scispace - formally typeset
Search or ask a question

Showing papers by "Tsinghua University published in 2016"


Journal ArticleDOI
B. P. Abbott1, Richard J. Abbott1, T. D. Abbott2, Matthew Abernathy1  +1008 moreInstitutions (96)
TL;DR: This is the first direct detection of gravitational waves and the first observation of a binary black hole merger, and these observations demonstrate the existence of binary stellar-mass black hole systems.
Abstract: On September 14, 2015 at 09:50:45 UTC the two detectors of the Laser Interferometer Gravitational-Wave Observatory simultaneously observed a transient gravitational-wave signal. The signal sweeps upwards in frequency from 35 to 250 Hz with a peak gravitational-wave strain of $1.0 \times 10^{-21}$. It matches the waveform predicted by general relativity for the inspiral and merger of a pair of black holes and the ringdown of the resulting single black hole. The signal was observed with a matched-filter signal-to-noise ratio of 24 and a false alarm rate estimated to be less than 1 event per 203 000 years, equivalent to a significance greater than 5.1 {\sigma}. The source lies at a luminosity distance of $410^{+160}_{-180}$ Mpc corresponding to a redshift $z = 0.09^{+0.03}_{-0.04}$. In the source frame, the initial black hole masses are $36^{+5}_{-4} M_\odot$ and $29^{+4}_{-4} M_\odot$, and the final black hole mass is $62^{+4}_{-4} M_\odot$, with $3.0^{+0.5}_{-0.5} M_\odot c^2$ radiated in gravitational waves. All uncertainties define 90% credible intervals.These observations demonstrate the existence of binary stellar-mass black hole systems. This is the first direct detection of gravitational waves and the first observation of a binary black hole merger.

9,596 citations


Proceedings Article
15 Feb 2016
TL;DR: Deep Compression as mentioned in this paper proposes a three-stage pipeline: pruning, quantization, and Huffman coding to reduce the storage requirement of neural networks by 35x to 49x without affecting their accuracy.
Abstract: Neural networks are both computationally intensive and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources. To address this limitation, we introduce "deep compression", a three stage pipeline: pruning, trained quantization and Huffman coding, that work together to reduce the storage requirement of neural networks by 35x to 49x without affecting their accuracy. Our method first prunes the network by learning only the important connections. Next, we quantize the weights to enforce weight sharing, finally, we apply Huffman coding. After the first two steps we retrain the network to fine tune the remaining connections and the quantized centroids. Pruning, reduces the number of connections by 9x to 13x; Quantization then reduces the number of bits that represent each connection from 32 to 5. On the ImageNet dataset, our method reduced the storage required by AlexNet by 35x, from 240MB to 6.9MB, without loss of accuracy. Our method reduced the size of VGG-16 by 49x from 552MB to 11.3MB, again with no loss of accuracy. This allows fitting the model into on-chip SRAM cache rather than off-chip DRAM memory. Our compression method also facilitates the use of complex neural networks in mobile applications where application size and download bandwidth are constrained. Benchmarked on CPU, GPU and mobile GPU, compressed network has 3x to 4x layerwise speedup and 3x to 7x better energy efficiency.

7,256 citations


Journal ArticleDOI
Daniel J. Klionsky1, Kotb Abdelmohsen2, Akihisa Abe3, Joynal Abedin4  +2519 moreInstitutions (695)
TL;DR: In this paper, the authors present a set of guidelines for the selection and interpretation of methods for use by investigators who aim to examine macro-autophagy and related processes, as well as for reviewers who need to provide realistic and reasonable critiques of papers that are focused on these processes.
Abstract: In 2008 we published the first set of guidelines for standardizing research in autophagy. Since then, research on this topic has continued to accelerate, and many new scientists have entered the field. Our knowledge base and relevant new technologies have also been expanding. Accordingly, it is important to update these guidelines for monitoring autophagy in different organisms. Various reviews have described the range of assays that have been used for this purpose. Nevertheless, there continues to be confusion regarding acceptable methods to measure autophagy, especially in multicellular eukaryotes. For example, a key point that needs to be emphasized is that there is a difference between measurements that monitor the numbers or volume of autophagic elements (e.g., autophagosomes or autolysosomes) at any stage of the autophagic process versus those that measure flux through the autophagy pathway (i.e., the complete process including the amount and rate of cargo sequestered and degraded). In particular, a block in macroautophagy that results in autophagosome accumulation must be differentiated from stimuli that increase autophagic activity, defined as increased autophagy induction coupled with increased delivery to, and degradation within, lysosomes (in most higher eukaryotes and some protists such as Dictyostelium) or the vacuole (in plants and fungi). In other words, it is especially important that investigators new to the field understand that the appearance of more autophagosomes does not necessarily equate with more autophagy. In fact, in many cases, autophagosomes accumulate because of a block in trafficking to lysosomes without a concomitant change in autophagosome biogenesis, whereas an increase in autolysosomes may reflect a reduction in degradative activity. It is worth emphasizing here that lysosomal digestion is a stage of autophagy and evaluating its competence is a crucial part of the evaluation of autophagic flux, or complete autophagy. Here, we present a set of guidelines for the selection and interpretation of methods for use by investigators who aim to examine macroautophagy and related processes, as well as for reviewers who need to provide realistic and reasonable critiques of papers that are focused on these processes. These guidelines are not meant to be a formulaic set of rules, because the appropriate assays depend in part on the question being asked and the system being used. In addition, we emphasize that no individual assay is guaranteed to be the most appropriate one in every situation, and we strongly recommend the use of multiple assays to monitor autophagy. Along these lines, because of the potential for pleiotropic effects due to blocking autophagy through genetic manipulation, it is imperative to target by gene knockout or RNA interference more than one autophagy-related protein. In addition, some individual Atg proteins, or groups of proteins, are involved in other cellular pathways implying that not all Atg proteins can be used as a specific marker for an autophagic process. In these guidelines, we consider these various methods of assessing autophagy and what information can, or cannot, be obtained from them. Finally, by discussing the merits and limits of particular assays, we hope to encourage technical innovation in the field.

5,187 citations


Proceedings Article
20 May 2016
TL;DR: R-FCN as mentioned in this paper proposes position-sensitive score maps to address the dilemma between translation-invariance in image classification and translation-variance in object detection, and achieves state-of-the-art performance on the PASCAL VOC dataset.
Abstract: We present region-based, fully convolutional networks for accurate and efficient object detection. In contrast to previous region-based detectors such as Fast/Faster R-CNN that apply a costly per-region subnetwork hundreds of times, our region-based detector is fully convolutional with almost all computation shared on the entire image. To achieve this goal, we propose position-sensitive score maps to address a dilemma between translation-invariance in image classification and translation-variance in object detection. Our method can thus naturally adopt fully convolutional image classifier backbones, such as the latest Residual Networks (ResNets), for object detection. We show competitive results on the PASCAL VOC datasets (e.g., 83.6% mAP on the 2007 set) with the 101-layer ResNet. Meanwhile, our result is achieved at a test-time speed of 170ms per image, 2.5-20 times faster than the Faster R-CNN counterpart. Code is made publicly available at: https://github.com/daijifeng001/r-fcn.

4,618 citations


Journal ArticleDOI
B. P. Abbott1, Richard J. Abbott1, T. D. Abbott2, M. R. Abernathy3  +970 moreInstitutions (114)
TL;DR: This second gravitational-wave observation provides improved constraints on stellar populations and on deviations from general relativity.
Abstract: We report the observation of a gravitational-wave signal produced by the coalescence of two stellar-mass black holes. The signal, GW151226, was observed by the twin detectors of the Laser Interferometer Gravitational-Wave Observatory (LIGO) on December 26, 2015 at 03:38:53 UTC. The signal was initially identified within 70 s by an online matched-filter search targeting binary coalescences. Subsequent off-line analyses recovered GW151226 with a network signal-to-noise ratio of 13 and a significance greater than 5 σ. The signal persisted in the LIGO frequency band for approximately 1 s, increasing in frequency and amplitude over about 55 cycles from 35 to 450 Hz, and reached a peak gravitational strain of 3.4+0.7−0.9×10−22. The inferred source-frame initial black hole masses are 14.2+8.3−3.7M⊙ and 7.5+2.3−2.3M⊙ and the final black hole mass is 20.8+6.1−1.7M⊙. We find that at least one of the component black holes has spin greater than 0.2. This source is located at a luminosity distance of 440+180−190 Mpc corresponding to a redshift 0.09+0.03−0.04. All uncertainties define a 90 % credible interval. This second gravitational-wave observation provides improved constraints on stellar populations and on deviations from general relativity.

3,448 citations


Proceedings ArticleDOI
13 Aug 2016
TL;DR: This paper proposes a Structural Deep Network Embedding method, namely SDNE, which first proposes a semi-supervised deep model, which has multiple layers of non-linear functions, thereby being able to capture the highly non- linear network structure and exploits the first-order and second-order proximity jointly to preserve the network structure.
Abstract: Network embedding is an important method to learn low-dimensional representations of vertexes in networks, aiming to capture and preserve the network structure. Almost all the existing network embedding methods adopt shallow models. However, since the underlying network structure is complex, shallow models cannot capture the highly non-linear network structure, resulting in sub-optimal network representations. Therefore, how to find a method that is able to effectively capture the highly non-linear network structure and preserve the global and local structure is an open yet important problem. To solve this problem, in this paper we propose a Structural Deep Network Embedding method, namely SDNE. More specifically, we first propose a semi-supervised deep model, which has multiple layers of non-linear functions, thereby being able to capture the highly non-linear network structure. Then we propose to exploit the first-order and second-order proximity jointly to preserve the network structure. The second-order proximity is used by the unsupervised component to capture the global network structure. While the first-order proximity is used as the supervised information in the supervised component to preserve the local network structure. By jointly optimizing them in the semi-supervised deep model, our method can preserve both the local and global network structure and is robust to sparse networks. Empirically, we conduct the experiments on five real-world networks, including a language network, a citation network and three social networks. The results show that compared to the baselines, our method can reconstruct the original network significantly better and achieves substantial gains in three applications, i.e. multi-label classification, link prediction and visualization.

2,238 citations


Journal ArticleDOI
TL;DR: All of the major steps in RNA-seq data analysis are reviewed, including experimental design, quality control, read alignment, quantification of gene and transcript levels, visualization, differential gene expression, alternative splicing, functional analysis, gene fusion detection and eQTL mapping.
Abstract: RNA-sequencing (RNA-seq) has a wide variety of applications, but no single analysis pipeline can be used in all cases. We review all of the major steps in RNA-seq data analysis, including experimental design, quality control, read alignment, quantification of gene and transcript levels, visualization, differential gene expression, alternative splicing, functional analysis, gene fusion detection and eQTL mapping. We highlight the challenges associated with each step. We discuss the analysis of small RNAs and the integration of RNA-seq with other functional genomics techniques. Finally, we discuss the outlook for novel technologies that are changing the state of the art in transcriptomics.

1,963 citations


Journal ArticleDOI
TL;DR: Durability tests revealed that the Co single atoms exhibit outstanding chemical stability during electrocatalysis and thermal stability that resists sintering at 900 °C, which could facilitate new discoveries at the atomic scale in condensed materials.
Abstract: A new strategy for achieving stable Co single atoms (SAs) on nitrogen-doped porous carbon with high metal loading over 4 wt % is reported. The strategy is based on a pyrolysis process of predesigned bimetallic Zn/Co metal–organic frameworks, during which Co can be reduced by carbonization of the organic linker and Zn is selectively evaporated away at high temperatures above 800 °C. The spherical aberration correction electron microscopy and extended X-ray absorption fine structure measurements both confirm the atomic dispersion of Co atoms stabilized by as-generated N-doped porous carbon. Surprisingly, the obtained Co-Nx single sites exhibit superior ORR performance with a half-wave potential (0.881 V) that is more positive than commercial Pt/C (0.811 V) and most reported non-precious metal catalysts. Durability tests revealed that the Co single atoms exhibit outstanding chemical stability during electrocatalysis and thermal stability that resists sintering at 900 °C. Our findings open up a new routine for general and practical synthesis of a variety of materials bearing single atoms, which could facilitate new discoveries at the atomic scale in condensed materials.

1,779 citations


Proceedings ArticleDOI
01 Nov 2016
TL;DR: This paper reveals that the sentiment polarity of a sentence is not only determined by the content but is also highly related to the concerned aspect, and proposes an Attention-based Long Short-Term Memory Network for aspect-level sentiment classification.
Abstract: Aspect-level sentiment classification is a fine-grained task in sentiment analysis. Since it provides more complete and in-depth results, aspect-level sentiment analysis has received much attention these years. In this paper, we reveal that the sentiment polarity of a sentence is not only determined by the content but is also highly related to the concerned aspect. For instance, “The appetizers are ok, but the service is slow.”, for aspect taste, the polarity is positive while for service, the polarity is negative. Therefore, it is worthwhile to explore the connection between an aspect and the content of a sentence. To this end, we propose an Attention-based Long Short-Term Memory Network for aspect-level sentiment classification. The attention mechanism can concentrate on different parts of a sentence when different aspects are taken as input. We experiment on the SemEval 2014 dataset and results show that our model achieves state-ofthe-art performance on aspect-level sentiment classification.

1,722 citations


Book ChapterDOI
08 Oct 2016
TL;DR: Stochastic depth is proposed, a training procedure that enables the seemingly contradictory setup to train short networks and use deep networks at test time and reduces training time substantially and improves the test error significantly on almost all data sets that were used for evaluation.
Abstract: Very deep convolutional networks with hundreds of layers have led to significant reductions in error on competitive benchmarks. Although the unmatched expressiveness of the many layers can be highly desirable at test time, training very deep networks comes with its own set of challenges. The gradients can vanish, the forward flow often diminishes, and the training time can be painfully slow. To address these problems, we propose stochastic depth, a training procedure that enables the seemingly contradictory setup to train short networks and use deep networks at test time. We start with very deep networks but during training, for each mini-batch, randomly drop a subset of layers and bypass them with the identity function. This simple approach complements the recent success of residual networks. It reduces training time substantially and improves the test error significantly on almost all data sets that we used for evaluation. With stochastic depth we can increase the depth of residual networks even beyond 1200 layers and still yield meaningful improvements in test error (4.91 % on CIFAR-10).

1,485 citations


Journal ArticleDOI
B. P. Abbott1, Richard J. Abbott1, T. D. Abbott2, M. R. Abernathy1  +976 moreInstitutions (107)
TL;DR: It is found that the final remnant's mass and spin, as determined from the low-frequency and high-frequency phases of the signal, are mutually consistent with the binary black-hole solution in general relativity.
Abstract: The LIGO detection of GW150914 provides an unprecedented opportunity to study the two-body motion of a compact-object binary in the large-velocity, highly nonlinear regime, and to witness the final merger of the binary and the excitation of uniquely relativistic modes of the gravitational field. We carry out several investigations to determine whether GW150914 is consistent with a binary black-hole merger in general relativity. We find that the final remnant’s mass and spin, as determined from the low-frequency (inspiral) and high-frequency (postinspiral) phases of the signal, are mutually consistent with the binary black-hole solution in general relativity. Furthermore, the data following the peak of GW150914 are consistent with the least-damped quasinormal mode inferred from the mass and spin of the remnant black hole. By using waveform models that allow for parametrized general-relativity violations during the inspiral and merger phases, we perform quantitative tests on the gravitational-wave phase in the dynamical regime and we determine the first empirical bounds on several high-order post-Newtonian coefficients. We constrain the graviton Compton wavelength, assuming that gravitons are dispersed in vacuum in the same way as particles with mass, obtaining a 90%-confidence lower bound of 1013 km. In conclusion, within our statistical uncertainties, we find no evidence for violations of general relativity in the genuinely strong-field regime of gravity.

Journal ArticleDOI
TL;DR: A multidisciplinary approach is highly required to enable the formation of robust SEI for highly efficient energy storage systems.
Abstract: Lithium metal batteries (LMBs) are among the most promising candidates of high-energy-density devices for advanced energy storage. However, the growth of dendrites greatly hinders the practical applications of LMBs in portable electronics and electric vehicles. Constructing stable and efficient solid electrolyte interphase (SEI) is among the most effective strategies to inhibit the dendrite growth and thus to achieve a superior cycling performance. In this review, the mechanisms of SEI formation and models of SEI structure are briefly summarized. The analysis methods to probe the surface chemistry, surface morphology, electrochemical property, dynamic characteristics of SEI layer are emphasized. The critical factors affecting the SEI formation, such as electrolyte component, temperature, current density, are comprehensively debated. The efficient methods to modify SEI layer with the introduction of new electrolyte system and additives, ex-situ-formed protective layer, as well as electrode design, are summarized. Although these works afford new insights into SEI research, robust and precise routes for SEI modification with well-designed structure, as well as understanding of the connection between structure and electrochemical performance, is still inadequate. A multidisciplinary approach is highly required to enable the formation of robust SEI for highly efficient energy storage systems.

Proceedings Article
05 Dec 2016
TL;DR: Empirical evidence shows that the new approach to domain adaptation in deep networks that can jointly learn adaptive classifiers and transferable features from labeled data in the source domain and unlabeledData in the target domain outperforms state of the art methods on standard domain adaptation benchmarks.
Abstract: The recent success of deep neural networks relies on massive amounts of labeled data. For a target task where labeled data is unavailable, domain adaptation can transfer a learner from a different source domain. In this paper, we propose a new approach to domain adaptation in deep networks that can jointly learn adaptive classifiers and transferable features from labeled data in the source domain and unlabeled data in the target domain. We relax a shared-classifier assumption made by previous methods and assume that the source classifier and target classifier differ by a residual function. We enable classifier adaptation by plugging several layers into deep network to explicitly learn the residual function with reference to the target classifier. We fuse features of multiple layers with tensor product and embed them into reproducing kernel Hilbert spaces to match distributions for feature adaptation. The adaptation can be achieved in most feed-forward models by extending them with new residual layers and loss functions, which can be trained efficiently via back-propagation. Empirical evidence shows that the new approach outperforms state of the art methods on standard domain adaptation benchmarks.

Journal ArticleDOI
TL;DR: The propelling redox reaction is not limited to Li-S system, and it is foresee the reported strategy herein can be applied in other high-power devices through the systems with controllable redox reactions.
Abstract: Lithium–sulfur (Li–S) battery system is endowed with tremendous energy density, resulting from the complex sulfur electrochemistry involving multielectron redox reactions and phase transformations. Originated from the slow redox kinetics of polysulfide intermediates, the flood of polysulfides in the batteries during cycling induced low sulfur utilization, severe polarization, low energy efficiency, deteriorated polysulfide shuttle, and short cycling life. Herein, sulfiphilic cobalt disulfide (CoS2) was incorporated into carbon/sulfur cathodes, introducing strong interaction between lithium polysulfides and CoS2 under working conditions. The interfaces between CoS2 and electrolyte served as strong adsorption and activation sites for polar polysulfides and therefore accelerated redox reactions of polysulfides. The high polysulfide reactivity not only guaranteed effective polarization mitigation and promoted energy efficiency by 10% but also promised high discharge capacity and stable cycling performance dur...

Journal ArticleDOI
18 Jun 2016
TL;DR: This work proposes a novel PIM architecture, called PRIME, to accelerate NN applications in ReRAM based main memory, and distinguishes itself from prior work on NN acceleration, with significant performance improvement and energy saving.
Abstract: Processing-in-memory (PIM) is a promising solution to address the "memory wall" challenges for future computer systems. Prior proposed PIM architectures put additional computation logic in or near memory. The emerging metal-oxide resistive random access memory (ReRAM) has showed its potential to be used for main memory. Moreover, with its crossbar array structure, ReRAM can perform matrix-vector multiplication efficiently, and has been widely studied to accelerate neural network (NN) applications. In this work, we propose a novel PIM architecture, called PRIME, to accelerate NN applications in ReRAM based main memory. In PRIME, a portion of ReRAM crossbar arrays can be configured as accelerators for NN applications or as normal memory for a larger memory space. We provide microarchitecture and circuit designs to enable the morphable functions with an insignificant area overhead. We also design a software/hardware interface for software developers to implement various NNs on PRIME. Benefiting from both the PIM architecture and the efficiency of using ReRAM for NN computation, PRIME distinguishes itself from prior work on NN acceleration, with significant performance improvement and energy saving. Our experimental results show that, compared with a state-of-the-art neural processing unit design, PRIME improves the performance by ~2360× and the energy consumption by ~895×, across the evaluated machine learning benchmarks.

Posted Content
TL;DR: In this article, the authors address the problem of 3D reconstruction from a single image, generating a straight-forward form of output -point cloud coordinates. But the groundtruth shape for an input image may be ambiguous, and they design architecture, loss function and learning paradigm that are novel and effective.
Abstract: Generation of 3D data by deep neural network has been attracting increasing attention in the research community. The majority of extant works resort to regular representations such as volumetric grids or collection of images; however, these representations obscure the natural invariance of 3D shapes under geometric transformations and also suffer from a number of other issues. In this paper we address the problem of 3D reconstruction from a single image, generating a straight-forward form of output -- point cloud coordinates. Along with this problem arises a unique and interesting issue, that the groundtruth shape for an input image may be ambiguous. Driven by this unorthodox output form and the inherent ambiguity in groundtruth, we design architecture, loss function and learning paradigm that are novel and effective. Our final solution is a conditional shape sampler, capable of predicting multiple plausible 3D point clouds from an input image. In experiments not only can our system outperform state-of-the-art methods on single image based 3d reconstruction benchmarks; but it also shows a strong performance for 3d shape completion and promising ability in making multiple plausible predictions.

Proceedings ArticleDOI
21 Feb 2016
TL;DR: This paper presents an in-depth analysis of state-of-the-art CNN models and shows that Convolutional layers are computational-centric and Fully-Connected layers are memory-centric, and proposes a CNN accelerator design on embedded FPGA for Image-Net large-scale image classification.
Abstract: In recent years, convolutional neural network (CNN) based methods have achieved great success in a large number of applications and have been among the most powerful and widely used techniques in computer vision. However, CNN-based methods are com-putational-intensive and resource-consuming, and thus are hard to be integrated into embedded systems such as smart phones, smart glasses, and robots. FPGA is one of the most promising platforms for accelerating CNN, but the limited bandwidth and on-chip memory size limit the performance of FPGA accelerator for CNN.In this paper, we go deeper with the embedded FPGA platform on accelerating CNNs and propose a CNN accelerator design on embedded FPGA for Image-Net large-scale image classification. We first present an in-depth analysis of state-of-the-art CNN models and show that Convolutional layers are computational-centric and Fully-Connected layers are memory-centric.Then the dynamic-precision data quantization method and a convolver design that is efficient for all layer types in CNN are proposed to improve the bandwidth and resource utilization. Results show that only 0.4% accuracy loss is introduced by our data quantization flow for the very deep VGG16 model when 8/4-bit quantization is used. A data arrangement method is proposed to further ensure a high utilization of the external memory bandwidth. Finally, a state-of-the-art CNN, VGG16-SVD, is implemented on an embedded FPGA platform as a case study. VGG16-SVD is the largest and most accurate network that has been implemented on FPGA end-to-end so far. The system on Xilinx Zynq ZC706 board achieves a frame rate at 4.45 fps with the top-5 accuracy of 86.66% using 16-bit quantization. The average performance of convolutional layers and the full CNN is 187.8 GOP/s and 137.0 GOP/s under 150MHz working frequency, which outperform previous approaches significantly.

Journal ArticleDOI
B. P. Abbott1, Richard J. Abbott1, T. D. Abbott2, Matthew Abernathy3  +978 moreInstitutions (112)
TL;DR: The first observational run of the Advanced LIGO detectors, from September 12, 2015 to January 19, 2016, saw the first detections of gravitational waves from binary black hole mergers as discussed by the authors.
Abstract: The first observational run of the Advanced LIGO detectors, from September 12, 2015 to January 19, 2016, saw the first detections of gravitational waves from binary black hole mergers. In this paper we present full results from a search for binary black hole merger signals with total masses up to 100M⊙ and detailed implications from our observations of these systems. Our search, based on general-relativistic models of gravitational wave signals from binary black hole systems, unambiguously identified two signals, GW150914 and GW151226, with a significance of greater than 5σ over the observing period. It also identified a third possible signal, LVT151012, with substantially lower significance, which has a 87% probability of being of astrophysical origin. We provide detailed estimates of the parameters of the observed systems. Both GW150914 and GW151226 provide an unprecedented opportunity to study the two-body motion of a compact-object binary in the large velocity, highly nonlinear regime. We do not observe any deviations from general relativity, and place improved empirical bounds on several high-order post-Newtonian coefficients. From our observations we infer stellar-mass binary black hole merger rates lying in the range 9−240Gpc−3yr−1. These observations are beginning to inform astrophysical predictions of binary black hole formation rates, and indicate that future observing runs of the Advanced detector network will yield many more gravitational wave detections.

Proceedings ArticleDOI
13 Aug 2016
TL;DR: A novel graph embedding algorithm, High-Order Proximity preserved Embedding (HOPE for short), is developed, which is scalable to preserve high-order proximities of large scale graphs and capable of capturing the asymmetric transitivity.
Abstract: Graph embedding algorithms embed a graph into a vector space where the structure and the inherent properties of the graph are preserved The existing graph embedding methods cannot preserve the asymmetric transitivity well, which is a critical property of directed graphs Asymmetric transitivity depicts the correlation among directed edges, that is, if there is a directed path from u to v, then there is likely a directed edge from u to v Asymmetric transitivity can help in capturing structures of graphs and recovering from partially observed graphs To tackle this challenge, we propose the idea of preserving asymmetric transitivity by approximating high-order proximity which are based on asymmetric transitivity In particular, we develop a novel graph embedding algorithm, High-Order Proximity preserved Embedding (HOPE for short), which is scalable to preserve high-order proximities of large scale graphs and capable of capturing the asymmetric transitivity More specifically, we first derive a general formulation that cover multiple popular high-order proximity measurements, then propose a scalable embedding algorithm to approximate the high-order proximity measurements based on their general formulation Moreover, we provide a theoretical upper bound on the RMSE (Root Mean Squared Error) of the approximation Our empirical experiments on a synthetic dataset and three real-world datasets demonstrate that HOPE can approximate the high-order proximities significantly better than the state-of-art algorithms and outperform the state-of-art algorithms in tasks of reconstruction, link prediction and vertex recommendation

Proceedings ArticleDOI
Yankai Lin1, Shiqi Shen1, Zhiyuan Liu1, Huanbo Luan1, Maosong Sun1 
01 Aug 2016
TL;DR: A sentence-level attention-based model for relation extraction that employs convolutional neural networks to embed the semantics of sentences and dynamically reduce the weights of those noisy instances.
Abstract: Distant supervised relation extraction has been widely used to find novel relational facts from text. However, distant supervision inevitably accompanies with the wrong labelling problem, and these noisy data will substantially hurt the performance of relation extraction. To alleviate this issue, we propose a sentence-level attention-based model for relation extraction. In this model, we employ convolutional neural networks to embed the semantics of sentences. Afterwards, we build sentence-level attention over multiple instances, which is expected to dynamically reduce the weights of those noisy instances. Experimental results on real-world datasets show that, our model can make full use of all informative sentences and effectively reduce the influence of wrong labelled instances. Our model achieves significant and consistent improvements on relation extraction as compared with baselines. The source code of this paper can be obtained from https: //github.com/thunlp/NRE.

Posted Content
TL;DR: The Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion, and has several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters.
Abstract: Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with L layers have L connections - one between each layer and its subsequent layer - our network has L(L+1)/2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less computation to achieve high performance. Code and pre-trained models are available at this https URL .

Journal ArticleDOI
TL;DR: The classes, functions and the representatives of the frequently detected PPCPs in aquatic environments were summarized and suggestions are made for future study of P PCPs.

Posted Content
TL;DR: Stochastic depth as discussed by the authors randomly drops a subset of layers during training and bypasses them with the identity function, which can increase the depth of residual networks even beyond 1200 layers and still yield meaningful improvements in test error.
Abstract: Very deep convolutional networks with hundreds of layers have led to significant reductions in error on competitive benchmarks. Although the unmatched expressiveness of the many layers can be highly desirable at test time, training very deep networks comes with its own set of challenges. The gradients can vanish, the forward flow often diminishes, and the training time can be painfully slow. To address these problems, we propose stochastic depth, a training procedure that enables the seemingly contradictory setup to train short networks and use deep networks at test time. We start with very deep networks but during training, for each mini-batch, randomly drop a subset of layers and bypass them with the identity function. This simple approach complements the recent success of residual networks. It reduces training time substantially and improves the test error significantly on almost all data sets that we used for evaluation. With stochastic depth we can increase the depth of residual networks even beyond 1200 layers and still yield meaningful improvements in test error (4.91% on CIFAR-10).

Proceedings ArticleDOI
27 Jun 2016
TL;DR: This work proposes an energy minimization approach that places object candidates in 3D using the fact that objects should be on the ground-plane, and achieves the best detection performance on the challenging KITTI benchmark, among published monocular competitors.
Abstract: The goal of this paper is to perform 3D object detection from a single monocular image in the domain of autonomous driving. Our method first aims to generate a set of candidate class-specific object proposals, which are then run through a standard CNN pipeline to obtain highquality object detections. The focus of this paper is on proposal generation. In particular, we propose an energy minimization approach that places object candidates in 3D using the fact that objects should be on the ground-plane. We then score each candidate box projected to the image plane via several intuitive potentials encoding semantic segmentation, contextual information, size and location priors and typical object shape. Our experimental evaluation demonstrates that our object proposal generation approach significantly outperforms all monocular approaches, and achieves the best detection performance on the challenging KITTI benchmark, among published monocular competitors.

Book ChapterDOI
08 Oct 2016
TL;DR: It is shown that CNN in classification mode can be trained from scratch using the consecutive bounding boxes of each identity, and the learned CNN embedding outperforms other competing methods considerably and has good generalization ability on other video re-id datasets upon fine-tuning.
Abstract: This paper considers person re-identification (re-id) in videos. We introduce a new video re-id dataset, named Motion Analysis and Re-identification Set (MARS), a video extension of the Market-1501 dataset. To our knowledge, MARS is the largest video re-id dataset to date. Containing 1,261 IDs and around 20,000 tracklets, it provides rich visual information compared to image-based datasets. Meanwhile, MARS reaches a step closer to practice. The tracklets are automatically generated by the Deformable Part Model (DPM) as pedestrian detector and the GMMCP tracker. A number of false detection/tracking results are also included as distractors which would exist predominantly in practical video databases. Extensive evaluation of the state-of-the-art methods including the space-time descriptors and CNN is presented. We show that CNN in classification mode can be trained from scratch using the consecutive bounding boxes of each identity. The learned CNN embedding outperforms other competing methods considerably and has good generalization ability on other video re-id datasets upon fine-tuning.

Proceedings ArticleDOI
Rui Fu1, Zuo Zhang1, Li Li1
01 Nov 2016
TL;DR: This paper uses Long Short Term Memory and Gated Recurrent Units (GRU) neural network methods to predict short-term traffic flow, and experiments demonstrate that Recurrent Neural Network (RNN) based deep learning methods such as LSTM and GRU perform better than auto regressive integrated moving average (ARIMA) model.
Abstract: Accurate and real-time traffic flow prediction is important in Intelligent Transportation System (ITS), especially for traffic control. Existing models such as ARMA, ARIMA are mainly linear models and cannot describe the stochastic and nonlinear nature of traffic flow. In recent years, deep-learning-based methods have been applied as novel alternatives for traffic flow prediction. However, which kind of deep neural networks is the most appropriate model for traffic flow prediction remains unsolved. In this paper, we use Long Short Term Memory (LSTM) and Gated Recurrent Units (GRU) neural network (NN) methods to predict short-term traffic flow, and experiments demonstrate that Recurrent Neural Network (RNN) based deep learning methods such as LSTM and GRU perform better than auto regressive integrated moving average (ARIMA) model. To the best of our knowledge, this is the first time that GRU is applied to traffic flow prediction.

Journal ArticleDOI
B. P. Abbott1, Richard J. Abbott1, T. D. Abbott2, Matthew Abernathy1  +984 moreInstitutions (116)
TL;DR: The data around the time of the event were analyzed coherently across the LIGO network using a suite of accurate waveform models that describe gravitational waves from a compact binary system in general relativity.
Abstract: On September 14, 2015, the Laser Interferometer Gravitational-wave Observatory (LIGO) detected a gravitational-wave transient (GW150914); we characterise the properties of the source and its parameters. The data around the time of the event were analysed coherently across the LIGO network using a suite of accurate waveform models that describe gravitational waves from a compact binary system in general relativity. GW150914 was produced by a nearly equal mass binary black hole of $36^{+5}_{-4} M_\odot$ and $29^{+4}_{-4} M_\odot$ (for each parameter we report the median value and the range of the 90% credible interval). The dimensionless spin magnitude of the more massive black hole is bound to be $0.7$ (at 90% probability). The luminosity distance to the source is $410^{+160}_{-180}$ Mpc, corresponding to a redshift $0.09^{+0.03}_{-0.04}$ assuming standard cosmology. The source location is constrained to an annulus section of $590$ deg$^2$, primarily in the southern hemisphere. The binary merges into a black hole of $62^{+4}_{-4} M_\odot$ and spin $0.67^{+0.05}_{-0.07}$. This black hole is significantly more massive than any other known in the stellar-mass regime.

Journal ArticleDOI
TL;DR: Li dendrite-free growth is achieved by employing glass fiber with large polar functional groups as the interlayer of Li metal anode and separator to uniformly distribute Li ions.
Abstract: Li dendrite-free growth is achieved by employing glass fiber with large polar functional groups as the interlayer of Li metal anode and separator to uniformly distribute Li ions. The evenly distributed Li ions render the dendrite-free Li deposits at high rates (10 mA cm(-2)) and high lithiation capacity (2.0 mAh cm(-2)).

Journal ArticleDOI
TL;DR: In this paper, the missing source of sulfate and particulate matter can be explained by reactive nitrogen chemistry in aerosol water, where the alkaline aerosol components trap SO 2, which is oxidized by NO 2 to form sulfate, whereby high reaction rates are sustained by the high neutralizing capacity of the atmosphere.
Abstract: Fine-particle pollution associated with winter haze threatens the health of more than 400 million people in the North China Plain. Sulfate is a major component of fine haze particles. Record sulfate concentrations of up to ~300 μg m −3 were observed during the January 2013 winter haze event in Beijing. State-of-the-art air quality models that rely on sulfate production mechanisms requiring photochemical oxidants cannot predict these high levels because of the weak photochemistry activity during haze events. We find that the missing source of sulfate and particulate matter can be explained by reactive nitrogen chemistry in aerosol water. The aerosol water serves as a reactor, where the alkaline aerosol components trap SO 2 , which is oxidized by NO 2 to form sulfate, whereby high reaction rates are sustained by the high neutralizing capacity of the atmosphere in northern China. This mechanism is self-amplifying because higher aerosol mass concentration corresponds to higher aerosol water content, leading to faster sulfate production and more severe haze pollution.

Journal ArticleDOI
Fengpeng An1, Guangpeng An, Qi An2, Vito Antonelli3  +226 moreInstitutions (55)
TL;DR: The Jiangmen Underground Neutrino Observatory (JUNO) as mentioned in this paper is a 20kton multi-purpose underground liquid scintillator detector with the determination of neutrino mass hierarchy (MH) as a primary physics goal.
Abstract: The Jiangmen Underground Neutrino Observatory (JUNO), a 20 kton multi-purpose underground liquid scintillator detector, was proposed with the determination of the neutrino mass hierarchy (MH) as a primary physics goal. The excellent energy resolution and the large fiducial volume anticipated for the JUNO detector offer exciting opportunities for addressing many important topics in neutrino and astro-particle physics. In this document, we present the physics motivations and the anticipated performance of the JUNO detector for various proposed measurements. Following an introduction summarizing the current status and open issues in neutrino physics, we discuss how the detection of antineutrinos generated by a cluster of nuclear power plants allows the determination of the neutrino MH at a 3–4σ significance with six years of running of JUNO. The measurement of antineutrino spectrum with excellent energy resolution will also lead to the precise determination of the neutrino oscillation parameters ${\mathrm{sin}}^{2}{\theta }_{12}$, ${\rm{\Delta }}{m}_{21}^{2}$, and $| {\rm{\Delta }}{m}_{{ee}}^{2}| $ to an accuracy of better than 1%, which will play a crucial role in the future unitarity test of the MNSP matrix. The JUNO detector is capable of observing not only antineutrinos from the power plants, but also neutrinos/antineutrinos from terrestrial and extra-terrestrial sources, including supernova burst neutrinos, diffuse supernova neutrino background, geoneutrinos, atmospheric neutrinos, and solar neutrinos. As a result of JUNO's large size, excellent energy resolution, and vertex reconstruction capability, interesting new data on these topics can be collected. For example, a neutrino burst from a typical core-collapse supernova at a distance of 10 kpc would lead to ∼5000 inverse-beta-decay events and ∼2000 all-flavor neutrino–proton ES events in JUNO, which are of crucial importance for understanding the mechanism of supernova explosion and for exploring novel phenomena such as collective neutrino oscillations. Detection of neutrinos from all past core-collapse supernova explosions in the visible universe with JUNO would further provide valuable information on the cosmic star-formation rate and the average core-collapse neutrino energy spectrum. Antineutrinos originating from the radioactive decay of uranium and thorium in the Earth can be detected in JUNO with a rate of ∼400 events per year, significantly improving the statistics of existing geoneutrino event samples. Atmospheric neutrino events collected in JUNO can provide independent inputs for determining the MH and the octant of the ${\theta }_{23}$ mixing angle. Detection of the (7)Be and (8)B solar neutrino events at JUNO would shed new light on the solar metallicity problem and examine the transition region between the vacuum and matter dominated neutrino oscillations. Regarding light sterile neutrino topics, sterile neutrinos with ${10}^{-5}\,{{\rm{eV}}}^{2}\lt {\rm{\Delta }}{m}_{41}^{2}\lt {10}^{-2}\,{{\rm{eV}}}^{2}$ and a sufficiently large mixing angle ${\theta }_{14}$ could be identified through a precise measurement of the reactor antineutrino energy spectrum. Meanwhile, JUNO can also provide us excellent opportunities to test the eV-scale sterile neutrino hypothesis, using either the radioactive neutrino sources or a cyclotron-produced neutrino beam. The JUNO detector is also sensitive to several other beyondthe-standard-model physics. Examples include the search for proton decay via the $p\to {K}^{+}+\bar{ u }$ decay channel, search for neutrinos resulting from dark-matter annihilation in the Sun, search for violation of Lorentz invariance via the sidereal modulation of the reactor neutrino event rate, and search for the effects of non-standard interactions. The proposed construction of the JUNO detector will provide a unique facility to address many outstanding crucial questions in particle and astrophysics in a timely and cost-effective fashion. It holds the great potential for further advancing our quest to understanding the fundamental properties of neutrinos, one of the building blocks of our Universe.