given an observation sequence, the two level decoding enables the reuse of character likelihoods to decode all words in the lexicon, avoiding repeated computation of state sequences. In an 80,000-word recognition task, the proposed decoding algorithm is about 15 times faster than a conventional Viterbi algorithm, while maintaining the same recognition accuracy.

/pdf/fast-two-level-hmm-decoding-algorithm-for-large-vocabulary-yzg891k87b.pdf

Fast two-level HMM decoding algorithm for large vocabulary handwriting recognition

In document recognition, it is often important to obtain high accuracy or reliability and to reject patterns that cannot be classified with high confidence. This is the case for applications such as the processing of financial documents in which errors can be very costly and therefore far less tolerable than rejections. This paper presents a new approach based on Linear Discriminant Analysis (LDA) to reject less reliable classifier outputs. To implement the rejection, which can be considered a two-class problem of accepting the classification result or otherwise, an LDA-based measurement is used to determine a new rejection threshold. This measurement (LDAM) is designed to take into consideration the confidence values of the classifier outputs and the relations between them, and it represents a more comprehensive measurement than traditional rejection measurements such as First Rank Measurement and First Two Ranks Measurement. Experiments are conducted on the CENPARMI database of numerals, the CENPARMI Arabic Isolated Numerals Database, and the numerals in the NIST Special Database 19. The results show that LDAM is more effective, and it can achieve a higher reliability while maintaining a high recognition rate on these databases of very different origins and sizes.

Rejection measurement based on linear discriminant analysis for document recognition

This book constitutes the refereed proceedings of the 6th IAPR TC3 International Workshop on Artificial Neural Networks in Pattern Recognition, ANNPR 2014, held in Montreal, QC, Canada, in October 2014. The 24 revised full papers presented were carefully reviewed and selected from 37 submissions for inclusion in this volume. They cover a large range of topics in the field of learning algorithms and architectures and discussing the latest research, results, and ideas in these areas.

Ching Y. Suen

Papers

Validation of preprocessing algorithms: A methodology and its application to the design of a thinning algorithm for handwritten characters

A review on Persian script and recognition techniques

Fast two-level HMM decoding algorithm for large vocabulary handwriting recognition

Rejection measurement based on linear discriminant analysis for document recognition

Artificial Neural Networks in Pattern Recognition: 6th IAPR TC 3 International Workshop, ANNPR 2014, Montreal, QC, Canada, October 6-8, 2014