Topic
Hybrid neural network
About: Hybrid neural network is a research topic. Over the lifetime, 1305 publications have been published within this topic receiving 18223 citations.
Papers published on a yearly basis
Papers
More filters
01 Aug 2004
TL;DR: In this article, an advanced hybrid neural network (AHNN) is integrated with an automotive drivetrain model for system simulations to accurately predict the dynamic behaviors of transmission friction components over a broad operating range.
Abstract: In this research, the advanced hybrid neural network (AHNN) friction-component model, presented in Part 1 of this two-part paper, is integrated with an automotive drivetrain model for system simulations. The AHNN model accurately predicts the dynamic behaviours of transmission friction components over a broad operating range. It also allows variable sampling time steps in a numerical integration process. In this investigation, the AHNN model is trained using experimental data obtained from a powertrain dynamometer test stand. Since typical dynamometer measurements are acquired at locations away from friction components, a backtracking algorithm is developed to evaluate friction component torque during engagement. The trained AHNN model, together with a comprehensive drivetrain model, is implemented to simulate the shifting process of an automatic transmission system under various operating conditions, including different oil-temperature and engine-throttle levels. Simulation results demonstrate that the AHNN friction component model can be effectively utilized as a part of the drivetrain model to accurately predict transmission shift dynamics.
8 citations
TL;DR: Experiments gave strong evidence that in both simple and complex networks the introduction of a hyperbolic structure results in an improvement of the model accuracy, and the combined use ofhyperbolic and Euclidean layers showed superior performance in almost all the classification tasks.
Abstract: Prior approaches for multimodal sentiment and emotion recognition (SER) exploit input data representations and neural networks based on the classical Euclidean geometry. Recently, however, the hyperbolic metric proved to be a powerful tool for data mapping, being able to capture the hierarchical structure of the relations among elements in the data. In this paper we propose the use of hyperbolic learning for SER, and show that the inclusion in the neural network of hyperbolic structures mapping the input into the hyperbolic space can improve the quality of the predictions. The benefits brought by the hyperbolic features are evaluated by developing extensions of existing methods following two approaches. From one side, we modified state-of-the-art models by including hyperbolic output layers. From the other, we generated hybrid neural network architectures by combining hyperbolic and Euclidean layers according to different schemes. The proposed hyperbolic models were tested on several classification tasks applied to benchmark multimodal SER datasets. Experiments gave strong evidence that in both simple and complex networks the introduction of a hyperbolic structure results in an improvement of the model accuracy. Specifically, the combined use of hyperbolic and Euclidean layers showed superior performance in almost all the classification tasks.
8 citations
09 May 1995
TL;DR: An integrated hybrid neural network and hidden Markov model (HMM) classifier that combines the time normalization property of the HMM classifier with the superior discriminative ability of the neural net (NN) is presented.
Abstract: Presents an integrated hybrid neural network and hidden Markov model (HMM) classifier that combines the time normalization property of the HMM classifier with the superior discriminative ability of the neural net (NN). Sonar signals display a strong time varying characteristic. Although the neural net has been successful in classifying transient like sonar signals, the success is achieved either by using a bigger net architecture or by incorporating a detection mechanism in the classification procedure. The present authors propose an integrated hybrid HMM and neural net classifier where a left-to-right HMM module is used first. The HMM module segments the observation sequence belonging to every exemplar into a fixed number of states starting from the left. After this segmentation, all the frames belonging to the same state are replaced by one average frame. Thus, every exemplar, irrespective of its time scale variation, is transformed into a fixed number of frames, i.e., a static pattern. The multilayer perceptron (MLP) neural net is then used as the classifier for these time normalized exemplars. For successful modeling and classification, each frame is succinctly represented by a feature vector. Two feature extraction schemes are considered-the first one is based on the FFT power spectral coefficient, and the second one is based on the quadrature mirror filter (QMF) bank based subband decomposition. Finally, some experimental results are provided to demonstrate the superiority of the hybrid integrated classifier.
8 citations
29 Sep 2004
TL;DR: Simulations performed on a bilingual dictionary show the improvements in terms of phoneme accuracy of the method against the approach that uses a single neural network for multilingual TTP.
Abstract: Text-to-phoneme (TTP) mapping is a preliminary step in text-to-speech synthesis and it affects the naturalness and understandability of synthetic speech In this paper, we propose a hybrid neural network/rule based system for bilingual text-to-phoneme mapping Our system uses three neural networks and a simple rule to perform the phoneme transcription The first network is trained to convert the letters from the first language into their corresponding phonemes, the second one is used to obtain the phonemes for the second language whereas the third neural network together with a simple rule is responsible of the language recognition The proposed approach can be easily extended for multilingual applications when more neural networks are introduced Simulations performed on a bilingual dictionary (English+French) show the improvements in terms of phoneme accuracy of our method against the approach that uses a single neural network for multilingual TTP
8 citations
01 Jan 1994
TL;DR: The WISARD (Wilkie, Stonham, Aleksander Recognition Device) is an implementation in hardware or software of an n-tuple sampling technique and results suggest that being comparable to the above mentioned three systems, training time has been significantly reduced.
Abstract: Clinical electromyography (EMG) provides useful information for the diagnosis of neuromuscular disorders. The utility of artificial neural networks trained with the backpropagation, the Kohonen's self-organizing feature maps algorithm, and the genetics based machine learning (GBML) in classifying EMG data has been demonstrated. A hybrid diagnostic system was also introduced that combines the above neural network and GBML models. In this paper the WISARD net is applied on the same set of EMG data. The WISARD (Wilkie, Stonham, Aleksander Recognition Device) is an implementation in hardware or software of an n-tuple sampling technique. Results suggest that although the diagnostic performance of the WISARD models is of the order of 80%, that being comparable to the above mentioned three systems, training time has been significantly reduced. In addition, the hardware or software implementation of the WISARD net is simpler than the other three systems. >
8 citations