arXiv subjects

Jie Lin

Publications and source records attributed to Jie Lin.

At least 127 records · Page 7Linked to original sources

arXiv2020

Trusted detector noise analysis for discrete modulation schemes of continuous-variable quantum key distribution

Discrete-modulated continuous-variable quantum key distribution protocols are promising candidates for large-scale deployment due to the large technological overlap with deployed modern optical communication devices. The security of discrete modulation schemes has previously analyzed in the ideal detector scenario in the asymptotic limit. In this work, we calculate asymptotic key rates against collective attacks in the trusted detector noise scenario. Our results show that we can thus cut out most of the effect of detector noise and obtain asymptotic key rates similar to those had we access to ideal detectors.

quant-ph↗

arXiv2020

The energetic thermonuclear bursts in SAX J1712.6-3739

The neutron star low-mass X-ray binary SAX J1712.6-3739 is known for its very long and hard thermonuclear bursts in past X-ray observations. Its thermonuclear bursts are so energetic that they can even last for tens of minutes in the hard X-ray observations of Swift/BAT (E > 15 keV). For exploring the natures of SAX J1712.6-3739 and its thermonuclear bursts, we analyzed its four bursts which were all captured by Swift/BAT. The peak flux and fluence for these bursts are derived by applying the archive observation data of Swift/BAT and Swift/XRT. Based on the peak fluxes observed by Swift, the distance of SAX J1712.6-3739 is estimated as 4.6 kpc, which is much closer than the estimate by the BeppoSAX observations. The corresponding absolute magnitude and average accretion rate agree its ultra-compact X-ray binary nature. By the measurements of effective duration, we determined that the 2010 burst corresponds to a normal X-ray burst, the 2011 burst is consistent with an intermediate-duration burst while the 2014 and the 2018 bursts are more energetic than common intermediate-duration bursts but less energetic than superbursts. Since the average mass accretion rate of SAX J1712.6-3739 was only 0.6% Eddington accretion rate, and current theory predict no carbon production in the bursters under very low accretion rate. Therefore, the 2014 and the 2018 bursts are very likely to be deep helium bursts. In the Swift observations, SAX J1712.6-3739 has shown at least three different levels of duration in thermonuclear bursts. They were possibly induced by variable accretion rates, because their ignition column depths roughly follow the prediction of ignition models for pure helium bursts at corresponding accretion rates.

astro-ph.HE↗

arXiv2020

Towards the AlexNet Moment for Homomorphic Encryption: HCNN, theFirst Homomorphic CNN on Encrypted Data with GPUs

Deep Learning as a Service (DLaaS) stands as a promising solution for cloud-based inference applications. In this setting, the cloud has a pre-learned model whereas the user has samples on which she wants to run the model. The biggest concern with DLaaS is user privacy if the input samples are sensitive data. We provide here an efficient privacy-preserving system by employing high-end technologies such as Fully Homomorphic Encryption (FHE), Convolutional Neural Networks (CNNs) and Graphics Processing Units (GPUs). FHE, with its widely-known feature of computing on encrypted data, empowers a wide range of privacy-concerned applications. This comes at high cost as it requires enormous computing power. In this paper, we show how to accelerate the performance of running CNNs on encrypted data with GPUs. We evaluated two CNNs to classify homomorphically the MNIST and CIFAR-10 datasets. Our solution achieved a sufficient security level (> 80 bit) and reasonable classification accuracy (99%) and (77.55%) for MNIST and CIFAR-10, respectively. In terms of latency, we could classify an image in 5.16 seconds and 304.43 seconds for MNIST and CIFAR-10, respectively. Our system can also classify a batch of images (> 8,000) without extra overhead.

cs.CR↗

arXiv2020

Role-Wise Data Augmentation for Knowledge Distillation

Knowledge Distillation (KD) is a common method for transferring the ``knowledge'' learned by one machine learning model (the \textit{teacher}) into another model (the \textit{student}), where typically, the teacher has a greater capacity (e.g., more parameters or higher bit-widths). To our knowledge, existing methods overlook the fact that although the student absorbs extra knowledge from the teacher, both models share the same input data -- and this data is the only medium by which the teacher's knowledge can be demonstrated. Due to the difference in model capacities, the student may not benefit fully from the same data points on which the teacher is trained. On the other hand, a human teacher may demonstrate a piece of knowledge with individualized examples adapted to a particular student, for instance, in terms of her cultural background and interests. Inspired by this behavior, we design data augmentation agents with distinct roles to facilitate knowledge distillation. Our data augmentation agents generate distinct training data for the teacher and student, respectively. We find empirically that specially tailored data points enable the teacher's knowledge to be demonstrated more effectively to the student. We compare our approach with existing KD methods on training popular neural architectures and demonstrate that role-wise data augmentation improves the effectiveness of KD over strong prior approaches. The code for reproducing our results can be found at https://github.com/bigaidream-projects/role-kd

cs.LG↗

arXiv2020

GRB 111209A/SN 2011kl: Collapse of a supramassive magnetar with r-mode oscillation and fall-back accretion onto a newborn black hole

Ultra-long-duration gamma-ray burst GRB 111209A was found to be associated with a very luminous supernovae (SNe) SN 2011kl. The physics of GRB 111209A/SN 2011kl has been extensively studied in the literures, but does not settle down yet. By investigating in detail the characteristics of the X-ray light curve of GRB 111209A, coupled with the temporal and spectral features observed in SN 2011kl, we argue that a short-living supramassive magnetar can be responsible for the initial shallow X-ray emission. Then the electromagnetic extraction of spin energy from a black hole results in the steeply declining X-ray flux when the magnetar collapses into a black hole (BH). A fraction of the envelope materials falls back and activates the accretion onto the newborn BH, which produces the X-ray rebrightening bump at late times. During this process, a centrifugally driven baryon-rich quasi-isotropic Blandford \& Payne outflow from the revived accretion disk deposits its kinetic energy on the SN ejecta, which powers luminous SN 2011kl. Finally, we place a limitation on the magnetar's physical parameters based on the observations.

astro-ph.HE↗

arXiv2020

Deeply Activated Salient Region for Instance Search

The performance of instance search depends heavily on the ability to locate and describe a wide variety of object instances in a video/image collection. Due to the lack of proper mechanism in locating instances and deriving feature representation, instance search is generally only effective for retrieving instances of known object categories. In this paper, a simple but effective instance-level feature representation is presented. Different from other approaches, the issues in class-agnostic instance localization and distinctive feature representation are considered. The former is achieved by detecting salient instance regions from an image by a layer-wise back-propagation process. The back-propagation starts from the last convolution layer of a pre-trained CNN that is originally used for classification. The back-propagation proceeds layer-by-layer until it reaches the input layer. This allows the salient instance regions in the input image from both known and unknown categories to be activated. Each activated salient region covers the full or more usually a major range of an instance. The distinctive feature representation is produced by average-pooling on the feature map of certain layer with the detected instance region. Experiments show that such kind of feature representation demonstrates considerably better performance over most of the existing approaches. In addition, we show that the proposed feature descriptor is also suitable for content-based image search.

cs.CV↗

arXiv2020

From single-cell variability to population growth

Single-cell experiments have revealed cell-to-cell variability in generation times and growth rates for genetically identical cells. Theoretical models relating the fluctuating generation times of single cells to the population growth rate are usually based on the assumption that the generation times of mother and daughter cells are uncorrelated. This assumption, however, is inconsistent with the exponential growth of cell volume in time observed for many cell types. Here we develop a more general and biologically relevant model in which cells grow exponentially and generation times are correlated in a manner which controls cell size. In addition to the fluctuating generation times, we also allow the single-cell growth rates to fluctuate and account for their correlations across the lineage tree. Surprisingly, we find that the population growth rate only depends on the distribution of single-cell growth rates and their correlations.

q-bio.PE↗

arXiv2019

Asymptotic security analysis of discrete-modulated continuous-variable quantum key distribution

Continuous-variable quantum key distribution (CV QKD) protocols with discrete modulation are interesting due to their experimental simplicity and their great potential for massive deployment in the quantum-secured networks, but their security analysis is less advanced than that of Gaussian modulation schemes. In this work, we apply a numerical method to analyze the security of discrete-modulation protocols against collective attacks in the asymptotic limit, paving the way for a full security proof with finite-size effects. While our method is general for discrete-modulation schemes, we focus on two variants of the CV QKD protocol with quaternary modulation. Interestingly, thanks to the tightness of our proof method, we show that this protocol is capable of achieving much higher key rates over significantly longer distances with experimentally feasible parameters compared with previous security proofs of binary and ternary modulation schemes and also yielding key rates comparable to Gaussian modulation schemes. Furthermore, as our security analysis method is versatile, it allows us to evaluate variations of the discrete-modulated protocols, including direct and reverse reconciliation, and postselection strategies. In particular, we demonstrate that postselection of data in combination with reverse reconciliation can improve the key rates.

quant-ph↗

arXiv2019

Quantum search for unknown number of target items hybridizing the fixed-point method with the trail-and-error method

For the unsorted database quantum search with the unknown fraction $λ$ of target items, there are mainly two kinds of methods, i.e., fixed-point or trail-and-error. (i) In terms of the fixed-point method, Yoder et al. [Phys. Rev. Lett. 113, 210501 (2014)] claimed that the quadratic speedup over classical algorithms has been achieved. However, in this paper, we point out that this is not the case, because the query complexity of Yoder's algorithm is actually in $O(1/\sqrt{λ_0})$ rather than $O(1/\sqrtλ)$, where $λ_0$ is a known lower bound of $λ$. (ii) In terms of the trail-and-error method, currently the algorithm without randomness has to take more than 1 times queries or iterations than the algorithm with randomly selected parameters. For the above problems, we provide the first hybrid quantum search algorithm based on the fixed-point and trail-and-error methods, where the matched multiphase Grover operations are trialed multiple times and the number of iterations increases exponentially along with the number of trials. The upper bound of expected queries as well as the optimal parameters are derived. Compared with Yoder's algorithm, the query complexity of our algorithm indeed achieves the optimal scaling in $λ$ for quantum search, which reconfirms the practicality of the fixed-point method. In addition, our algorithm also does not contain randomness, and compared with the existing deterministic algorithm, the query complexity can be reduced by about 1/3. Our work provides an new idea for the research on fixed-point and trial-and-error quantum search.

quant-ph↗

arXiv2019

TEA-DNN: the Quest for Time-Energy-Accuracy Co-optimized Deep Neural Networks

Embedded deep learning platforms have witnessed two simultaneous improvements. First, the accuracy of convolutional neural networks (CNNs) has been significantly improved through the use of automated neural-architecture search (NAS) algorithms to determine CNN structure. Second, there has been increasing interest in developing hardware accelerators for CNNs that provide improved inference performance and energy consumption compared to GPUs. Such embedded deep learning platforms differ in the amount of compute resources and memory-access bandwidth, which would affect performance and energy consumption of CNNs. It is therefore critical to consider the available hardware resources in the network architecture search. To this end, we introduce TEA-DNN, a NAS algorithm targeting multi-objective optimization of execution time, energy consumption, and classification accuracy of CNN workloads on embedded architectures. TEA-DNN leverages energy and execution time measurements on embedded hardware when exploring the Pareto-optimal curves across accuracy, execution time, and energy consumption and does not require additional effort to model the underlying hardware. We apply TEA-DNN for image classification on actual embedded platforms (NVIDIA Jetson TX2 and Intel Movidius Neural Compute Stick). We highlight the Pareto-optimal operating points that emphasize the necessity to explicitly consider hardware characteristics in the search process. To the best of our knowledge, this is the most comprehensive study of Pareto-optimal models across a range of hardware platforms using actual measurements on hardware to obtain objective values.

cs.NE↗

arXiv2019

A search for short-term hard X-ray bursts in the direction of the repeating FRB 121102

The nature of fast radio bursts (FRBs), which occurs on millisecond time scales in the radio band, has not been well-understood. Among their unknown observational properties are their broadband spectra and persistent and transient multi-wavelength counterparts. Well-localized FRBs provide us the opportunity to address these issues in archival observations. We have performed searches for 15-150 keV hard X-ray bursts on time scales as short as millisecond in the direction of the repeating FRB 121102 (with a spacial resolution of a few arcminutes) in the archival Swift/BAT data during the period between October 2016 and September 2017. We have found no significant (5 $σ$) hard X-ray bursts in the direction of the repeating FRB. We have derived an upper limit of the hard X-ray (15--150 keV) flux of any X-ray bursts on 1 ms time scale of around $1.01 \times 10^{-7}$erg cm$^{-2}$s$^{-1}$, if assuming a photo-index of 2 for potential X-ray flares in X-ray band. A plausible scenario for the repeating FRB as being associated with \emph{magnetar giant flare} is still far below the upper limit.

astro-ph.HE↗

arXiv2019

A*3D Dataset: Towards Autonomous Driving in Challenging Environments

With the increasing global popularity of self-driving cars, there is an immediate need for challenging real-world datasets for benchmarking and training various computer vision tasks such as 3D object detection. Existing datasets either represent simple scenarios or provide only day-time data. In this paper, we introduce a new challenging A*3D dataset which consists of RGB images and LiDAR data with significant diversity of scene, time, and weather. The dataset consists of high-density images ($\approx~10$ times more than the pioneering KITTI dataset), heavy occlusions, a large number of night-time frames ($\approx~3$ times the nuScenes dataset), addressing the gaps in the existing datasets to push the boundaries of tasks in autonomous driving research to more challenging highly diverse environments. The dataset contains $39\text{K}$ frames, $7$ classes, and $230\text{K}$ 3D object annotations. An extensive 3D object detection benchmark evaluation on the A*3D dataset for various attributes such as high density, day-time/night-time, gives interesting insights into the advantages and limitations of training and testing 3D object detection in real-world setting.

cs.CV↗

arXiv2019

Quantum-enhanced least-square support vector machine: simplified quantum algorithm and sparse solutions

Quantum algorithms can enhance machine learning in different aspects. Here, we study quantum-enhanced least-square support vector machine (LS-SVM). Firstly, a novel quantum algorithm that uses continuous variable to assist matrix inversion is introduced to simplify the algorithm for quantum LS-SVM, while retaining exponential speed-up. Secondly, we propose a hybrid quantum-classical version for sparse solutions of LS-SVM. By encoding a large dataset into a quantum state, a much smaller transformed dataset can be extracted using quantum matrix toolbox, which is further processed in classical SVM. We also incorporate kernel methods into the above quantum algorithms, which uses both exponential growth Hilbert space of qubits and infinite dimensionality of continuous variable for quantum feature maps. The quantum LS-SVM exploits quantum properties to explore important themes for SVM such as sparsity and kernel methods, and stresses its quantum advantages ranging from speed-up to the potential capacity to solve classically difficult machine learning tasks.

quant-ph↗

arXiv2019

CaRENets: Compact and Resource-Efficient CNN for Homomorphic Inference on Encrypted Medical Images

Convolutional neural networks (CNNs) have enabled significant performance leaps in medical image classification tasks. However, translating neural network models for clinical applications remains challenging due to data privacy issues. Fully Homomorphic Encryption (FHE) has the potential to address this challenge as it enables the use of CNNs on encrypted images. However, current HE technology poses immense computational and memory overheads, particularly for high-resolution images such as those seen in the clinical context. We present CaRENets: Compact and Resource-Efficient CNNs for high performance and resource-efficient inference on high-resolution encrypted images in practical applications. At the core, CaRENets comprises a new FHE compact packing scheme that is tightly integrated with CNN functions. CaRENets offers dual advantages of memory efficiency (due to compact packing of images and CNN activations) and inference speed (due to the reduction in the number of ciphertexts created and the associated mathematical operations) over standard interleaved packing schemes. We apply CaRENets to perform homomorphic abnormality detection with 80-bit security level in two clinical conditions - Retinopathy of Prematurity (ROP) and Diabetic Retinopathy (DR). The ROP dataset comprises 96 x 96 grayscale images, while the DR dataset comprises 256 x 256 RGB images. We demonstrate over 45x improvement in memory efficiency and 4-5x speedup in inference over the interleaved packing schemes. As our approach enables memory-efficient low-latency HE inference without imposing additional communication burden, it has implications for practical and secure deep learning inference in clinical imaging.

cs.CR↗

arXiv2019

Quasi-derivations of Lie-Yamaguti algebras

The concept of derivation for Lie-Yamaguti algebras is generalized in this paper. A quasi-derivation of an LY-algebra is embedded as derivation in a larger LY-algebra. The relationship between quasi-derivations and robustness of Lie-Yamaguti algebras has been studied.

math.RA↗

arXiv2019

Dataflow-based Joint Quantization of Weights and Activations for Deep Neural Networks

This paper addresses a challenging problem - how to reduce energy consumption without incurring performance drop when deploying deep neural networks (DNNs) at the inference stage. In order to alleviate the computation and storage burdens, we propose a novel dataflow-based joint quantization approach with the hypothesis that a fewer number of quantization operations would incur less information loss and thus improve the final performance. It first introduces a quantization scheme with efficient bit-shifting and rounding operations to represent network parameters and activations in low precision. Then it restructures the network architectures to form unified modules for optimization on the quantized model. Extensive experiments on ImageNet and KITTI validate the effectiveness of our model, demonstrating that state-of-the-art results for various tasks can be achieved by this quantized model. Besides, we designed and synthesized an RTL model to measure the hardware costs among various quantization methods. For each quantization operation, it reduces area cost by about 15 times and energy consumption by about 9 times, compared to a strong baseline.

cs.LG↗

arXiv2019

The relation between outburst rate and orbital period in low-mass X-ray binary transients

We have investigated the outburst properties of low-mass X-ray binary transients (LMXBTs) based on a comprehensive study of the outbursts observed in the past few decades. The outburst rates were estimated based on the X-ray monitoring data from Swift/BAT, RXTE/ASM and MAXI, and previous reports in the literature. We found that almost all LMXBTs with the orbital period below $\sim$12 hr showed only one outburst in these observations. There are systematic difference in the outburst rate between long-period ($P_{\rm orb} \gtrsim$ 12 hr) and short-period ($P_{\rm orb} \lesssim$ 12 hr) systems. We infer that mass transfer rate is responsible for the systematic difference, since the disk instability model (DIM) suggested that the mass transfer rate is a key factor affecting the quiescence time. The difference in outburst rate between long-period and short-period LMXBTs is probably due to the different mass transfer mechanism at different evolutionary stages of the donors. Based on the evolutionary tracks of single stars, we derived the critical orbital period for X-ray binaries that harbor a subgiant donor in various metallicity. The critical orbital period ($P_{\rm orb,crit}=$12.4 hr) is consistent with the above orbital period boundary obtained from the statistics of outburst rates. Furthermore, we found a negative correlation between the outburst rate and the orbital period in the samples for which the luminosity class of the donor star is III/IV. The best-fitting power-law index for the black hole subsamples is roughly consistent with the theoretical prediction for those systems with a donor star evolved off the main sequence.

astro-ph.HE↗

arXiv2018

Complementary-multiphase quantum search for all numbers of target items

Grover's algorithm achieves a quadratic speedup over classical algorithms, but it is considered necessary to know the value of $λ$ exactly [Phys. Rev. Lett. 95, 150501 (2005); Phys. Rev. Lett. 113, 210501 (2014)], where $λ$ is the fraction of target items in the database. In this paper, we find out that the Grover algorithm can actually apply to the case where one can identify the range that $λ$ belongs to from a given series of disjoint ranges. However, Grover's algorithm still cannot maintain high success probability when there exist multiple target items. For this problem, we proposed a complementary-multiphase quantum search algorithm, %with general iterations, in which multiple phases complement each other so that the overall high success probability can be maintained. Compared to the existing algorithms, in the case defined above, for the first time our algorithm achieves the following three goals simultaneously: (1) the success probability can be no less than any given value between 0 and 1, (2) the algorithm is applicable to the entire range of $λ$, and (3) the number of iterations is almost the same as that of Grover's algorithm. Especially compared to the optimal fixed-point algorithm [Phys. Rev. Lett. 113, 210501 (2014)], our algorithm uses fewer iterations to achieve success probability greater than 82.71\%, e.g., when the minimum success probability is required to be 99.25\%, the number of iterations can be reduced by 50\%.

quant-ph↗