In this paper we propose a two-stage duration model using neural networks for predicting the duration of syllables in Indian languages. The proposed model consists of three feedforward neural networks for predicting the duration of syllable in specific intervals and a syllable classifier, which has to predict the probability that a given syllable falls into an interval. Autoassociative neural network models and support vector machines are explored for syllable classification. Syllable duration prediction and analysis is performed on broadcast news data in Hindi, Telugu and Tamil. The input to the neural network consists of a set of phonological, positional and contextual features extracted from the text. From the studies it is found that about 80% of the syllable durations are predicted within a deviation of 25%. The performance of the duration model is evaluated using objective measures such as mean absolute error (μ), standard deviation (σ) and correlation coefficient (γ).
artificial neural networks yegnanarayana pdf download
About The Book Artificial Neural Networks Book Summary:Designed as an introductory level textbook on Artificial Neural Networks at the postgraduate and senior undergraduate levels in any branch of engineering, this self-contained and well-organized book highlights the need for new models of computing based on the fundamental principles of neural networks.
Professor Yegnanarayana compresses, into the covers of a single volume, his several years of rich experience, in teaching and research in the areas of speech processing, image processing, artificial intelligence and neural networks. He gives a masterly analysis of such topics as Basics of artificial neural networks, Functional units of artificial neural networks for pattern recognition tasks, Feedforward and Feedback neural networks, and Archi-tectures for complex pattern recognition tasks. Throughout, the emphasis is on the pattern processing feature of the neural networks. Besides, the presentation of real-world applications provides a practical thrust to the discussion.
The fairly large number of diagrams, the detailed Bibliography, and the provision of Review Questions and Problems at the end of each chapter should prove to be of considerable assistance to the reader. Besides students, practising engineers and research scientists would cherish this book which treats the emerging and exciting area of artificial neural networks in a rigorous yet lucid fashion.
Optical coherence tomography (OCT) images semi-transparent tissues noninvasively. Relying on backscatter and interferometry to calculate spatial relationships, OCT shares similarities with other pulse-echo modalities. There is considerable interest in using machine learning techniques for automated image classification, particularly among ophthalmologists who rely heavily on diagnostic OCT. Artificial neural networks (ANN) consist of interconnected nodes and can be employed as classifiers after training on large datasets. Conventionally, OCT scans are rendered as 2D or 3D human-readable images of which the smallest depth-resolved unit is the amplitude-scan reflectivity-function profile which is difficult for humans to interpret. We set out to determine whether amplitude-scan reflectivity-function profiles representing disease signatures could be distinguished and classified by a feed-forward ANN. Our classifier achieved high accuracies after training on only 24 eyes, with evidence of good generalization on unseen data. The repertoire of our classifier can now be expanded to include rare and unseen diseases and can be extended to other disciplines and industries.
An artificial neural network (ANN) is a computational construct consisting of layers of interconnected virtual nodes2,3,4. Each node mimics the behavior of a biological neuron by way of a transfer function that maps an input signal to an output response. Artificial neural networks can be configured to perform information-processing functions and have been shown to have advantages over deterministic techniques for various classification tasks. Artificial neural networks are trained on source data rather than being programmed using parameterized rules. Because training is incremental, building a classifier which generalizes well over unseen data typically requires large datasets.
Image classification is frequently approached using a convolutional neural network, a type of ANN in which the configuration of nodes and connections (topology) is inspired by the receptive fields and hierarchical processing understood to occur in sensory systems in living creatures7,8,9. Convolutional neural networks have found roles in many applications such as computer vision and speech and language recognition and abstraction, and are also employed in medical image classification. Gargeya and Leng have used a convolutional neural network to perform binary classification of color fundus photographs as normal or consistent with diabetic retinopathy, with sensitivity and specificity of 94% and 98%, respectively10. Their dataset consisted of 75,137 expert-labeled fundus photographs representing all stages of diabetic retinopathy, including absent retinopathy. Lee and associates recently published results of binary classification of OCT images as either normal or representing age-related macular degeneration, using a database of approximately 50,000 labeled images in each class, and reporting sensitivity and specificities around 93%11. Although the investigators in each of these studies were able to perform secondary analyses to identify the features in the images which made the images diagnostic, each study was confined to its respective overall disease of interest.
K.K.D., K.K.V., S.T.D., and J.C. drafted the manuscript. J.C. provided anonymized OCT data on all patients from his clinical practice. K.K.D. labeled the amplitude-scans. K.K.V., S.T.D. and S.J. designed and implemented the artificial neural network. K.B.F. provided valuable consultation on optimal OCT imaging and performed critical review and editing of the manuscript. 2ff7e9595c
Comments