An intelligent land vehicle utilizes onboard sensors to acquire observed states at a disorderly intersection. However, partial observation of the environment occurs due to sensor noise. This causes decision failure easily. A collision relationship-based driving behavior decision-making method via deep recurrent Q network (CR-DRQN) is proposed for intelligent land vehicles. First, the collision relationship between the intelligent land vehicle and surrounding vehicles is designed as the input. The collision relationship is extracted from the observed states with the sensor noise. This avoids a CR-DRQN dimension explosion and speeds up the network training. Then, DRQN is utilized to attenuate the impact of the input noise and achieve driving behavior decision-making. Finally, some comparative experiments are conducted to verify the effectiveness of the proposed method. CR-DRQN maintains a high decision success rate at a disorderly intersection with partially observable states. In addition, the proposed method is outstanding in the aspects of safety, the ability of collision risk prediction, and comfort.
Retinal vessel segmentation benefits significantly from deep learning. Its performance relies on sufficient training images with accurate ground-truth segmentation, which are usually manually annotated in the form of binary pixel-wise label maps. Manually annotated ground-truth label maps, more or less, contain errors for part of the pixels. Due to the thin structure of retina vessels, such errors are more frequent and serious in manual annotations, which negatively affect deep learning performance.
In this paper, we develop a new method to automatically and iteratively identify and correct such noisy segmentation labels in the process of network training. We consider historical predicted label maps of network-in-training from different epochs and jointly use them to self-supervise the predicted labels during training and dynamically correct the supervised labels with noises.
We conducted experiments on the three datasets of DRIVE, STARE and CHASE-DB1 with synthetic noises, pseudo-labeled noises, and manually labeled noises. For synthetic noise, the proposed method corrects the original noisy label maps to a more accurate label map by 4.0–$$9.8\%$$
on PR on three testing datasets. For the other two types of noise, the method could also improve the label map quality.
Experiment results verified that the proposed method could achieve better retinal image segmentation performance than many existing methods by simultaneously correcting the noise in the initial label map.
Aiming at recognizing small proportion, blurred and complex traffic sign in natural scenes, a traffic sign detection method based on RetinaNet-NeXt is proposed. First, to ensure the quality of dataset, the data were cleaned and enhanced to denoise. Secondly, a novel backbone network ResNeXt was employed to improve the detection accuracy and effection of RetinaNet. Finally, transfer learning and group normalization were adopted to accelerate our network training. Experimental results show that the precision, recall and mAP of our method, compared with the original RetinaNet, are improved by 9.08%, 9.09% and 7.32%, respectively. Our method can be effectively applied to traffic sign detection.
The availability of evidence-based therapies for abstract words is limited. Abstract Semantic Associative Network Training (AbSANT) is theoretically motivated and has been shown to not only improve directly trained abstract words, such as the word
in the category
but also promote generalization to related concrete words, such as the word
This tutorial provides step-by-step instructions, including cueing strategies, and material resources for conducting AbSANT. Importantly, this tutorial also explains the theoretical motivation behind AbSANT, as well as information regarding the population, dose, and environment characteristics of effective trials, to help clinicians make informed decisions regarding the applicability of this approach and to guide decision-making throughout the steps of therapy.
AbSANT is an effective, theoretically based treatment for abstract words. This tutorial provides all of the resources needed to conduct AbSANT with clients with aphasia.
Explicit feedback and implicit feedback are two important types of heterogeneous data for constructing a recommendation system. The combination of the two can effectively improve the performance of the recommendation system. However, most of the current deep learning recommendation models fail to fully exploit the complementary advantages of two types of data combined and usually only use binary implicit feedback data. Thus, this paper proposes a neural matrix factorization recommendation algorithm (EINMF) based on explicit-implicit feedback. First, neural network is used to learn nonlinear feature of explicit-implicit feedback of user-item interaction. Second, combined with the traditional matrix factorization, explicit feedback is used to accurately reflect the explicit preference and the potential preferences of users to build a recommendation model; a new loss function is designed based on explicit-implicit feedback to obtain the best parameters through the neural network training to predict the preference of users for items; finally, according to prediction results, personalized recommendation list is pushed to the user. The feasibility, validity, and robustness are fully demonstrated in comparison with multiple baseline models on two real datasets.
One of the most common metrics to evaluate neural network classiﬁers is the area under the receiver operating characteristic curve (AUC). However, optimisation of the AUC as the loss function during network training is not a standard procedure. Here we compare minimising the cross-entropy (CE) loss and optimising the AUC directly. In particular, we analyse the loss function landscape (LFL) of approximate AUC (appAUC) loss functions to discover the organisation of this solution space. We discuss various surrogates for AUC approximation and show their differences. We ﬁnd that the characteristics of the appAUC landscape are signiﬁcantly different from the CE landscape. The approximate AUC loss function improves testing AUC, and the appAUC landscape has substantially more minima, but these minima are less robust, with larger average Hessian eigenvalues. We provide a theoretical foundation to explain these results. To generalise our results, we lastly provide an overview of how the LFL can help to guide loss function analysis and selection.
The sedimentation problem is one of the critical issues affecting the long-term use of rivers, and the study of sediment variation in rivers is closely related to water resource, river ecosystem and estuarine delta siltation. Traditional research on sediment variation in rivers is mostly based on field measurements and experimental simulations, which requires a large amount of human and material resources, many influencing factors and other restrictions. With the development of computer technology, intelligent approaches have been applied to hydrological models to establish small information in river areas. In this paper, considering the influence of multiple factors on sediment transport, the validity of predicting sediment transport combined with wavelet transforms and neural network was analyzed. The rainfall and runoff cycles are extracted and decomposed into time series sub-signals by wavelet transforms; then, the data post-processing is used as the neural network training set to predict the sediment model. The results show that wavelet coupled neural network model effectively improves the accuracy of the predicted sediment model, which can provide a reference basis for river sediment prediction.
Recent progress in novel non-volatile memory-based synaptic device technologies and their feasibility for matrix-vector multiplication (MVM) has ignited active research on implementing analog neural network training accelerators with resistive crosspoint arrays. While significant performance boost as well as area- and power-efficiency is theoretically predicted, the realization of such analog accelerators is largely limited by non-ideal switching characteristics of crosspoint elements. One of the most performance-limiting non-idealities is the conductance update asymmetry which is known to distort the actual weight change values away from the calculation by error back-propagation and, therefore, significantly deteriorates the neural network training performance. To address this issue by an algorithmic remedy, Tiki-Taka algorithm was proposed and shown to be effective for neural network training with asymmetric devices. However, a systematic analysis to reveal the required asymmetry specification to guarantee the neural network performance has been unexplored. Here, we quantitatively analyze the impact of update asymmetry on the neural network training performance when trained with Tiki-Taka algorithm by exploring the space of asymmetry and hyper-parameters and measuring the classification accuracy. We discover that the update asymmetry level of the auxiliary array affects the way the optimizer takes the importance of previous gradients, whereas that of main array affects the frequency of accepting those gradients. We propose a novel calibration method to find the optimal operating point in terms of device and network parameters. By searching over the hyper-parameter space of Tiki-Taka algorithm using interpolation and Gaussian filtering, we find the optimal hyper-parameters efficiently and reveal the optimal range of asymmetry, namely the asymmetry specification. Finally, we show that the analysis and calibration method be applicable to spiking neural networks.
Due to refraction, absorption, and scattering of light by suspended particles in water, underwater images are characterized by low contrast, blurred details, and color distortion. In this paper, a fusion algorithm to restore and enhance underwater images is proposed. It consists of a color restoration module, an end-to-end defogging module and a brightness equalization module. In the color restoration module, a color balance algorithm based on CIE Lab color model is proposed to alleviate the effect of color deviation in underwater images. In the end-to-end defogging module, one end is the input image and the other end is the output image. A CNN network is proposed to connect these two ends and to improve the contrast of the underwater images. In the CNN network, a sub-network is used to reduce the depth of the network that needs to be designed to obtain the same features. Several depth separable convolutions are used to reduce the amount of calculation parameters required during network training. The basic attention module is introduced to highlight some important areas in the image. In order to improve the defogging network’s ability to extract overall information, a cross-layer connection and pooling pyramid module are added. In the brightness equalization module, a contrast limited adaptive histogram equalization method is used to coordinate the overall brightness. The proposed fusion algorithm for underwater image restoration and enhancement is verified by experiments and comparison with previous deep learning models and traditional methods. Comparison results show that the color correction and detail enhancement by the proposed method are superior.
AbstractWith the vigorous development of mobile Internet technology and the popularization of smart devices, while the amount of multimedia data has exploded, its forms have become more and more diversified. People’s demand for information is no longer satisfied with single-modal data retrieval, and cross-modal retrieval has become a research hotspot in recent years. Due to the strong feature learning ability of deep learning, cross-modal deep hashing has been extensively studied. However, the similarity of different modalities is difficult to measure directly because of the different distribution and representation of cross-modal. Therefore, it is urgent to eliminate the modal gap and improve retrieval accuracy. Some previous research work has introduced GANs in cross-modal hashing to reduce semantic differences between different modalities. However, most of the existing GAN-based cross-modal hashing methods have some issues such as network training is unstable and gradient disappears, which affect the elimination of modal differences. To solve this issue, this paper proposed a novel Semantic-guided Autoencoder Adversarial Hashing method for cross-modal retrieval (SAAH). First of all, two kinds of adversarial autoencoder networks, under the guidance of semantic multi-labels, maximize the semantic relevance of instances and maintain the immutability of cross-modal. Secondly, under the supervision of semantics, the adversarial module guides the feature learning process and maintains the modality relations. In addition, to maintain the inter-modal correlation of all similar pairs, this paper use two types of loss functions to maintain the similarity. To verify the effectiveness of our proposed method, sufficient experiments were conducted on three widely used cross-modal datasets (MIRFLICKR, NUS-WIDE and MS COCO), and compared with several representatives advanced cross-modal retrieval methods, SAAH achieved leading retrieval performance.