This paper presents a new deep neural network (DNN)-based speech enhancement algorithm by integrating the distilled knowledge from the traditional statistical-based method. Unlike the other DNN-based methods, which usually train many different models on the same data and then average their predictions, or use a large number of noise types to enlarge the simulated noisy speech, the proposed method does not train a whole ensemble of models and does not require a mass of simulated noisy speech. It first trains a discriminator network and a generator network simultaneously using the adversarial learning method. Then, the discriminator network and generator network are re-trained by distilling knowledge from the statistical method, which is inspired by the knowledge distillation in a neural network. Finally, the generator network is fine-tuned using real noisy speech. Experiments on CHiME4 data sets demonstrate that the proposed method achieves a more robust performance than the compared DNN-based method in terms of perceptual speech quality.

The technique of extracting different distinguishing features by locating different part regions to achieve fine-grained visual classification (FGVC) has made significant improvements. Utilizing attention mechanisms for feature extraction has become one of the mainstream methods in computer vision, but these methods have certain limitations. They typically focus on the most discriminative regions and directly combine the features of these parts, neglecting other less prominent yet still discriminative regions. Additionally, these methods may not fully explore the intrinsic connections between higher-order and lower-order features to optimize model classification performance. By considering the potential relationships between different higher-order feature representations in the object image, we can enable the integrated higher-order features to contribute more significantly to the model’s classification decision-making capabilities. To this end, we propose a saliency feature suppression and cross-feature fusion network model (SFSCF-Net) to explore the interaction learning between different higher-order feature representations. These include (1) an object-level image generator (OIG): the intersection of the output feature maps of the last two convolutional blocks of the backbone network is used as an object mask and mapped to the original image for cropping to obtain an object-level image, which can effectively reduce the interference caused by complex backgrounds. (2) A saliency feature suppression module (SFSM): the most distinguishing part of the object image is obtained by a feature extractor, and the part is masked by a two-dimensional suppression method, which improves the accuracy of feature suppression. (3) A cross-feature fusion method (CFM) based on inter-layer interaction: the output feature maps of different network layers are interactively integrated to obtain high-dimensional features, and then the high-dimensional features are channel compressed to obtain the inter-layer interaction feature representation, which enriches the output feature semantic information. The proposed SFSCF-Net can be trained end-to-end and achieves state-of-the-art or competitive results on four FGVC benchmark datasets.

Japan's largest platform for academic e-journals: J-STAGE is a full text database for reviewed academic papers published by Japanese societies

Convolutional neural networks (CNNS) have a strong ability to understand and judge images. However, the enormous parameters and computation of CNNS have limited its application in resource-limited devices. In this letter, we used the idea of parameter sharing and dense connection to compress the parameters in the convolution kernel channel direction, thus greatly reducing the number of model parameters. On this basis, we designed Shared and Dense Channel-wise Convolutional Networks (SDChannelNets), mainly composed of Depth-wise Separable SD-Channel-wise Convolution layer. The advantage of SDChannelNets is that the number of model parameters is greatly reduced without or with little loss of accuracy. We also introduced a hyperparameter that can effectively balance the number of parameters and the accuracy of a model. We evaluated the model proposed by us through two popular image recognition tasks (CIFAR-10 and CIFAR-100). The results showed that SDChannelNets had similar accuracy to other CNNs, but the number of parameters was greatly reduced.