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.

A Hybrid Neural Network for Fast Automatic Modulation Classification

Abstract: Automatic modulation classification (AMC) plays a key role in cognitive radio. For AMC, convolutional neural networks (CNNs) have been explored in previous works extensively and deliver the best performance. However, temporal dependencies of signals modeled by CNNs are inherently implicit and insufficient. As a result, models need more data to learn discriminative features automatically. In this work, we propose a hybrid model named HybridNet, where a bidirectional gated recurrent unit (Bi-GRU) is placed after CNN to capture temporal dependencies explicitly. In addition, we investigate why varying Signal-to-Noise Ratio (SNR) dataset makes performance deteriorate. By visualization, we discover that the increase of the intra-class divergence under sharply varying SNR is the central cause. To this end, channel-wise attention is adopted in HybridNet to learn different patterns existing in SNR, which does not require SNR labels in the training process or inference values of SNR. On RadioML2016.10b, our HybridNet obtains the best accuracy among all scales of training data. Especially, in small datasets, our model obtains 87.4% accuracy that is 9.7% higher than the baseline method.

Forecasting wavelet neural hybrid network with financial ensemble empirical mode decomposition and MCID evaluation

Abstract: By considering the properties of nonlinear data and the impact of historical data, this paper combines ensemble empirical mode decomposition (EEMD) into wavelet neural network with random time effective (WNNRT) to establish a hybrid neural network prediction model to improve the prediction accuracy of energy prices The EEMD is a noise-aided data analyze method, since it can effectively suppress pattern confusion and restore signal essence. Different from traditional models, the random time effective function that considers the timeliness of historical data and the random change of market environment is applied to the wavelet neural network to establish the WNNRT model. Moreover, multiscale complexity invariant distance (MCID) is utilized to evaluate the predicting performance of EEMD-WNNRT model. Further, the proposed model which is tested in predicting the impact on the global energy prices has carried on the empirical research, and it has also proved the corresponding superiority.

Combination of two-dimensional cochleogram and spectrogram features for deep learning-based ASR

Abstract: This paper explores the use of auditory features based on cochleograms; two dimensional speech features derived from gammatone filters within the convolutional neural network (CNN) framework. Furthermore, we also propose various possibilities to combine cochleogram features with log-mel filter banks or spectrogram features. In particular, we combine within low and high levels of CNN framework which we refer to as low-level and high-level feature combination. As comparison, we also construct the similar configuration with deep neural network (DNN). Performance was evaluated in the framework of hybrid neural network - hidden Markov model (NN-HMM) system on TIMIT phoneme sequence recognition task. The results reveal that cochleogram-spectrogram feature combination provides significant advantages. The best accuracy was obtained by high-level combination of two dimensional cochleogram-spectrogram features using CNN, achieved up to 8.2% relative phoneme error rate (PER) reduction from CNN single features or 19.7% relative PER reduction from DNN single features.

Retrospective optimization of time-dependent fermentation control strategies using time-independent historical data.

Abstract: We have previously shown the usefulness of historical data for fermentation process optimization. The methodology developed includes identification of important process inputs, training of an artificial neural network (ANN) process model, and ultimately use of the ANN model with a genetic algorithm to find the optimal values of each critical process input. However, this approach ignores the time-dependent nature of the system, and therefore, does not fully utilize the available information within a database. In this work, we propose a method for incorporating time-dependent optimization into our previously developed three-step optimization routine. This is achieved by an additional step that uses a fermentation model (consisting of coupled ordinary differential equations (ODE)) to interpret important time-course features of the collected data through adjustments in model parameters. Important process variables not explicitly included in the model were then identified for each model parameter using automatic relevance determination (ARD) with Gaussian process (GP) models. The developed GP models were then combined with the fermentation model to form a hybrid neural network model that predicted the time-course activity of the cell and protein concentrations of novel fermentation conditions. A hybrid-genetic algorithm was then used in conjunction with the hybrid model to suggest optimal time-dependent control strategies. The presented method was implemented upon an E. coli fermentation database generated in our laboratory. Optimization of two different criteria (final protein yield and a simplified economic criteria) was attempted. While the overall protein yield was not increased using this methodology, we were successful in increasing a simplified economic criterion by 15% compared to what had been previously observed. These process conditions included using 35% less arabinose (the inducer) and 33% less typtone in the media and reducing the time required to reach the maximum protein concentration by 10% while producing approximately the same level of protein as the previous optimum.