Search

Spiking Neural Network (SNN) has been recognized as the third generation of neural networks. Conventionally, a SNN can be converted from a pre-trained Artificial Neural Network (ANN) with less computation and memory than training from scratch. But, these converted SNNs are vulnerable to adversarial attacks. Numerical experiments demonstrate that the SNN trained by optimizing the loss function will be more adversarial robust, but the theoretical analysis for the mechanism of robustness is lacking. In this paper, we provide a theoretical explanation by analyzing the expected risk function. Starting by modeling the stochastic process introduced by the Poisson encoder, we prove that there is a positive semidefinite regularizer. Perhaps surprisingly, this regularizer can make the gradients of the output with respect to input closer to zero, thus resulting in inherent robustness against adversarial attacks. Extensive experiments on the CIFAR10 and CIFAR100 datasets support our point of view. For example, we find that the sum of squares of the gradients of the converted SNNs is 13 ∼ 160 times that of the trained SNNs. And, the smaller the sum of the squares of the gradients, the smaller the degradation of accuracy under adversarial attack. • The adversarial robustness of SNNs based on rate encoder is explained mathematically. • A regularizer distinguishes the directly trained SNN from the converted SNN. • The coding length can also change the gradient of the input to affect the robustness. • We compare more types of the trained SNN with the converted SNN. [ABSTRACT FROM AUTHOR]

Deep learning has accomplished huge success in computer vision applications such as self-driving vehicles, facial recognition, and controlling robots. A growing need for deploying systems on resource-limited or resource-constrained environments such as smart cameras, autonomous vehicles, robots, smartphones, and smart wearable devices drives one of the current mainstream developments of convolutional neural networks: reducing model complexity but maintaining fine accuracy. In this study, the proposed efficient light convolutional neural network (ELNet) comprises three convolutional modules which perform ELNet using fewer computations, which is able to be implemented in resource-constrained hardware equipment. The classification task using CIFAR-10 and CIFAR-100 datasets was used to verify the model performance. According to the experimental results, ELNet reached 92.3% and 69%, respectively, in CIFAR-10 and CIFAR-100 datasets; moreover, ELNet effectively lowered the computational complexity and parameters required in comparison with other CNN architectures. [ABSTRACT FROM AUTHOR]

In this paper, we investigate the effectiveness of applying deep neural network hidden Markov models, or DNN-HMMs, for acoustic modeling in the context of educational applications. Specifically, we focus on spoken responses from non-native and child speech that tend to show great acoustic variability. We perform comprehensive experiments to compare the performance between traditional Gaussian mixture model (GMM)-HMMs and DNN-HMMs in three large language assessment datasets that contain various spoken tasks, classified broadly as constrained and open-ended tasks. Our experimental results suggest useful conclusions that can help guide the design of real-life educational applications. DNN-HMMs outperform conventional GMM-HMMs by a large margin for all spoken tasks commonly used in spoken assessment applications. In our experiments, DNN-HMMs trained using 25 h of data can outperform GMM-HMMs trained with 6.7–9 times data. Specifically regarding overall performance, when all available training data were used (175, 227, 169 h respectively), we achieved a relative word error rate decrease of 20.4% for adult English and 29.3% for child English, and a relative character error rate decrease of 14.3% for adult Chinese, when switching from GMMs to DNNs. In comparing between types of tasks, we notice that the more challenging open-ended tasks benefit significantly more than constrained item types by the use of DNN-HMMs. For open-ended tasks, having large amounts of training data is the key, as DNN-HMMs can take full advantage of the added training data and further push performance. In contrast, the performance of constrained spoken tasks saturates at around 25 h of training data. At the same time, constrained spoken tasks require only a few hours of data (1 or 5 h) to build well-performing acoustic models. This is an encouraging observation, that indicates the potential to build reliable spoken assessment applications based on constrained tasks, when few domain specific training data are available. [ABSTRACT FROM AUTHOR]

Skeleton-based hand gesture recognition is an active research topic in computer graphics and computer vision and has a wide range of applications in VR/AR and robotics. Although the spatial–temporal graph convolutional network has been successfully used in skeleton-based hand gesture recognition, these works often use a fixed spatial graph according to the hand skeleton tree or use a fixed graph on the temporal dimension, which may not be optimal for hand gesture recognition. In this paper, we propose a two-stream graph attention convolutional network with spatial–temporal attention for hand gesture recognition. We adopt pose stream and motion stream as the two input streams for our network. In pose stream, we use the joint in each frame as the input; In motion stream, we use the joint offsets between neighboring frames as the input. We propose a new temporal graph attention module to model the temporal dependency and also use a spatial graph attention module to construct dynamic skeleton graph. For each stream, we adopt graph convolutional network with spatial–temporal attention to extract the features. Then, we concatenate the feature of the pose stream and motion stream for gesture recognition. We achieve the competitive performance on the main hand gesture recognition benchmark datasets, which demonstrates the effectiveness of our method. [ABSTRACT FROM AUTHOR]

• We propose a unified two-stage framework for automatic model compression. • We propose Dynamic BN to predict the performance of each compression policy. • We propose a solving algorithm to search for the compression ratio allocation. Deep Neural Networks have become ubiquitous in various domains. Meanwhile, the problems of massive storage and computation costs have hindered the deployment of these models to real-world applications. This paper proposes a novel and unified two-stage framework for automatic model compression. To determine the compression ratio of each layer, we improve the optimization from two aspects. First, to predict the performance of each compression policy, we propose Dynamic BN, which improves the correlation significantly with little computation overhead. Second, to search for the compression ratio allocation, we propose an efficient and hyperparameter-free solving algorithm based on the proposed Hessian matrix approximation and Knapsack problem reformulation. Moreover, comprehensive experiments and analyses are conducted on the CIFAR-100&ImageNet datasets and various network architectures to demonstrate its performance advantages over existing model compression methods under the quantization-only, pruning-only, and pruning-quantization settings. [ABSTRACT FROM AUTHOR]

Visualization from trained deep neural networks has drawn massive public attention in recent. One of the visualization approaches is to train images maximizing the activation of specific neurons. However, directly maximizing the activation would lead to unrecognizable images, which cannot provide any meaningful information. In this paper, we introduce a simple but effective technique to constrain the optimization route of the visualization. By adding two totally inverse transformations, image blurring and deblurring, to the optimization procedure, recognizable images can be created. Our algorithm is good at extracting the details in the images, which are usually filtered by previous methods in the visualizations. Extensive experiments on AlexNet, VGGNet and GoogLeNet illustrate that we can better understand the neural networks utilizing the knowledge obtained by the visualization. [ABSTRACT FROM AUTHOR]

Channel Pruning is one of the most widespread techniques used to compress deep neural networks while maintaining their performances. Currently, a typical pruning algorithm leverages neural architecture search to directly find networks with a configurable width, the key step of which is to identify representative subnet for various pruning ratios by training a supernet. However, current methods mainly follow a serial training strategy to optimize supernet, which is very time-consuming. In this work, we introduce PSE-Net, a novel parallel-subnets estimator for efficient channel pruning. Specifically, we propose a parallel-subnets training algorithm that simulate the forward–backward pass of multiple subnets by droping extraneous features on batch dimension, thus various subnets could be trained in one round. Our proposed algorithm facilitates the efficiency of supernet training and equips the network with the ability to interpolate the accuracy of unsampled subnets, enabling PSE-Net to effectively evaluate and rank the subnets. Over the trained supernet, we develop a prior-distributed-based sampling algorithm to boost the performance of classical evolutionary search. Such algorithm utilizes the prior information of supernet training phase to assist in the search of optimal subnets while tackling the challenge of discovering samples that satisfy resource constraints due to the long-tail distribution of network configuration. Extensive experiments demonstrate PSE-Net outperforms previous state-of-the-art channel pruning methods on the ImageNet dataset while retaining superior supernet training efficiency. For example, under 300M FLOPs constraint, our pruned MobileNetV2 achieves 75.2% Top-1 accuracy on ImageNet dataset, exceeding the original MobileNetV2 by 2.6 units while only cost 30%/16% times than BCNet/AutoAlim. • We propose PSE-Net, a novel parallel-subnets estimator for efficient channel pruning. • The prior-distributed-based sampling algorithm to facilitate subnets search. • 30%/16% total time for estimator training compared with BCNet/AutoSlim. • Exceeds the SOTAs on ImageNet and downstream tasks for a variety of FLOPs budgets. [ABSTRACT FROM AUTHOR]

The aim of this paper is to distinguish the vehicle detection and count the class number in each classification from the inputs. We proposed the use of Fuzzy Guided Scale Choice (FGSC)-based SSD deep neural network architecture for vehicle detection and class counting with parameter optimization. The 'FGSC' blocks are integrated into the convolutional layers of the model, which emphasize essential features while ignoring less important ones that are not significant for the operation. We created the passing detection lines and class counting windows and connected them with the proposed FGSC-SSD deep neural network model. The 'FGSC' blocks in the convolution layer emphasize essential features and find out unnecessary features by using the scale choice method at the training stage and eliminate that significant speedup of the model. In addition, FGSC blocks avoided many unusable parameters in the saturation interval and improved the performance efficiency. In addition, the Fuzzy Sigmoid Function (FSF) increases the activation interval through fuzzy logic. While performing operations, the FGSC-SSD model reduces the computational complexity of convolutional layers and their parameters. As a result, the model tested Frames Per Second (FPS) on edge artificial intelligence (AI) and reached a real-time processing speed of 38.4 and an accuracy rate of more than 94%. Therefore, this work might be considered an improvement to the traffic monitoring approach by using edge AI applications. [ABSTRACT FROM AUTHOR]