Search

We have witnessed great success of Single Image Super-Resolution (SISR) with convolutional neural networks (CNNs) in recent years. However, most existing Super-Resolution (SR) networks fail to utilize the multi-scale features of low-resolution (LR) images to further improve the representation capability for more accurate SR. In addition, most of them do not exploit the hierarchical features across networks for the final reconstruction. In this paper, we propose a novel multi-scale feature fusion residual network (MSFFRN) to fully exploit image features for SISR. Based on the residual learning, we propose a multi-scale feature fusion residual block (MSFFRB) with multiple intertwined paths to adaptively detect and fuse image features at different scales. Furthermore, the outputs of each MSFFRB and the shallow features are used as the hierarchical features for global feature fusion. Finally, we recover the high-resolution image based on the fused global features. Extensive experiments on four standard benchmarks demonstrate that our MSFFRN achieves better accuracy and visually pleasing than the current state-of-the-art methods.

Automatic math problem solving has recently attracted increasing attention as a long-standing AI benchmark. In this paper, we focus on solving geometric problems, which requires a comprehensive understanding of textual descriptions, visual diagrams, and theorem knowledge. However, the existing methods were highly dependent on handcraft rules and were merely evaluated on small-scale datasets. Therefore, we propose a Geometric Question Answering dataset GeoQA, containing 4,998 geometric problems with corresponding annotated programs, which illustrate the solving process of the given problems. Compared with another publicly available dataset GeoS, GeoQA is 25 times larger, in which the program annotations can provide a practical testbed for future research on explicit and explainable numerical reasoning. Moreover, we introduce a Neural Geometric Solver (NGS) to address geometric problems by comprehensively parsing multimodal information and generating interpretable programs. We further add multiple self-supervised auxiliary tasks on NGS to enhance cross-modal semantic representation. Extensive experiments on GeoQA validate the effectiveness of our proposed NGS and auxiliary tasks. However, the results are still significantly lower than human performance, which leaves large room for future research. Our benchmark and code are released at https://github.com/chen-judge/GeoQA ., Comment: Accepted to Findings of ACL 2021

Text classification is one of the most fundamental and important tasks in the field of natural language processing, which aims to identify the most relevant label for a given piece of text. Although deep learning-based text classification methods have achieved promising results, most researches mainly focus on the internal context information of the document, ignoring the available global information such as document hierarchy and label semantics. To address this problem, we propose a novel Label-Attentive Hierarchical Network (LAHN) for document classification. In particular, we integrate label information into the hierarchical structure of the document by calculating the word-label attention at word level and the sentence-label attention at sentence level respectively. We give full consideration to the global information during encoding the whole document, which makes the final document representation vector more discriminative for classification. Extensive experiments on several benchmark datasets show that our proposed LAHN surpasses several state-of-the-art methods.

Mobile edge computing (MEC) can alleviate computation and power limitation of user equipments (UEs) by offloading tasks to MEC servers or the remote cloud. Delays of finishing tasks are the most important indicators for a MEC system. However, existing researches in MEC on task allocation problems are decided by UEs or by centralized algorithms, imposing burden to UEs or centralize controllers. Most of them neglect real-time resource utilization of MEC system, which may affect the performance of executing offloaded tasks. To address the above problems, we propose a distributed game-theoretic task-offloading allocation (GTOA) algorithm by transforming a task allocation problem into a strategy game, turning the goal of maximizing deadline satisfaction and resource usage into a payoff function, which MEC server is eager to obtain. Simulations with a different number of MEC servers of system handling tasks for UEs showed that the algorithm can improve system resource utilization while meeting delay limit of most offloaded tasks.

In this paper, we introduce an automatic method for image quality enhancement. Our method relies on a multi-level connected generator and a pair-wise relativistic discriminator. Moreover, we also incorporate auxiliary loss functions to exhibit a high-quality image enhancement. Different from other image style transfer approaches, our method has two appealing properties: 1) different scale information not only pass on the same scale but also flow from encoder to decoder freely and 2) more contextual information is provided to help the discriminator to perform pair-wise discriminative learning. The learned generative model can then automatically enhance the low-quality image into a digital single lens reflex (DSLR) quality. The extensive experiments demonstrate that the proposed framework surpasses the state-of-the-art methods with a clear margin.

Automatically evaluating dialogue coherence is a challenging but high-demand ability for developing high-quality open-domain dialogue systems. However, current evaluation metrics consider only surface features or utterance-level semantics, without explicitly considering the fine-grained topic transition dynamics of dialogue flows. Here, we first consider that the graph structure constituted with topics in a dialogue can accurately depict the underlying communication logic, which is a more natural way to produce persuasive metrics. Capitalized on the topic-level dialogue graph, we propose a new evaluation metric GRADE, which stands for Graph-enhanced Representations for Automatic Dialogue Evaluation. Specifically, GRADE incorporates both coarse-grained utterance-level contextualized representations and fine-grained topic-level graph representations to evaluate dialogue coherence. The graph representations are obtained by reasoning over topic-level dialogue graphs enhanced with the evidence from a commonsense graph, including k-hop neighboring representations and hop-attention weights. Experimental results show that our GRADE significantly outperforms other state-of-the-art metrics on measuring diverse dialogue models in terms of the Pearson and Spearman correlations with human judgements. Besides, we release a new large-scale human evaluation benchmark to facilitate future research on automatic metrics., Comment: Long paper; EMNLP2020

In order to reduce the impact of the randomness and volatility of renewable energy on the economic operation of AC/DC hybrid microgrids, a multi-time-scale rolling optimization strategy is proposed for the grid-connected AC/DC hybrid microgrids. It considers the source-load uncertainty declined with time scale reduction, and the scheduling cooperation problem of different units on different time scales. In this paper, we propose a three-time-scale optimal strategy of the day-ahead, intraday and real-time dispatching stage and a two-level rolling optimal strategy of the intraday and real-time stage, aiming at minimizing the operating cost. We added the power penalty cost in the rolling optimization model to limit the energy state of the energy storage system in the constraint, and improve the power correction and tracking effect of the rolling optimization. A typical-structure AC/DC hybrid microgrid is analyzed in this paper and the simulation results are shown to demonstrate the feasibility and effectiveness of the proposed multi-time-scale rolling optimal dispatch.

As a cyber attack which leverages social engineering and other sophisticated techniques to steal sensitive information from users, phishing attack has been a critical threat to cyber security for a long time. Although researchers have proposed lots of countermeasures, phishing criminals figure out circumventions eventually since such countermeasures require substantial manual feature engineering and can not detect newly emerging phishing attacks well enough, which makes developing an efficient and effective phishing detection method an urgent need. In this work, we propose a novel phishing website detection approach by detecting the Uniform Resource Locator (URL) of a website, which is proved to be an effective and efficient detection approach. To be specific, our novel capsule-based neural network mainly includes several parallel branches wherein one convolutional layer extracts shallow features from URLs and the subsequent two capsule layers generate accurate feature representations of URLs from the shallow features and discriminate the legitimacy of URLs. The final output of our approach is obtained by averaging the outputs of all branches. Extensive experiments on a validated dataset collected from the Internet demonstrate that our approach can achieve competitive performance against other state-of-the-art detection methods while maintaining a tolerable time overhead.

This paper reviewed the 3rd NTIRE challenge on single-image super-resolution (restoration of rich details in a low-resolution image) with a focus on proposed solutions and results. The challenge had 1 track, which was aimed at the real-world single image super-resolution problem with an unknown scaling factor. Participants were mapping low-resolution images captured by a DSLR camera with a shorter focal length to their high-resolution images captured at a longer focal length. With this challenge, we introduced a novel real-world super-resolution dataset (RealSR). The track had 403 registered participants, and 36 teams competed in the final testing phase. They gauge the state-of-the-art in real-world single image super-resolution.

Recently, deep learning based single image super-resolution(SR) approaches have achieved great development. The state-of-the-art SR methods usually adopt a feed-forward pipeline to establish a non-linear mapping between low-res(LR) and high-res(HR) images. However, due to treating all image regions equally without considering the difficulty diversity, these approaches meet an upper bound for optimization. To address this issue, we propose a novel SR approach that discriminately processes each image region within an image by its difficulty. Specifically, we propose a dual-way SR network that one way is trained to focus on easy image regions and another is trained to handle hard image regions. To identify whether a region is easy or hard, we propose a novel image difficulty recognition network based on PSNR prior. Our SR approach that uses the region mask to adaptively enforce the dual-way SR network yields superior results. Extensive experiments on several standard benchmarks (e.g., Set5, Set14, BSD100, and Urban100) show that our approach achieves state-of-the-art performance., ICME2019(Oral), code and results are available at: https://github.com/xzwlx/Difficulty-SR

The offloading policy has attracted much attention in the study of Device-to-Device (D2D) Communication underlaying cellular network integrated with Mobile Edge Computing (MEC) service. D2D communication can improve energy efficiency and prolong the battery life of user equipment (UEs) by offloading computation tasks from a UE to a nearby UE via a D2D link. How to coordinate UEs to offload tasks in D2D underlay communication integrated with MEC Service for reducing interference and improving network energy efficiency is a very important subject. In this paper, we study a D2D underlay communication integrated with MEC service system that provides three offloading policies that are: (i) executing tasks in local device, (ii) offloading tasks to servers in local region, (iii) offloading tasks to a nearby UE. Furthermore, we formulate the computation offloading problem as a potential game and propose a D2D communication integrated with MEC service offloading computation algorithm for solving the computation offloading problem in a short period. The simulation results show that our proposed algorithm can reduce overall system energy consumption under time constraints.

This paper reviews the first challenge on efficient perceptual image enhancement with the focus on deploying deep learning models on smartphones. The challenge consisted of two tracks. In the first one, participants were solving the classical image super-resolution problem with a bicubic downscaling factor of 4. The second track was aimed at real-world photo enhancement, and the goal was to map low-quality photos from the iPhone 3GS device to the same photos captured with a DSLR camera. The target metric used in this challenge combined the runtime, PSNR scores and solutions’ perceptual results measured in the user study. To ensure the efficiency of the submitted models, we additionally measured their runtime and memory requirements on Android smartphones. The proposed solutions significantly improved baseline results defining the state-of-the-art for image enhancement on smartphones.

This paper reviews the first NTIRE challenge on perceptual image enhancement with the focus on proposed solutions and results. The participating teams were solving a real-world photo enhancement problem, where the goal was to map low-quality photos from the iPhone 3GS device to the same photos captured with Canon 70D DSLR camera. The considered problem embraced a number of computer vision subtasks, such as image denoising, image resolution and sharpness enhancement, image color/contrast/exposure adjustment, etc. The target metric used in this challenge combined PSNR and SSIM scores with solutions' perceptual results measured in the user study. The proposed solutions significantly improved baseline results, defining the state-of-the-art for practical image enhancement.

Mobile edge computing (MEC) is a promising paradigm to integrate computing and communication resources in mobile networks. MEC can improve mobile service quality and enhance Quality of Experience (QoE) by offloading computation tasks to MEC servers. However, a MEC server only can provide limited computational resources for users. In this paper, we consider a mobile edge computing system that provides three offloading policies that are: (i) executing tasks in local device, (ii) offloading tasks to servers in a local region, (iii)offloading tasks to servers in a nearby region. In the policy (iii), mobile user equipment can utilize computational resources of MEC servers in nearby regions to solve the problem of insufficient computational resources in local region servers. We formulate the computation offloading problem as a potential game and propose a Distributed Offloading strategy based on Jacobi algorithm (DOJ) for solving the computation offloading problem in a short period. The simulation results show that our proposed algorithm can reduce overall system costs and guarantee the QoE of users.

With renewable generation resources and multiple load demands increasing, the combined cooling, heating, and power (CCHP) microgrid energy management system has attracted much attention due to its high efficiency and low emissions. In order to realize the integration of substation resources and solve the problems of inaccurate, random, volatile and intermittent load forecasting, we propose a three-stage coordinated optimization scheduling strategy for a CCHP microgrid. The strategy contains three stages: a day-ahead economic scheduling stage, an intraday rolling optimization stage, and a real-time adjustment stage. Forecasting data with different accuracy at different time scales were used to carry out multilevel coordination and gradually improve the scheduling plan. A case study was used to verify that the proposed scheduling strategy can mitigate and eliminate the load forecasting error of renewable energy (for power balance and scheduling economy).