Charles C. Tappert

The state of the art in online handwriting recognition

This survey describes the state of the art of online handwriting recognition during a period of renewed activity in the field. It is based on an extensive review of the literature, including journal articles, conference proceedings, and patents. Online versus offline recognition, digitizer technology, and handwriting properties and recognition problems are discussed. Shape recognition algorithms, preprocessing and postprocessing techniques, experimental systems, and commercial products are examined. >

Cursive script recognition by elastic matching

Dynamic programming has been found useful for performing nonlinear time warping for matching patterns in automatic speech recognition. Here, this technique is applied to the problem of recognizing cursive script. The parameters used in the matching are derived from time sequences of x-y coordinate data of words handwritten on an electronic tablet. Chosen for their properties of invariance with respect to size and translation of the writing, these parameters are found particularly suitable for the elastic matching technique. A salient feature of the recognition system is the establishment, in a training procedure, of prototypes by each writer using the system. In this manner, the system is tailored to the user. Processing is performed on a word-by-word basis after the writing is separated into words. Using prototypes for each letter, the matching procedure allows any letter to follow any letter and finds the letter sequence which best fits the unknown word. A major advantage of this procedure is that it combines letter segmentation and recognition in one operation by, in essence, evaluating recognition at all possible segmentations, thus avoiding the usual segmentation-then-recognition philosophy. Results on cursive writing are presented where the alphabet is restricted to the lower-case letters. Letter recognition accuracy is over 95 percent for each of three writers.

Keystroke Biometric Recognition Studies on Long-Text Input under Ideal and Application-Oriented Conditions

A long-text-input keystroke biometric system was developed for applications such as identifying perpetrators of inappropriate e-mail or fraudulent Internet activity. A Java applet collected raw keystroke data over the Internet, appropriate long-text-input features were extracted, and a pattern classifier made identification decisions. Experiments were conducted on a total of 118 subjects using two input modes - copy and free-text input - and two keyboard types - desktop and laptop keyboards. Results indicate that the keystroke biometric can accurately identify an individual who sends inappropriate email (free text) if sufficient enrollment samples are available and if the same type of keyboard is used to produce the enrollment and questioned samples. For laptop keyboards we obtained 99.5% accuracy on 36 users, which decreased to 97.9% on a larger population of 47 users. For desktop keyboards we obtained 98.3% accuracy on 36 users, which decreased to 93.3% on a larger population of 93 users. Accuracy decreases significantly when subjects used different keyboard types or different input modes for enrollment and testing.

A Genetic Algorithm for Constructing Compact Binary Decision Trees

Tree-based classifiers are important in pattern recognitio n and have been well studied. Although the problem of finding an optimal decision tree has r eceived attention, it is a hard optimization problem. Here we propose utilizing a genetic algorithm to improve on the finding of compact, near-optimal decision trees. We present a method to encode and decode a decision tree to and from a chromosome where genetic operators such as mutation and crossover can be applied. Theoretical properties of decisi on trees, encoded chromosomes, and fitness functions are presented.

Enhancing Binary Feature Vector Similarity Measures

Similarity and dissimilarity measures play an important role in pattern classification and clustering. For a century, researchers have searched for a good measure. Here, we review, categorize, and evaluate various binary vector similarity / dissimilarity measures. One of the most contentious disputes in the similarity measure selection problem is whether the measure includes or excludes negative matches. While inner-product based similarity measures consider only positive matches, other conventional measures credit both positive and negative matches equally. Hence, we propose an enhanced similarity measure that gives variable credits and show that it is superior to conventional measures in IRIS biometric authentication and offline handwritten character recognition applications. Finally, the proposed similarity measure can be further boosted by applying weights and we demonstrate that it outperforms the weighted Hamming distance.

Memory and time improvements in a dynamic programming algorithm for matching speech patterns

Recently, dynamic programming has been found useful for performing nonlinear time warping in speech recognition. Although considerably faster than exhaustive search procedures, the dynamic programming procedure nevertheless requires substantial computation. Also, considerable storage is normally required for reference prototypes necessary in the matching process. This paper is concerned with methods for reducing this storage and computation. Empirical results indicate that one method yields 50 to 60 percent storage reduction and a factor of 4 to 6 in computational savings relative to conventional dynamic programming procedures without degradation in recognition accuracy.

An investigation of keystroke and stylometry traits for authenticating online test takers

The 2008 federal Higher Education Opportunity Act requires institutions of higher learning to make greater access control efforts for the purposes of assuring that students of record are those actually accessing the systems and taking exams in online courses by adopting identification technologies as they become more ubiquitous. To meet these needs, keystroke and stylometry biometrics were investigated towards developing a robust system to authenticate (verify) online test takers. Performance statistics on keystroke, stylometry, and combined keystroke-stylometry systems were obtained on data from 40 test-taking students enrolled in a university course. The best equal-error-rate performance on the keystroke system was 0.5% which is an improvement over earlier reported results on this system. The performance of the stylometry system, however, was rather poor and did not boost the performance of the keystroke system, indicating that stylometry is not suitable for text lengths of short-answer tests unless the features can be substantially improved, at least for the method employed.

On the individuality of the iris biometric

Biometric authentication has been considered a model for quantitatively establishing the discriminative power of biometric data. The dichotomy model classifies two biometric samples as coming either from the same person or from two different people. This paper reviews features, distance measures, and classifiers used in iris authentication. For feature extraction we compare simple binary and multi-level 2D wavelet features. For distance measures we examine scalar distances such as Hamming and Euclidean, feature vector and histogram distances. Finally, for the classifiers we compare Bayes decision rule, nearest neighbor, artificial neural network, and support vector machines. Of the eleven different combinations tested, the best one uses multi-level 2D wavelet features, the histogram distance, and a support vector machine classifier.

Behavioral biometric verification of student identity in online course assessment and authentication of authors in literary works

Keystroke and stylometry behavioral biometrics were investigated with the objective of developing a robust system to authenticate students taking online examinations. This work responds to the 2008 U.S. Higher Education Opportunity Act that requires institutions of higher learning undertake greater access control efforts, by adopting identification technologies as they become available, to assure that students of record are those actually accessing the systems and taking the exams in online courses. Performance statistics on keystroke, stylometry, and combined keystroke-stylometry systems were obtained on data from 30 students taking examinations in a university course. The performance of the keystroke system was 99.96% and 100.00%, while that of the stylometry system was considerably weaker at 74% and 78%, on test input of 500 and 1000 words, respectively. To further investigate the stylometry system, a separate study on 30 book authors achieved performance of 88.2% and 91.5% on samples of 5000 and 10000 words, respectively, and the varied performance over the population of authors was analyzed.

Developing a Keystroke Biometric System for Continual Authentication of Computer Users

Data windows of keyboard input are analyzed to continually authenticate computer users and verify that they are the authorized ones. Because the focus is on fast intruder detection, the authentication process operates on short bursts of roughly a minute of keystroke input, while the training process can be extensive and use hours of input. The biometric system consists of components for data capture, feature extraction, authentication classification, and receiver-operating-characteristic curve generation. Using keystroke data from 120 users, system performance was obtained as a function of two independent variables: the user population size and the number of keystrokes per sample. For each population size, the performance increased (and the equal error rate decreased) roughly logarithmically as the number of keystrokes per sample was increased. The best closed-system performance results of 99 percent on 14 participants and 96 percent on 30 participants indicate the potential of this approach.