Scientists have long dreamed of creating machines humans could interact with by voice. Although one no longer believes Turing's prophecy that machines will be able to converse like humans in the near future, real progress has been made in the voice and text-based human-machine interaction. This paper is a light introduction and survey of some deployed natural language systems and technologies and their historical evolution. We review two fundamental problems involving natural language: the language prediction problem and the language understanding problem. While describing in detail all these technologies is beyond our scope, we do comment on some aspects less discussed in the literature such as language prediction using huge models and semantic labeling using Marcus contextual grammars.

This paper aims to quantify the main error types the 2004 BBN speech recognition system made in the broadcast news (BN) and conversational telephone speech (CTS) DARPA EARS evaluations. We show that many of the remaining errors occur in clusters rather than isolated, have specific causes, and differ to some extent between the BN and CTS domains. The correctly recognized words are also clustered and are highly correlated with regions where the system produces a single hypothesized choice per word. A statistical analysis of some well-known error causes (out-of-vocabulary words, word fragments, hesitations, and unlikely language constructs) was performed in order to assess their contribution to the overall word error rate (WER). We conclude with a discussion of the lower bound on the WER introduced by the human annotator disagreement

We report on the results of the first evaluations for the BBN/LIMSI system under the new DARPA EARS program. The evaluations were carried out for conversational telephone speech (CTS) and broadcast news (BN) for three languages: English, Mandarin, and Arabic. In addition to providing system descriptions and evaluation results, the paper highlights methods that worked well across the two domains and those few that worked well on one domain but not the other. For the BN evaluations, which had to be run under 10 times real-time, we demonstrated that a joint BBN/LIMSI system with a time constraint achieved better results than either system alone.

Although Arabic is currently one of the most widely spoken languages in the world, there has been relatively little speech recognition research on Arabic compared to other languages. Moreover, most previous work has concentrated on the recognition of formal rather than dialectal Arabic. This paper reports on our project at the 2002 Johns Hopkins Summer Workshop, which focused on the recognition of dialectal Arabic. Three problems were addressed: (a) the lack of short vowels and other pronunciation information in Arabic texts; (b) the morphological complexity of Arabic; and (c) the discrepancies between dialectal and formal Arabic. We present novel approaches to automatic vowel restoration, morphology-based language modeling and the integration of out-of-corpus language model data, and report significant word error rate improvements on the LDC Arabic CallHome task.

A procedure for automated 2D shape model design is presented. The system is given a set of training example shapes defined by contour point coordinates. The shapes are automatically aligned using Procrustes analysis and clustered to obtain cluster prototypes (typical objects) and statistical information about intracluster shape variation. One difference from previous methods is that the training set is first automatically clustered and shapes considered to be outliers are discarded. In this way, cluster prototypes are not distorted by outliers. A second difference is in the manner in which registered sets of points are extracted from each shape contour. We propose a flexible point matching technique that takes into account both pose/scale differences and nonlinear shape differences. The matching method is independent of the objects' initial relative position/scale and does not require any manually tuned parameters. Our shape model design method was used to learn 11 different shapes from contours that were manually traced in MR brain images. The resulting model was then employed to segment several MR brain images that were not included in the shape-training set. A quantitative analysis of our shape registration approach, within the main cluster of each structure, demonstrated results that compare very well to those achieved by manual registration; achieving an average registration error of about 1 pixel. Our approach can serve as a fully automated substitute to the tedious and time-consuming manual 2D shape registration and analysis.

General deformable models have reduced the need for hand crafting new models for every new problem, but still most of the general models rely on manual interaction by an expert, when applied to a new problem, e.g. for selecting parameters and initialization. We propose a full and unified scheme for applying the general deformable template model proposed by (Grenander et al., 1991) to a new problem with minimal manual interaction, beside supplying a training set, which can be done by a non-expert user. The main contributions compared to previous work are a supervised learning scheme for the model parameters, a very fast general initialization algorithm and an adaptive likelihood model based on local means. The model parameters are trained by a combination of a 2D shape learning algorithm and a maximum likelihood based criteria. The fast initialization algorithm is based on a search approach using a filter interpretation of the likelihood model.

The goal of this study is to detect the main road network in high resolution, panchromatic SPOT satellite images. We describe an automatic procedure for road detection which has the following advantages over the previous approaches: 1) it does not require manual initialization; 2) it is able to detect some of the secondary roads in addition to the main highways; and 3) the detection time is small (/spl sim/3 min) even on large images.

A new fully automated shape learning method is presented. It is based on clustering a set of training shapes in the original shape space (defined by the coordinates of the contour points) and performing a Procrustes analysis on each cluster to obtain cluster prototypes and information about shape variation. The main difference from previously reported methods is that the training set is first automatically clustered and those shapes considered to be outliers are discarded. The second difference is in the manner in which registered sets of points are extracted from each shape contour. As a direct application of our shape learning method, an 11-structure shape model of brain substructures was extracted from MR image data, an eigen-shape model was automatically trained, and employed to segment several MR brain images not present in the shape-training set. A quantitative analysis of our shape registration approach, within the main cluster of each structure, shows that our results compare very well to those achieved by manual registration; achieving an average rms error of about 1 pixel. Our approach can serve as a fully automated substitute to the tedious and time-consuming manual shape registration and analysis.

Magnetic resonance imaging (MRI) of the brain, followed by automated segmentation of the corpus callosum (CC) in midsagittal sections has important applications in neurology and neurocognitive research since the size and shape of the CC are shown to be correlated to sex, age, neurodegenerative diseases and various lateralized behavior in man. Moreover, whole head, multispectral 3D MRI recordings enable voxel-based tissue classification and estimation of total brain volumes, in addition to CC morphometric parameters. We propose a new algorithm that uses both multispectral MRI measurements (intensity values) and prior information about shape (CC template) to segment CC in midsagittal slices with very little user interaction. The algorithm has been successfully tested on a sample of 10 subjects scanned with multispectral 3D MRI, collected for a study of dyslexia. We conclude that the proposed method for CC segmentation is promising for clinical use when multispectral MR images are recorded.

An automated method for left ventricle detection in MR cardiac images is presented. Ventricle detection is the first step in a fully automated segmentation system used to compute volumetric information about the heart. Our method is based on learning the gray level appearance of the ventricle by maximizing the discrimination between positive and negative examples in a training set. The main differences from previously reported methods are feature definition and solution to the optimization problem involved in the learning process. Our method was trained on a set of 1,350 MR cardiac images from which 101,250 positive examples and 123,096 negative examples were generated. The detection results on a test set of 887 different images demonstrate an excellent performance: 98% detection rate, a false alarm rate of 0.05% of the number of windows analyzed (10 false alarms per image) and a detection time of 2 seconds per 256/spl times/256 image on a Sun Ultra 10 for an 8-scale search. The false alarms ore eventually eliminated by a position/scale consistency check along all the images that represent the same anatomical slice.

We present a method for personal authentication based on deformable matching of hand shapes. Authentication systems are already employed in domains that require some sort of user verification. Unlike previous methods on hand shape based verification, our method aligns the hand shapes before extracting a feature set. We also base the verification decision on the shape distance which is automatically computed during the alignment stage. The shape distance proves to be a more reliable classification criterion than the handcrafted feature sets used by previous systems. Our verification system attained a high level of accuracy: 96.5% genuine accept rate vs. false accept rate. This performance is further improved by learning an enrolment template shape for each user.

Presents a learning approach for the face detection problem. The problem can be stated as follows: given an arbitrary black and white, still image, find the location and size of every human face it contains. Numerous applications of automatic face detection have attracted considerable interest in this problem, but no present face detection system is completely satisfactory from the point of view of detection rate, false alarm rate and detection time. We describe an inductive learning-based detection method that produces a maximally specific hypothesis consistent with the training data. Three different sets of features were considered for defining the concept of a human face. The performance achieved is as follows: 85% detection rate, a false alarm rate of 0.04% of the number of windows analyzed and 1 minute detection table for a 320/spl times/240 image on a Sun Ultrasparc 1.

This paper reports a novel method for fully automated segmentation that is based on description of shape and its variation using point distribution models (PDM's). An improvement of the active shape procedure introduced by Cootes and Taylor to find new examples of previously learned shapes using PDM's is presented. The new method for segmentation and interpretation of deep neuroanatomic structures such as thalamus, putamen, ventricular system, etc. incorporates a priori knowledge about shapes of the neuroanatomic structures to provide their robust segmentation and labeling in magnetic resonance (MR) brain images. The method was trained in eight MR brain images and tested in 19 brain images by comparison to observer-defined independent standards. Neuroanatomic structures in all testing images were successfully identified. Computer-identified and observer-defined neuroanatomic structures agreed well. The average labeling error was 7%+/-3%. Border positioning errors were quite small, with the average border positioning error of 0.8+/-0.1 pixels in 256 x 256 MR images. The presented method was specifically developed for segmentation of neuroanatomic structures in MR brain images. However, it is generally applicable to virtually any task involving deformable shape analysis.