arXiv subjects

Yu Zhou

Publications and source records attributed to Yu Zhou.

At least 379 records · Page 21Linked to original sources

arXiv2020

CodeGRU: Context-aware Deep Learning with Gated Recurrent Unit for Source Code Modeling

Recently deep learning based Natural Language Processing (NLP) models have shown great potential in the modeling of source code. However, a major limitation of these approaches is that they take source code as simple tokens of text and ignore its contextual, syntactical and structural dependencies. In this work, we present CodeGRU, a gated recurrent unit based source code language model that is capable of capturing source code's contextual, syntactical and structural dependencies. We introduce a novel approach which can capture the source code context by leveraging the source code token types. Further, we adopt a novel approach which can learn variable size context by taking into account source code's syntax, and structural information. We evaluate CodeGRU with real-world data set and it shows that CodeGRU outperforms the state-of-the-art language models and help reduce the vocabulary size up to 24.93\%. Unlike previous works, we tested CodeGRU with an independent test set which suggests that our methodology does not requisite the source code comes from the same domain as training data while providing suggestions. We further evaluate CodeGRU with two software engineering applications: source code suggestion, and source code completion. Our experiment confirms that the source code's contextual information can be vital and can help improve the software language models. The extensive evaluation of CodeGRU shows that it outperforms the state-of-the-art models. The results further suggest that the proposed approach can help reduce the vocabulary size and is of practical use for software developers.

cs.NE↗

arXiv2020

DeepVS: An Efficient and Generic Approach for Source Code Modeling Usage

The source code suggestions provided by current IDEs are mostly dependent on static type learning. These suggestions often end up proposing irrelevant suggestions for a peculiar context. Recently, deep learning-based approaches have shown great potential in the modeling of source code for various software engineering tasks. However, these techniques lack adequate generalization and resistance to acclimate the use of such models in a real-world software development environment. This letter presents \textit{DeepVS}, an end-to-end deep neural code completion tool that learns from existing codebases by exploiting the bidirectional Gated Recurrent Unit (BiGRU) neural net. The proposed tool is capable of providing source code suggestions instantly in an IDE by using pre-trained BiGRU neural net. The evaluation of this work is two-fold, quantitative and qualitative. Through extensive evaluation on ten real-world open-source software systems, the proposed method shows significant performance enhancement and its practicality. Moreover, the results also suggest that \textit{DeepVS} tool is capable of suggesting zero-day (unseen) code tokens by learning coding patterns from real-world software systems.

cs.NE↗

arXiv2020

Deep Transfer Learning for Source Code Modeling

In recent years, deep learning models have shown great potential in source code modeling and analysis. Generally, deep learning-based approaches are problem-specific and data-hungry. A challenging issue of these approaches is that they require training from starch for a different related problem. In this work, we propose a transfer learning-based approach that significantly improves the performance of deep learning-based source code models. In contrast to traditional learning paradigms, transfer learning can transfer the knowledge learned in solving one problem into another related problem. First, we present two recurrent neural network-based models RNN and GRU for the purpose of transfer learning in the domain of source code modeling. Next, via transfer learning, these pre-trained (RNN and GRU) models are used as feature extractors. Then, these extracted features are combined into attention learner for different downstream tasks. The attention learner leverages from the learned knowledge of pre-trained models and fine-tunes them for a specific downstream task. We evaluate the performance of the proposed approach with extensive experiments with the source code suggestion task. The results indicate that the proposed approach outperforms the state-of-the-art models in terms of accuracy, precision, recall, and F-measure without training the models from scratch.

cs.LG↗

arXiv2020

Expert Training: Task Hardness Aware Meta-Learning for Few-Shot Classification

Deep neural networks are highly effective when a large number of labeled samples are available but fail with few-shot classification tasks. Recently, meta-learning methods have received much attention, which train a meta-learner on massive additional tasks to gain the knowledge to instruct the few-shot classification. Usually, the training tasks are randomly sampled and performed indiscriminately, often making the meta-learner stuck into a bad local optimum. Some works in the optimization of deep neural networks have shown that a better arrangement of training data can make the classifier converge faster and perform better. Inspired by this idea, we propose an easy-to-hard expert meta-training strategy to arrange the training tasks properly, where easy tasks are preferred in the first phase, then, hard tasks are emphasized in the second phase. A task hardness aware module is designed and integrated into the training procedure to estimate the hardness of a task based on the distinguishability of its categories. In addition, we explore multiple hardness measurements including the semantic relation, the pairwise Euclidean distance, the Hausdorff distance, and the Hilbert-Schmidt independence criterion. Experimental results on the miniImageNet and tieredImageNetSketch datasets show that the meta-learners can obtain better results with our expert training strategy.

cs.CV↗

arXiv2020

FC2RN: A Fully Convolutional Corner Refinement Network for Accurate Multi-Oriented Scene Text Detection

Recent scene text detection works mainly focus on curve text detection. However, in real applications, the curve texts are more scarce than the multi-oriented ones. Accurate detection of multi-oriented text with large variations of scales, orientations, and aspect ratios is of great significance. Among the multi-oriented detection methods, direct regression for the geometry of scene text shares a simple yet powerful pipeline and gets popular in academic and industrial communities, but it may produce imperfect detections, especially for long texts due to the limitation of the receptive field. In this work, we aim to improve this while keeping the pipeline simple. A fully convolutional corner refinement network (FC2RN) is proposed for accurate multi-oriented text detection, in which an initial corner prediction and a refined corner prediction are obtained at one pass. With a novel quadrilateral RoI convolution operation tailed for multi-oriented scene text, the initial quadrilateral prediction is encoded into the feature maps which can be further used to predict offset between the initial prediction and the ground-truth as well as output a refined confidence score. Experimental results on four public datasets including MSRA-TD500, ICDAR2017-RCTW, ICDAR2015, and COCO-Text demonstrate that FC2RN can outperform the state-of-the-art methods. The ablation study shows the effectiveness of corner refinement and scoring for accurate text localization.

cs.CV↗

arXiv2020

Video Playback Rate Perception for Self-supervisedSpatio-Temporal Representation Learning

In self-supervised spatio-temporal representation learning, the temporal resolution and long-short term characteristics are not yet fully explored, which limits representation capabilities of learned models. In this paper, we propose a novel self-supervised method, referred to as video Playback Rate Perception (PRP), to learn spatio-temporal representation in a simple-yet-effective way. PRP roots in a dilated sampling strategy, which produces self-supervision signals about video playback rates for representation model learning. PRP is implemented with a feature encoder, a classification module, and a reconstructing decoder, to achieve spatio-temporal semantic retention in a collaborative discrimination-generation manner. The discriminative perception model follows a feature encoder to prefer perceiving low temporal resolution and long-term representation by classifying fast-forward rates. The generative perception model acts as a feature decoder to focus on comprehending high temporal resolution and short-term representation by introducing a motion-attention mechanism. PRP is applied on typical video target tasks including action recognition and video retrieval. Experiments show that PRP outperforms state-of-the-art self-supervised models with significant margins. Code is available at github.com/yuanyao366/PRP

cs.CV↗

arXiv2020

Defect-Correlated Skyrmions and Controllable Generation in Perpendicularly Magnetized CoFeB Ultrathin Films

Skyrmions have attracted significant interest due to their topological spin structures and fascinating physical features. The skyrmion phase arises in materials with Dzyaloshinskii-Moriya (DM) interaction at interfaces or in volume of non-centrosymmetric materials. However, although skyrmions were generated experimentally, one critical intrinsic relationship between fabrication, microstructures, magnetization and the existence of skyrmions remains to be established. Here, two series of CoFeB ultrathin films with controlled atomic scale structures are employed to reveal this relationship. By inverting the growth order, the amount of defects can be artificially tuned, and skyrmions are shown to be preferentially formed at defect sites. The stable region and the density of the skyrmions can be efficiently controlled in the return magnetization loops by utilizing first-order reversal curves to reach various metastable states. These findings establish the general and intrinsic relationship from sample preparation to skyrmion generation, offering an universal method to control skyrmion density.

cond-mat.mes-hall↗

arXiv2020

Self-Training for Domain Adaptive Scene Text Detection

Though deep learning based scene text detection has achieved great progress, well-trained detectors suffer from severe performance degradation for different domains. In general, a tremendous amount of data is indispensable to train the detector in the target domain. However, data collection and annotation are expensive and time-consuming. To address this problem, we propose a self-training framework to automatically mine hard examples with pseudo-labels from unannotated videos or images. To reduce the noise of hard examples, a novel text mining module is implemented based on the fusion of detection and tracking results. Then, an image-to-video generation method is designed for the tasks that videos are unavailable and only images can be used. Experimental results on standard benchmarks, including ICDAR2015, MSRA-TD500, ICDAR2017 MLT, demonstrate the effectiveness of our self-training method. The simple Mask R-CNN adapted with self-training and fine-tuned on real data can achieve comparable or even superior results with the state-of-the-art methods.

cs.CV↗

arXiv2020

SEED: Semantics Enhanced Encoder-Decoder Framework for Scene Text Recognition

Scene text recognition is a hot research topic in computer vision. Recently, many recognition methods based on the encoder-decoder framework have been proposed, and they can handle scene texts of perspective distortion and curve shape. Nevertheless, they still face lots of challenges like image blur, uneven illumination, and incomplete characters. We argue that most encoder-decoder methods are based on local visual features without explicit global semantic information. In this work, we propose a semantics enhanced encoder-decoder framework to robustly recognize low-quality scene texts. The semantic information is used both in the encoder module for supervision and in the decoder module for initializing. In particular, the state-of-the art ASTER method is integrated into the proposed framework as an exemplar. Extensive experiments demonstrate that the proposed framework is more robust for low-quality text images, and achieves state-of-the-art results on several benchmark datasets.

cs.CV↗

arXiv2020

Artificial intelligence control of a turbulent jet

An artificial intelligence (AI) control system is developed to maximize the mixing rate of a turbulent jet. This system comprises six independently operated unsteady minijet actuators, two hot-wire sensors placed in the jet, and genetic programming for the unsupervised learning of a near-optimal control law. The ansatz of this law includes multi-frequency open-loop forcing, sensor-feedback and nonlinear combinations thereof. Mixing performance is quantified by the decay rate of the centreline mean velocity of jet. Intriguingly, the learning process of AI control discovers the classical forcings, i.e. axisymmetric, helical and flapping achievable from conventional control techniques, one by one in the order of increased performance, and finally converges to a hitherto unexplored forcing. Careful examination of the control landscape unveils typical control laws, generated in the learning process, and their evolutions. The best AI forcing produces a complex turbulent flow structure that is characterized by periodically generated mushroom structures, helical motion and oscillating jet column, all enhancing the mixing rate and vastly outperforming others. Being never reported before, this flow structure is examined in various aspects, including the velocity spectra, mean and fluctuating velocity fields and their downstream evolution, and flow visualization images in three orthogonal planes, all compared with other classical flow structures. Along with the knowledge of the minijet-produced flow and its effect on the initial condition of the main jet, these aspects cast valuable insight into the physics behind the highly effective mixing of this newly found flow structure. The results point to the great potential of AI in conquering the vast opportunity space of control laws for many actuators and sensors and in optimizing turbulence.

physics.flu-dyn↗

arXiv2020

User Review-Based Change File Localization for Mobile Applications

In the current mobile app development, novel and emerging DevOps practices (e.g., Continuous Delivery, Integration, and user feedback analysis) and tools are becoming more widespread. For instance, the integration of user feedback (provided in the form of user reviews) in the software release cycle represents a valuable asset for the maintenance and evolution of mobile apps. To fully make use of these assets, it is highly desirable for developers to establish semantic links between the user reviews and the software artefacts to be changed (e.g., source code and documentation), and thus to localize the potential files to change for addressing the user feedback. In this paper, we propose RISING (Review Integration via claSsification, clusterIng, and linkiNG), an automated approach to support the continuous integration of user feedback via classification, clustering, and linking of user reviews. RISING leverages domain-specific constraint information and semi-supervised learning to group user reviews into multiple fine-grained clusters concerning similar users' requests. Then, by combining the textual information from both commit messages and source code, it automatically localizes potential change files to accommodate the users' requests. Our empirical studies demonstrate that the proposed approach outperforms the state-of-the-art baseline work in terms of clustering and localization accuracy, and thus produces more reliable results.

cs.SE↗

arXiv2020

Subtracting non-critical fluctuations in higher cumulants of conserved charges

Using the sample produced by the AMPT default model, we construct a corresponding mixed sample by the method of mixed events. The mixed sample provides an effective estimation for non-critical fluctuations which are caused by global and systematic effects. The dynamical cumulants of conserved charges are defined as the cumulants of the original sample minus the cumulants of the mixed sample. It is demonstrated that dynamical cumulants are subtracted statistical fluctuations, and centrality bin width or detection efficiency independent, in consistent with formulae corrected cumulants. Therefore, dynamical cumulants are helpful in obtaining critical fluctuations at the RHIC BES.

hep-ph↗

arXiv2020

Video Cloze Procedure for Self-Supervised Spatio-Temporal Learning

We propose a novel self-supervised method, referred to as Video Cloze Procedure (VCP), to learn rich spatial-temporal representations. VCP first generates "blanks" by withholding video clips and then creates "options" by applying spatio-temporal operations on the withheld clips. Finally, it fills the blanks with "options" and learns representations by predicting the categories of operations applied on the clips. VCP can act as either a proxy task or a target task in self-supervised learning. As a proxy task, it converts rich self-supervised representations into video clip operations (options), which enhances the flexibility and reduces the complexity of representation learning. As a target task, it can assess learned representation models in a uniform and interpretable manner. With VCP, we train spatial-temporal representation models (3D-CNNs) and apply such models on action recognition and video retrieval tasks. Experiments on commonly used benchmarks show that the trained models outperform the state-of-the-art self-supervised models with significant margins.

cs.CV↗

arXiv2019

TANet: Robust 3D Object Detection from Point Clouds with Triple Attention

In this paper, we focus on exploring the robustness of the 3D object detection in point clouds, which has been rarely discussed in existing approaches. We observe two crucial phenomena: 1) the detection accuracy of the hard objects, e.g., Pedestrians, is unsatisfactory, 2) when adding additional noise points, the performance of existing approaches decreases rapidly. To alleviate these problems, a novel TANet is introduced in this paper, which mainly contains a Triple Attention (TA) module, and a Coarse-to-Fine Regression (CFR) module. By considering the channel-wise, point-wise and voxel-wise attention jointly, the TA module enhances the crucial information of the target while suppresses the unstable cloud points. Besides, the novel stacked TA further exploits the multi-level feature attention. In addition, the CFR module boosts the accuracy of localization without excessive computation cost. Experimental results on the validation set of KITTI dataset demonstrate that, in the challenging noisy cases, i.e., adding additional random noisy points around each object,the presented approach goes far beyond state-of-the-art approaches. Furthermore, for the 3D object detection task of the KITTI benchmark, our approach ranks the first place on Pedestrian class, by using the point clouds as the only input. The running speed is around 29 frames per second.

cs.CV↗

arXiv2019

Tunable Microwave Single-photon Source Based on Transmon Qubit with High Efficiency

Single-photon sources are of great interest because they are key elements in different promising applications of quantum technologies. Here we demonstrate a highly efficient tunable on-demand microwave single-photon source based on a transmon qubit with the intrinsic emission efficiency above 98$\%$. The high efficiency ensures a negligible pure dephasing rate and the necessary condition for generation of indistinguishable photons. We provide an extended discussion and analysis of the efficiency of the photon generation. To further experimentally confirm the single-photon property of the source, correlation functions of the emission field are also measured using linear detectors with a GPU-enhanced signal processing technique. Our results experimentally demonstrate that frequency tunability and negligible pure dephasing rate can be achieved simultaneously and show that such a tunable single-photon source can be good for various practical applications in quantum communication, simulations and information processing in the microwave regime.

quant-ph↗

arXiv2019

Room temperature nanocavity laser with interlayer excitons in 2D heterostructures

Atomically thin layered two dimensional (2D) material has provided a rich library for both fundamental research and device applications. One of the special advantages is that, bandgap engineering and controlled material response can be achieved by stacking different 2D materials. Recently several types of excitonic lasers have been reported based on Transition metal dichalcogenide (TMDC) monolayers, however, the emission is still the intrinsic energy bandgap of the monolayers and lasers harnessing the flexibility of Van der Waals heterostructures have not been demonstrated yet. Here, we report for the first time a room temperature interlayer exciton laser with MoS2/WSe2 heterostructures. The onset of lasing action was identified by a combination of distinct kink in the 'L-L' curve and the noticeable collapse of spectral linewidth. Different from visible emission of intralayer excitons for both MoS2 and WSe2, our interlayer exciton laser works in the infrared range, which is fully compatible with the well-established technologies in silicon photonics. Thanks to the long lifetime of interlayer excitons, the requirement of the cavity quality factor is relaxed by orders of magnitude. The demonstration of room temperature interlayer exciton laser might open new perspectives for the development of coherent light source with tailored optical properties on silicon photonics platform.

cond-mat.mes-hall↗

arXiv2019

Locating fixed points in the phase plane

The critical point is a fixed point in finite-size scaling. To quantify the behaviour of such a fixed point, we define, at a given temperature and scaling exponent ratio, the width of scaled observables for different sizes. The minimum of the width reveals the position of fixed point, its corresponding phase transition temperature, and scaling exponent ratio. The value of this ratio tells the nature of fixed point, which can be a critical point, a point of the first order phase transition line, or a point of the crossover region. To demonstrate the effectiveness of this method, we apply it to three typical samples produced by the 3D three-state Potts model. Results show the method to be more precise and effective than conventional methods. Finally, we discuss a possible application at the beam energy scan program of relativistic heavy-Ion collision.

hep-ph↗

arXiv2019

The method of mixed events for higher cumulants of conserved charges

Higher cumulants of conserved charges are sensitive observables of quantum chromodynamics phase transitions. The sample of mixed events provides a background to estimate non-critical effects of cumulants. Four possible methods for constructing the sample of mixed events are suggested. The effectiveness of each method is examined. It is showed that the method of most random or least constrain is the best, rather than the conventional method.

hep-ph↗