Abstract Description and classification of face images is a significant task of computer vision, machine learning and pattern recognition communities. In the past, researchers have made tremendous efforts in this task. Previous researchers always seek high-resolution face images for better image classification. However, with this paper, we present and demonstrate a new opinion that in some cases the

Abstract Reinforcement learning (RL) is a double-edged sword: it frees the human trainer from having to provide voluminous supervised training data or from even knowing a solution. On the other hand, a common complaint about RL is that learning is slow. Deep Q-learning (DQN), a somewhat recent development, has allowed practitioners and scientists to solve tasks previously thought unsolvable by a reinforcement

Abstract Noisy and nonstationary real-world time series predictions (TSPs) are challenging tasks. Confronted with these challenging tasks, the predictive power of traditional shallow models is commonly not satisfactory enough. While the research on deep learning (DL) has made milestone breakthrough in recent years, and DL paradigm has gradually become indispensable for accomplishing these complex tasks

Abstract Emotion recognition has attracted great interest. Numerous emotion recognition approaches have been proposed, most of which focus on visual, acoustic or psychophysiological information individually. Although more recent research has considered multimodal approaches, individual modalities are often combined only by simple fusion or are directly fused with deep learning networks at the feature

Abstract Recently, rain streaks removal from a single image has attracted much research attention to alleviate the degenerated performance of computer vision tasks implemented on rainy images. In this paper, we provide a thorough review for current single-image-based rain removal techniques, which can be mainly categorized into three classes: early filter-based, conventional prior-based, and recent

Abstract Deep hashing has proven to be efficient and effective for large-scale image retrieval due to the strong representation capability of deep networks. Existing deep hashing methods only utilize a single deep hash table. In order to achieve both higher retrieval recall and precision, longer hash codes can be used but at the expense of higher space usage. To address this issue, a novel deep hashing

Abstract Poor road conditions, such as potholes, are a nuisance to society, which would annoy passengers, damage vehicles, and even cause accidents. Thus, detecting potholes is an important step toward pavement maintenance and rehabilitation to improve road conditions. Potholes have different shapes, scales, shadows, and illumination effects, and highly complicated backgrounds can be involved. Therefore

Abstract Deep learning (henceforth DL) has become most powerful machine learning methodology. Under specific circumstances recognition rates even surpass those obtained by humans. Despite this, several works have shown that deep learning produces outputs that are very far from human responses when confronted with the same task. This the case of the so-called “adversarial examples” (henceforth AE).

Abstract The material property of an object’s surface is critical for the tasks of robotic manipulation or interaction with its surrounding environment. Tactile sensing can provide rich information about the material characteristics of an object’s surface. Hence, it is important to convey and interpret tactile information of material properties to the users during interaction. In this paper, we propose

Abstract Transcription factors are cis-regulatory molecules that bind to specific sub-regions of DNA promoters and initiate transcription, the process that regulates the conversion of genetic information from DNA to RNA. Several computational methods have been developed to predict DNA–protein binding sites in DNA sequence using convolutional neural network (CNN). However, these techniques could indicate

Abstract Deep learning based facial expression recognition becomes more successful in many applications. However, the lack of labeled data is still a bottleneck for better recognition performance. Thus, it is of practical significance to exploit the rich unlabeled data for training deep neural networks (DNNs). In this paper, we propose a novel discriminative deep association learning (DDAL) framework

Abstract Due to the discovery of innovative and practical neural architectures, deep learning has achieved bright successes in many fields, such as computer vision, natural language processing, recommendation systems, etc. To reach high performance, researchers have to adjust neural architectures and choose training tricks very carefully. The manual trial-and-error process for discovering the best

Abstract User basic data (e.g. user gender, user age and user ID, etc.) and item basic data (e.g. item name, item category, etc.) are important side data that can be used to enhance the performance of recommendation algorithms, whereas attempts concerning this issue are still relatively scarce. In this study, a deep neural networks based recommendation algorithm is proposed where user average rating

Abstract Flow-based generative models (flow models) are conceptually attractive due to tractability of the exact log-likelihood and the exact latent-variable inference. In order to generate sharper images and extend the Gaussian prior of Flow models to other discrete forms, we propose an adversarial non-volume preserving flow model with Boltzmann priors (ANVP) for modeling complex high-dimensional

Abstract Learning both hierarchical and temporal dependencies can be crucial for recurrent neural networks (RNNs) to deeply understand sequences. To this end, a unified RNN framework is required that can ease the learning of both the deep hierarchical and temporal structures by allowing gradients to propagate back from both ends without being vanished. The residual learning (RL) has appeared as an

Abstract Ensemble methods have shown to improve the results of statistical classifiers by combining multiple single learners into a strong one. In this paper, we explore the use of ensemble methods at the level of the objective function of a deep neural network. We propose a novel objective function that is a linear combination of single losses and integrate the proposed objective function into a deep

Abstract The robustness to outliers, noises, and corruptions has been paid more attention recently to increase the performance in linear feature extraction and image classification. As one of the most effective subspace learning methods, low-rank representation (LRR) can improve the robustness of an algorithm by exploring the global representative structure information among the samples. However, the

Abstract The computational detection of lung lobes from computed tomography images is a challenging segmentation problem with important respiratory healthcare applications, including emphysema, chronic bronchitis, and asthma. This paper proposes a progressive random forest-based random walk approach for interactive semi-automated pulmonary lobes segmentation. First, our model performs automated segmentation

Abstract In medical diagnosis, how to select an optimal medicine from some medicines with similar efficacy values to treat diseases has become common problems between doctors and patients. To solve this problem, we describe it as a multi-attribute decision-making (MADM) in a finite fuzzy covering approximation space. This paper aims to propose two pairs of covering-based variable precision fuzzy rough

Abstract Dismantling of complex networks is a problem to find a minimal set of nodes in which the removal leaves the network broken into connected components of sub-extensive size. It has a wide spectrum of important applications, including network immunization and network destruction. Due to its NP-hard computational complexity, this problem cannot be solved exactly with polynomial time. Traditional

Abstract Optimal scale selection is an important issue in the study of multi-scale decision tables. Most existing optimal scale selection methods have been designed from the perspective of consistency or uncertainty, and cost as well as user requirements or preferences in practical applications has not been considered. It is well known that the uncertainty of decision making in different levels of