arXiv subjects

Kai Li

Publications and source records attributed to Kai Li.

At least 415 records · Page 23Linked to original sources

arXiv2018

Challenges of measuring the impact of software: an examination of the lme4 R package

The rise of software as a research object is mirrored in the increasing interests towards quantitative studies of scientific software. However, due to the inconsistent practice of citing software, most of the existing studies analyzing the impact of scientific software are based on identification of software name mentions in full-text publications. Despite its limitations, citation data have a much larger quantity and broader coverage of scientific fields than full-text data and thus could support findings in much larger scopes. This paper presents an analysis aiming to evaluate the extent to which citations data can be used to reconstruct the impact of software. Specifically, we identified the variety of citable objects related to the lme4 R package and examined how the package's impact is scattered across these objects. Our results reveal a little-discussed challenge of using citation data to measure the impact of software, that even within the category of formal citation, there might be different forms in which the same software object is cited. This challenge can be mitigated by more carefully selecting objects as the proxy of software. However, it cannot be fully solved until we have one-software-one-proxy policy for software citation.

cs.DL↗

arXiv2018

Cooperation Enforcement and Collusion Resistance in Repeated Public Goods Games

Enforcing cooperation among substantial agents is one of the main objectives for multi-agent systems. However, due to the existence of inherent social dilemmas in many scenarios, the free-rider problem may arise during agents' long-run interactions and things become even severer when self-interested agents work in collusion with each other to get extra benefits. It is commonly accepted that in such social dilemmas, there exists no simple strategy for an agent whereby she can simultaneously manipulate on the utility of each of her opponents and further promote mutual cooperation among all agents. Here, we show that such strategies do exist. Under the conventional repeated public goods game, we novelly identify them and find that, when confronted with such strategies, a single opponent can maximize his utility only via global cooperation and any colluding alliance cannot get the upper hand. Since a full cooperation is individually optimal for any single opponent, a stable cooperation among all players can be achieved. Moreover, we experimentally show that these strategies can still promote cooperation even when the opponents are both self-learning and collusive.

cs.GT↗

arXiv2018

Noncanonical warm inflation: a model with a general Lagrangian density

Warm inflation is extended from the canonical field to the noncanonical field with a general Lagrangian density having coupling form of kinetic and potential terms. We develop the motion equation of the noncanonical inflaton field, and notice the motion equation has an annoying coupling term $\mathcal{L}_{Xφ}\dotφ^2$, which makes the inflation difficult to solve. It is found that a special field redefinition exists, making the motion equation of inflaton uncoupled. Then we can solve inflation in the special field representation for convenience. The relation of field redefinition between the special uncoupled field and general noncanonical field is obtained. The motion equations and slow roll approximations of noncanonical warm inflation in the special field representation are developed. Then we give two examples to show how to translate a general noncanonical Lagrangian density in the coupled $φ$ representation, to the uncoupled $ϕ$ representation. Finally, the cosmological perturbations generated by the new kind of warm inflation is calculated.

gr-qc↗

arXiv2018

FFT Convolutions are Faster than Winograd on Modern CPUs, Here is Why

Winograd-based convolution has quickly gained traction as a preferred approach to implement convolutional neural networks (ConvNet) on various hardware platforms because it requires fewer floating point operations than FFT-based or direct convolutions. This paper compares three highly optimized implementations (regular FFT--, Gauss--FFT--, and Winograd--based convolutions) on modern multi-- and many--core CPUs. Although all three implementations employed the same optimizations for modern CPUs, our experimental results with two popular ConvNets (VGG and AlexNet) show that the FFT--based implementations generally outperform the Winograd--based approach, contrary to the popular belief. To understand the results, we use a Roofline performance model to analyze the three implementations in detail, by looking at each of their computation phases and by considering not only the number of floating point operations, but also the memory bandwidth and the cache sizes. The performance analysis explains why, and under what conditions, the FFT--based implementations outperform the Winograd--based one, on modern CPUs.

cs.PF↗

arXiv2018

Support Neighbor Loss for Person Re-Identification

Person re-identification (re-ID) has recently been tremendously boosted due to the advancement of deep convolutional neural networks (CNN). The majority of deep re-ID methods focus on designing new CNN architectures, while less attention is paid on investigating the loss functions. Verification loss and identification loss are two types of losses widely used to train various deep re-ID models, both of which however have limitations. Verification loss guides the networks to generate feature embeddings of which the intra-class variance is decreased while the inter-class ones is enlarged. However, training networks with verification loss tends to be of slow convergence and unstable performance when the number of training samples is large. On the other hand, identification loss has good separating and scalable property. But its neglect to explicitly reduce the intra-class variance limits its performance on re-ID, because the same person may have significant appearance disparity across different camera views. To avoid the limitations of the two types of losses, we propose a new loss, called support neighbor (SN) loss. Rather than being derived from data sample pairs or triplets, SN loss is calculated based on the positive and negative support neighbor sets of each anchor sample, which contain more valuable contextual information and neighborhood structure that are beneficial for more stable performance. To ensure scalability and separability, a softmax-like function is formulated to push apart the positive and negative support sets. To reduce intra-class variance, the distance between the anchor's nearest positive neighbor and furthest positive sample is penalized. Integrating SN loss on top of Resnet50, superior re-ID results to the state-of-the-art ones are obtained on several widely used datasets.

cs.CV↗

arXiv2018

Transiting Exoplanet Monitoring Project (TEMP). I. Refined System Parameters and Transit Timing Variations of HAT-P-29b

We report the photometry of six transits of the hot Jupiter HAT-P-29b obtained from 2013 October to 2015 January. We analyze the new light curves, in combination with the published photometric, and Doppler velocimetric, and spectroscopic measurements, finding an updated orbital ephemeris for the HAT-P-29 system, $T_{\rm C}[0]= 2456170.5494(15)\,[\rm{BJD_{TDB}}]$ and $P=5.723390(13) \,{\rm days}$. It is $17.63\,{\rm s}$ ($4.0\,σ$) longer than the previously published value, amounting to errors exceeding $2.5\,\rm{hrs}$ at the time of writing (on UTC 2018 June 1). The measured transit mid-times for HAT-P-29b show no compelling evidence of timing anomalies from a linear model, which rules out the presence of a perturbers with masses greater than $0.6$, $0.7$, $0.5$, and $0.4\,{\rm M_\oplus}$ near the $1:2$, $2:3$, $3:2$, and $2:1$ resonances with HAT-P-29b, respectively.

astro-ph.EP↗

arXiv2018

Payoff Control in the Iterated Prisoner's Dilemma

Repeated game has long been the touchstone model for agents' long-run relationships. Previous results suggest that it is particularly difficult for a repeated game player to exert an autocratic control on the payoffs since they are jointly determined by all participants. This work discovers that the scale of a player's capability to unilaterally influence the payoffs may have been much underestimated. Under the conventional iterated prisoner's dilemma, we develop a general framework for controlling the feasible region where the players' payoff pairs lie. A control strategy player is able to confine the payoff pairs in her objective region, as long as this region has feasible linear boundaries. With this framework, many well-known existing strategies can be categorized and various new strategies with nice properties can be further identified. We show that the control strategies perform well either in a tournament or against a human-like opponent.

cs.GT↗

arXiv2018

Scheduling Computation Graphs of Deep Learning Models on Manycore CPUs

For a deep learning model, efficient execution of its computation graph is key to achieving high performance. Previous work has focused on improving the performance for individual nodes of the computation graph, while ignoring the parallelization of the graph as a whole. However, we observe that running multiple operations simultaneously without interference is critical to efficiently perform parallelizable small operations. The attempt of executing the computation graph in parallel in deep learning frameworks usually involves much resource contention among concurrent operations, leading to inferior performance on manycore CPUs. To address these issues, in this paper, we propose Graphi, a generic and high-performance execution engine to efficiently execute a computation graph in parallel on manycore CPUs. Specifically, Graphi minimizes the interference on both software/hardware resources, discovers the best parallel setting with a profiler, and further optimizes graph execution with the critical-path first scheduling. Our experiments show that the parallel execution consistently outperforms the sequential one. The training times on four different neural networks with Graphi are 2.1x to 9.5x faster than those with TensorFlow on a 68-core Intel Xeon Phi processor.

cs.DC↗

arXiv2018

Image Super-Resolution Using Very Deep Residual Channel Attention Networks

Convolutional neural network (CNN) depth is of crucial importance for image super-resolution (SR). However, we observe that deeper networks for image SR are more difficult to train. The low-resolution inputs and features contain abundant low-frequency information, which is treated equally across channels, hence hindering the representational ability of CNNs. To solve these problems, we propose the very deep residual channel attention networks (RCAN). Specifically, we propose a residual in residual (RIR) structure to form very deep network, which consists of several residual groups with long skip connections. Each residual group contains some residual blocks with short skip connections. Meanwhile, RIR allows abundant low-frequency information to be bypassed through multiple skip connections, making the main network focus on learning high-frequency information. Furthermore, we propose a channel attention mechanism to adaptively rescale channel-wise features by considering interdependencies among channels. Extensive experiments show that our RCAN achieves better accuracy and visual improvements against state-of-the-art methods.

cs.CV↗

arXiv2018

Sparse Multi-Output Gaussian Processes for Medical Time Series Prediction

In the scenario of real-time monitoring of hospital patients, high-quality inference of patients' health status using all information available from clinical covariates and lab tests is essential to enable successful medical interventions and improve patient outcomes. Developing a computational framework that can learn from observational large-scale electronic health records (EHRs) and make accurate real-time predictions is a critical step. In this work, we develop and explore a Bayesian nonparametric model based on Gaussian process (GP) regression for hospital patient monitoring. We propose MedGP, a statistical framework that incorporates 24 clinical and lab covariates and supports a rich reference data set from which relationships between observed covariates may be inferred and exploited for high-quality inference of patient state over time. To do this, we develop a highly structured sparse GP kernel to enable tractable computation over tens of thousands of time points while estimating correlations among clinical covariates, patients, and periodicity in patient observations. MedGP has a number of benefits over current methods, including (i) not requiring an alignment of the time series data, (ii) quantifying confidence regions in the predictions, (iii) exploiting a vast and rich database of patients, and (iv) inferring interpretable relationships among clinical covariates. We evaluate and compare results from MedGP on the task of online prediction for three patient subgroups from two medical data sets across 8,043 patients. We found MedGP improves online prediction over baseline methods for nearly all covariates across different disease subgroups and studies. The publicly available code is at https://github.com/bee-hive/MedGP.

stat.ML↗

arXiv2018

Deep Ordinal Hashing with Spatial Attention

Hashing has attracted increasing research attentions in recent years due to its high efficiency of computation and storage in image retrieval. Recent works have demonstrated the superiority of simultaneous feature representations and hash functions learning with deep neural networks. However, most existing deep hashing methods directly learn the hash functions by encoding the global semantic information, while ignoring the local spatial information of images. The loss of local spatial structure makes the performance bottleneck of hash functions, therefore limiting its application for accurate similarity retrieval. In this work, we propose a novel Deep Ordinal Hashing (DOH) method, which learns ordinal representations by leveraging the ranking structure of feature space from both local and global views. In particular, to effectively build the ranking structure, we propose to learn the rank correlation space by exploiting the local spatial information from Fully Convolutional Network (FCN) and the global semantic information from the Convolutional Neural Network (CNN) simultaneously. More specifically, an effective spatial attention model is designed to capture the local spatial information by selectively learning well-specified locations closely related to target objects. In such hashing framework,the local spatial and global semantic nature of images are captured in an end-to-end ranking-to-hashing manner. Experimental results conducted on three widely-used datasets demonstrate that the proposed DOH method significantly outperforms the state-of-the-art hashing methods.

cs.CV↗

arXiv2018

Transiting Exoplanet Monitoring Project (TEMP). IV. Refined System Parameters, Transit Timing Variations and Orbital Stability of the Transiting Planetary System HAT-P-25

We present eight new light curves of the transiting extra-solar planet HAT-P-25b obtained from 2013 to 2016 with three telescopes at two observatories. We use the new light curves, along with recent literature material, to estimate the physical and orbital parameters of the transiting planet. Specifically, we determine the mid-transit times (T$_{C}$) and update the linear ephemeris, T$_{C[0]}$=2456418.80996$\pm$0.00025 [$\mathrm{BJD}_\mathrm{TDB}$] and P=3.65281572$\pm$0.00000095 days. We carry out a search for transit timing variations (TTVs), and find no significant TTV signal at the $ΔT=$80 s-level, placing a limit on the possible strength of planet-planet interactions ($\mathrm{TTV_{G}}$). In the course of our analysis, we calculate the upper mass-limits of the potential nearby perturbers. Near the 1:2, 2:1, and 3:1 resonances with HAT-P-25b, perturbers with masses greater than 0.5, 0.3, and 0.5 $\mathrm{M_{\oplus}}$ respectively, can be excluded. Furthermore, based on the analysis of TTVs caused by light travel time effect (LTTE) we also eliminate the possibility that a long-period perturber exists with $M_{\rm p}> 3000 \,\mathrm{M_{J}}$ within $a=11.2\,{\rm AU}$ of the parent star.

astro-ph.EP↗

arXiv2018

Gain-guided X-ray free-electron laser oscillator

The X-ray free-electron laser oscillator (XFELO) has recently been reconsidered a promising candidate for producing high-brightness, fully coherent pulses in the hard X-ray regime. This letter proposes a gain-guided XFELO scheme, avoiding the external focusing elements required for conventional configuration. Self-consistent theoretical analysis and three-dimensional numerical simulation results verify that the X-ray transverse mode in such an "unstable" cavity is stable and robust due to electron beam gain-guiding. This scheme is capable of generating 14.3 keV photons of FEL radiation with peak power on the order of a gigawatt using Shanghai coherent light source parameters. The output pulse energy and transverse and longitudinal coherence are comparable with those from conventional XFELOs. This promising scheme is expected to contribute significantly to the construction and operation of a real XFELO.

physics.acc-ph↗

arXiv2018

Atomic inner-shell radiation seeded free-electron lasers

In order to effectively improve the output quality of X-ray free electron laser (XFEL), we theoretically propose an XFEL scheme seeded by atomic inner-shell laser. As well known, an atomic inner-shell laser based on neutral atoms and pumped by an XFEL has been experimentally demonstrated, which produced sub-femtosecond X-ray pulses with increased temporal coherence. It shows that, by using the inner-shell laser as a seed to modulate the electron bunch, very stable and almost fully-coherent short-wavelength XFEL pulses can be generated. The proposed scheme holds promising prospects in X-ray wavelengths, and even shorter.

physics.acc-ph↗

arXiv2018

LCD: Low Latency Command Dissemination for A Platoon of Vehicles

In a vehicular platoon, a lead vehicle that is responsible for managing the platoon's moving directions and velocity periodically disseminates control commands to following vehicles based on vehicle-to-vehicle communications. However, reducing command dissemination latency with multiple vehicles while ensuring successful message delivery to the tail vehicle is challenging. We propose a new linear dynamic programming algorithm using backward induction and interchange arguments to minimize the dissemination latency of the vehicles. Furthermore, a closed form of dissemination latency in vehicular platoon is obtained by utilizing Markov chain with M/M/1 queuing model. Simulation results confirm that the proposed dynamic programming algorithm improves the dissemination rate by at least 50.9%, compared to similar algorithms in the literature. Moreover, it also approximates the best performance with the maximum gap of up to 0.2 second in terms of latency.

cs.PF↗

arXiv2017

Transiting Exoplanet Monitoring Project (TEMP). III. On the Relocation of the Kepler-9~b Transit

The Kepler-9 system harbors three known transiting planets. The system holds significant interest for several reasons. First, the outer two planets exhibit a period ratio that is close to a 2:1 orbital commensurability, with attendant dynamical consequences. Second, both planets lie in the planetary mass "desert" that is generally associated with the rapid gas agglomeration phase of the core accretion process. Third, there exist attractive prospects for accurately measuring both the sky-projected stellar spin-orbit angles as well as the mutual orbital inclination between the planets in the system. Following the original \textit{Kepler} detection announcement in 2010, the initially reported orbital ephemerides for Kepler-9~b and c have degraded significantly, due to the limited time base-line of observations on which the discovery of the system rested. Here, we report new ground-based photometric observations and extensive dynamical modeling of the system. These efforts allow us to photometrically recover the transit of Kepler-9~b, and thereby greatly improve the predictions for upcoming transit mid-times. Accurate ephemerides of this system are important in order to confidently schedule follow-up observations of this system, for both in-transit Doppler measurements as well as for atmospheric transmission spectra taken during transit.

astro-ph.EP↗

arXiv2017

Wireless Power Transfer and Data Collection in Wireless Sensor Networks

In a rechargeable wireless sensor network, the data packets are generated by sensor nodes at a specific data rate, and transmitted to a base station. Moreover, the base station transfers power to the nodes by using Wireless Power Transfer (WPT) to extend their battery life. However, inadequately scheduling WPT and data collection causes some of the nodes to drain their battery and have their data buffer overflow, while the other nodes waste their harvested energy, which is more than they need to transmit their packets. In this paper, we investigate a novel optimal scheduling strategy, called EHMDP, aiming to minimize data packet loss from a network of sensor nodes in terms of the nodes' energy consumption and data queue state information. The scheduling problem is first formulated by a centralized MDP model, assuming that the complete states of each node are well known by the base station. This presents the upper bound of the data that can be collected in a rechargeable wireless sensor network. Next, we relax the assumption of the availability of full state information so that the data transmission and WPT can be semi-decentralized. The simulation results show that, in terms of network throughput and packet loss rate, the proposed algorithm significantly improves the network performance.

eess.SP↗

arXiv2017

Quantifying Atom-scale Dopant Movement and Electrical Activation in Si:P Monolayers

Advanced hydrogen lithography techniques and low-temperature epitaxial overgrowth enable patterning of highly phosphorus-doped silicon (Si:P) monolayers (ML) with atomic precision. This approach to device fabrication has made Si:P monolayer systems a testbed for multiqubit quantum computing architectures and atomically precise 2-D superlattice designs whose behaviors are directly tied to the deterministic placement of single dopants. However, dopant segregation, diffusion, surface roughening, and defect formation during the encapsulation overgrowth introduce large uncertainties to the exact dopant placement and activation ratio. In this study, we develop a unique method by combining dopant segregation/diffusion models with sputter profiling simulation to monitor and control, at the atomic scale, dopant movement using room-temperature grown locking layers (LL). We explore the impact of LL growth rate, thickness, rapid thermal anneal, surface accumulation, and growth front roughness on dopant confinement, local crystalline quality, and electrical activation within Si:P 2-D systems. We demonstrate that dopant movement can be more efficiently suppressed by increasing the LL growth rate than by increasing LL thickness. We find that the dopant segregation length can be suppressed below a single Si lattice constant by increasing LL growth rates at room temperature while maintaining epitaxy. Although dopant diffusivity within the LL is found to remain high even below the hydrogen desorption temperature, we demonstrate that exceptionally sharp dopant confinement with high electrical quality within Si:P monolayers can be achieved by combining a high LL growth rate with a low-temperature LL rapid thermal anneal.

cond-mat.mtrl-sci↗