Searcharxiv⌕ Search

arXiv subjects

Jun Sun

Publications and source records attributed to Jun Sun.

At least 271 records · Page 15Linked to original sources
arXiv2017

On Study of the Reliable Fully Convolutional Networks with Tree Arranged Outputs (TAO-FCN) for Handwritten String Recognition

The handwritten string recognition is still a challengeable task, though the powerful deep learning tools were introduced. In this paper, based on TAO-FCN, we proposed an end-to-end system for handwritten string recognition. Compared with the conventional methods, there is no preprocess nor manually designed rules employed. With enough labelled data, it is easy to apply the proposed method to different applications. Although the performance of the proposed method may not be comparable with the state-of-the-art approaches, it's usability and robustness are more meaningful for practical applications.

cs.CV↗
arXiv2017

Analyzing the Impact of Cognitive Load in Evaluating Gaze-based Typing

Gaze-based virtual keyboards provide an effective interface for text entry by eye movements. The efficiency and usability of these keyboards have traditionally been evaluated with conventional text entry performance measures such as words per minute, keystrokes per character, backspace usage, etc. However, in comparison to the traditional text entry approaches, gaze-based typing involves natural eye movements that are highly correlated with human brain cognition. Employing eye gaze as an input could lead to excessive mental demand, and in this work we argue the need to include cognitive load as an eye typing evaluation measure. We evaluate three variations of gaze-based virtual keyboards, which implement variable designs in terms of word suggestion positioning. The conventional text entry metrics indicate no significant difference in the performance of the different keyboard designs. However, STFT (Short-time Fourier Transform) based analysis of EEG signals indicate variances in the mental workload of participants while interacting with these designs. Moreover, the EEG analysis provides insights into the user's cognition variation for different typing phases and intervals, which should be considered in order to improve eye typing usability.

cs.HC↗
arXiv2017

The Problem of Action at a Distance in Networks and the Emergence of Preferential Attachment from Triadic Closure

In this paper, we characterise the notion of preferential attachment in networks as action at a distance, and argue that it can only be an emergent phenomenon -- the actual mechanism by which networks grow always being the closing of triangles. After a review of the concepts of triangle closing and preferential attachment, we present our argument, as well as a simplified model in which preferential attachment can be derived mathematically from triangle closing. Additionally, we perform experiments on synthetic graphs to demonstrate the emergence of preferential attachment in graph growth models based only on triangle closing.

cs.SI↗
arXiv2017

Quality-Driven Disorder Handling for M-way Sliding Window Stream Joins

Sliding window join is one of the most important operators for stream applications. To produce high quality join results, a stream processing system must deal with the ubiquitous disorder within input streams which is caused by network delay, asynchronous source clocks, etc. Disorder handling involves an inevitable tradeoff between the latency and the quality of produced join results. To meet different requirements of stream applications, it is desirable to provide a user-configurable result-latency vs. result-quality tradeoff. Existing disorder handling approaches either do not provide such configurability, or support only user-specified latency constraints. In this work, we advocate the idea of quality-driven disorder handling, and propose a buffer-based disorder handling approach for sliding window joins, which minimizes sizes of input-sorting buffers, thus the result latency, while respecting user-specified result-quality requirements. The core of our approach is an analytical model which directly captures the relationship between sizes of input buffers and the produced result quality. Our approach is generic. It supports m-way sliding window joins with arbitrary join conditions. Experiments on real-world and synthetic datasets show that, compared to the state of the art, our approach can reduce the result latency incurred by disorder handling by up to 95% while providing the same level of result quality.

cs.DB↗
arXiv2017

Should We Learn Probabilistic Models for Model Checking? A New Approach and An Empirical Study

Many automated system analysis techniques (e.g., model checking, model-based testing) rely on first obtaining a model of the system under analysis. System modeling is often done manually, which is often considered as a hindrance to adopt model-based system analysis and development techniques. To overcome this problem, researchers have proposed to automatically "learn" models based on sample system executions and shown that the learned models can be useful sometimes. There are however many questions to be answered. For instance, how much shall we generalize from the observed samples and how fast would learning converge? Or, would the analysis result based on the learned model be more accurate than the estimation we could have obtained by sampling many system executions within the same amount of time? In this work, we investigate existing algorithms for learning probabilistic models for model checking, propose an evolution-based approach for better controlling the degree of generalization and conduct an empirical study in order to answer the questions. One of our findings is that the effectiveness of learning may sometimes be limited.

cs.SE↗
arXiv2017

Building Fast and Compact Convolutional Neural Networks for Offline Handwritten Chinese Character Recognition

Like other problems in computer vision, offline handwritten Chinese character recognition (HCCR) has achieved impressive results using convolutional neural network (CNN)-based methods. However, larger and deeper networks are needed to deliver state-of-the-art results in this domain. Such networks intuitively appear to incur high computational cost, and require the storage of a large number of parameters, which renders them unfeasible for deployment in portable devices. To solve this problem, we propose a Global Supervised Low-rank Expansion (GSLRE) method and an Adaptive Drop-weight (ADW) technique to solve the problems of speed and storage capacity. We design a nine-layer CNN for HCCR consisting of 3,755 classes, and devise an algorithm that can reduce the networks computational cost by nine times and compress the network to 1/18 of the original size of the baseline model, with only a 0.21% drop in accuracy. In tests, the proposed algorithm surpassed the best single-network performance reported thus far in the literature while requiring only 2.3 MB for storage. Furthermore, when integrated with our effective forward implementation, the recognition of an offline character image took only 9.7 ms on a CPU. Compared with the state-of-the-art CNN model for HCCR, our approach is approximately 30 times faster, yet 10 times more cost efficient.

cs.CV↗
arXiv2016

An Automated CNN Recommendation System for Image Classification Tasks

Nowadays the CNN is widely used in practical applications for image classification task. However the design of the CNN model is very professional work and which is very difficult for ordinary users. Besides, even for experts of CNN, to select an optimal model for specific task may still need a lot of time (to train many different models). In order to solve this problem, we proposed an automated CNN recommendation system for image classification task. Our system is able to evaluate the complexity of the classification task and the classification ability of the CNN model precisely. By using the evaluation results, the system can recommend the optimal CNN model and which can match the task perfectly. The recommendation process of the system is very fast since we don't need any model training. The experiment results proved that the evaluation methods are very accurate and reliable.

cs.CV↗
arXiv2016

Learning Likely Invariants to Explain Why a Program Fails

Debugging is difficult. Recent studies show that automatic bug localization techniques have limited usefulness. One of the reasons is that programmers typically have to understand why the program fails before fixing it. In this work, we aim to help programmers understand a bug by automatically generating likely invariants which are violated in the failed tests. Given a program with an initial assertion and at least one test case failing the assertion, we first generate random test cases, identify potential bug locations through bug localization, and then generate program state mutation based on active learning techniques to identify a predicate "explaining" the cause of the bug. The predicate is a classifier for the passed test cases and failed test cases. Our main contribution is the application of invariant learning for bug explanation, as well as a novel approach to overcome the problem of lack of test cases in practice. We apply our method to real-world bugs and show the generated invariants are often correlated to the actual bug fixes.

cs.SE↗
arXiv2016

Taming intermittent plasticity at small scales

The extreme miniaturization in modern technology calls for deeper insights into the non-conventional, fluctuation dominated mechanics of materials at micro- to nano-scales. Both experiments and simulations show that sub-micron face-centered-cubic (FCC) crystals exhibit high yield strength, which is, however, compromised by intermittent, power law distributed strain fluctuations, typical of wild plasticity. At macro-scales, the same bulk materials show mild plasticity characterized by bounded, uncorrelated fluctuations. Both anomalous strength and intermittency appear therefore as size effects. While the former is highly desirable, the latter is detrimental because stochastic dislocation avalanches interfere with forming processes and endanger structural stability. In this paper we show that defectiveness, which is used in classical metallurgy to harden materials, can be used at sub-micron scales to suppress intermittent fluctuations. We quantify the coexistence of wild and mild fluctuations in compressed Al alloys micro-pillars, demonstrate that the statistical nature of fluctuations is determined by sample size, and propose quantitative strategies allowing one to temper plastic intermittency by artificially tailored disorder. Our experimental results are rationalized using a theoretical framework which quantifies the competition between external (size related) and internal (disorder related) length scales.

cond-mat.mtrl-sci↗
arXiv2016

Towards Learning and Verifying Invariants of Cyber-Physical Systems by Code Mutation

Cyber-physical systems (CPS), which integrate algorithmic control with physical processes, often consist of physically distributed components communicating over a network. A malfunctioning or compromised component in such a CPS can lead to costly consequences, especially in the context of public infrastructure. In this short paper, we argue for the importance of constructing invariants (or models) of the physical behaviour exhibited by CPS, motivated by their applications to the control, monitoring, and attestation of components. To achieve this despite the inherent complexity of CPS, we propose a new technique for learning invariants that combines machine learning with ideas from mutation testing. We present a preliminary study on a water treatment system that suggests the efficacy of this approach, propose strategies for establishing confidence in the correctness of invariants, then summarise some research questions and the steps we are taking to investigate them.

cs.SE↗
arXiv2016

Towards Concolic Testing for Hybrid Systems

Hybrid systems exhibit both continuous and discrete behavior. Analyzing hybrid systems is known to be hard. Inspired by the idea of concolic testing (of programs), we investigate whether we can combine random sampling and symbolic execution in order to effectively verify hybrid systems. We identify a sufficient condition under which such a combination is more effective than random sampling. Furthermore, we analyze different strategies of combining random sampling and symbolic execution and propose an algorithm which allows us to dynamically switch between them so as to reduce the overall cost. Our method has been implemented as a web-based checker named HyChecker. HyChecker has been evaluated with benchmark hybrid systems and a water treatment system in order to test its effectiveness.

cs.SE↗
arXiv2016

Effects of entrance channel on fusion probability in hot fusion reactions

Within the framework of the dinuclear system (DNS) model, the fusion reactions leading to the compound nuclei 274Hs and 286Cn are investigated. The fusion probability as a function of DNS excitation energy is studied. The calculated results are in good agreement with the available experimental data. The obtained results show that the fusion probabilities are obviously enhanced for the reactions located at high place in potential energy surface, although these reactions may have small values of mass asymmetry. It is found that the enhancement is due to the large potential energy of the initial DNS.

nucl-th↗
arXiv2016

The deformation of symplectic critical surfaces in a Kähler surface-II---Compactness

In this paper we consider the compactness of $β$-symplectic critical surfaces in a Kähler surface. Let $M$ be a compact Kähler surface and $Σ_i\subset M$ be a sequence of closed $β_i$-symplectic critical surfaces with $β_i\toβ_0\in (0,\infty)$. Suppose the quantity $\int_{Σ_i}\frac{1}{\cos^qα_i}dμ_i$ (for some $q>4$) and the genus of $Σ_{i}$ are bounded, then there exists a finite set of points ${\mathcal S}\subset M$ and a subsequence $Σ_{i'}$ that converges uniformly in the $C^l$ topology (for any $l<\infty$) on compact subsets of $M\backslash {\mathcal S}$ to a $β_0$-symplectic critical surface $Σ\subset M$, each connected component of $Σ\setminus {\mathcal S}$ can be extended smoothly across ${\mathcal S}$.

math.DG↗
arXiv2016

A CNN Based Scene Chinese Text Recognition Algorithm With Synthetic Data Engine

Scene text recognition plays an important role in many computer vision applications. The small size of available public available scene text datasets is the main challenge when training a text recognition CNN model. In this paper, we propose a CNN based Chinese text recognition algorithm. To enlarge the dataset for training the CNN model, we design a synthetic data engine for Chinese scene character generation, which generates representative character images according to the fonts use frequency of Chinese texts. As the Chinese text is more complex, the English text recognition CNN architecture is modified for Chinese text. To ensure the small size nature character dataset and the large size artificial character dataset are comparable in training, the CNN model are trained progressively. The proposed Chinese text recognition algorithm is evaluated with two Chinese text datasets. The algorithm achieves better recognize accuracy compared to the baseline methods.

cs.CV↗
arXiv2016

A Novel Scene Text Detection Algorithm Based On Convolutional Neural Network

Candidate text region extraction plays a critical role in convolutional neural network (CNN) based text detection from natural images. In this paper, we propose a CNN based scene text detection algorithm with a new text region extractor. The so called candidate text region extractor I-MSER is based on Maximally Stable Extremal Region (MSER), which can improve the independency and completeness of the extracted candidate text regions. Design of I-MSER is motivated by the observation that text MSERs have high similarity and are close to each other. The independency of candidate text regions obtained by I-MSER is guaranteed by selecting the most representative regions from a MSER tree which is generated according to the spatial overlapping relationship among the MSERs. A multi-layer CNN model is trained to score the confidence value of the extracted regions extracted by the I-MSER for text detection. The new text detection algorithm based on I-MSER is evaluated with wide-used ICDAR 2011 and 2013 datasets and shows improved detection performance compared to the existing algorithms.

cs.CV↗
arXiv2016

On Study of the Binarized Deep Neural Network for Image Classification

Recently, the deep neural network (derived from the artificial neural network) has attracted many researchers' attention by its outstanding performance. However, since this network requires high-performance GPUs and large storage, it is very hard to use it on individual devices. In order to improve the deep neural network, many trials have been made by refining the network structure or training strategy. Unlike those trials, in this paper, we focused on the basic propagation function of the artificial neural network and proposed the binarized deep neural network. This network is a pure binary system, in which all the values and calculations are binarized. As a result, our network can save a lot of computational resource and storage. Therefore, it is possible to use it on various devices. Moreover, the experimental results proved the feasibility of the proposed network.

cs.NE↗