arXiv subjects

Alex Liu

Publications and source records attributed to Alex Liu.

34 records · Page 2Linked to original sources

arXiv2022

Hybrid Inlining: A Compositional and Context Sensitive Static Analysis Framework

Context sensitivity is essential for achieving the precision in inter-procedural static analysis. To be (fully) context sensitive, top-down analysis needs to fully inline all statements of the callees at each callsite, leading to statement explosion. Compositional analysis, which inlines summaries of the callees, scales up but often loses precision, as it is not strictly context sensitive. We propose a compositional and strictly context sensitive framework for static analysis. This framework is based on one key observation: a compositional static analysis often loses precision only on some critical statements that need to be analyzed context sensitively. Our approach hybridly inlines the critical statements and the summaries of non-critical statements of each callee, thus avoiding the re-analysis of non-critical ones. In addition, our analysis lazily summarizes the critical statements, by stopping propagating the critical statements once the calling context accumulated is adequate. Hybrid Inlining can be as precise as context sensitive top-down analysis. We have designed and implemented a pointer analysis based on this framework. It can analyze large Java programs from the Dacapo benchmark suite and industry in minutes. In our evaluation, compared to context insensitive analysis, Hybrid Inlining just brings 65% and 1% additional time overhead on Dacapo and industrial applications respectively.

cs.LO↗

arXiv2022

Merlin HugeCTR: GPU-accelerated Recommender System Training and Inference

In this talk, we introduce Merlin HugeCTR. Merlin HugeCTR is an open source, GPU-accelerated integration framework for click-through rate estimation. It optimizes both training and inference, whilst enabling model training at scale with model-parallel embeddings and data-parallel neural networks. In particular, Merlin HugeCTR combines a high-performance GPU embedding cache with an hierarchical storage architecture, to realize low-latency retrieval of embeddings for online model inference tasks. In the MLPerf v1.0 DLRM model training benchmark, Merlin HugeCTR achieves a speedup of up to 24.6x on a single DGX A100 (8x A100) over PyTorch on 4x4-socket CPU nodes (4x4x28 cores). Merlin HugeCTR can also take advantage of multi-node environments to accelerate training even further. Since late 2021, Merlin HugeCTR additionally features a hierarchical parameter server (HPS) and supports deployment via the NVIDIA Triton server framework, to leverage the computational capabilities of GPUs for high-speed recommendation model inference. Using this HPS, Merlin HugeCTR users can achieve a 5~62x speedup (batch size dependent) for popular recommendation models over CPU baseline implementations, and dramatically reduce their end-to-end inference latency.

cs.DC↗

arXiv2022

Regularized Graph Structure Learning with Semantic Knowledge for Multi-variates Time-Series Forecasting

Multivariate time-series forecasting is a critical task for many applications, and graph time-series network is widely studied due to its capability to capture the spatial-temporal correlation simultaneously. However, most existing works focus more on learning with the explicit prior graph structure, while ignoring potential information from the implicit graph structure, yielding incomplete structure modeling. Some recent works attempt to learn the intrinsic or implicit graph structure directly while lacking a way to combine explicit prior structure with implicit structure together. In this paper, we propose Regularized Graph Structure Learning (RGSL) model to incorporate both explicit prior structure and implicit structure together, and learn the forecasting deep networks along with the graph structure. RGSL consists of two innovative modules. First, we derive an implicit dense similarity matrix through node embedding, and learn the sparse graph structure using the Regularized Graph Generation (RGG) based on the Gumbel Softmax trick. Second, we propose a Laplacian Matrix Mixed-up Module (LM3) to fuse the explicit graph and implicit graph together. We conduct experiments on three real-word datasets. Results show that the proposed RGSL model outperforms existing graph forecasting algorithms with a notable margin, while learning meaningful graph structure simultaneously. Our code and models are made publicly available at https://github.com/alipay/RGSL.git.

cs.LG↗

arXiv2022

Ultra-Low Temperature Li/CFx Batteries Enabled by Fast-transport and Anion-pairing Liquefied Gas Electrolytes

Lithium fluorinated carbon is one of the most promising chemistries for high-energy-density primary energy storage systems in applications where rechargeability is not required. Though Li/CFx demonstrates high energy density under ambient conditions, achieving such a high energy density when exposed to subzero temperatures remains a challenge, particularly under high current density. Here, we report a liquefied gas electrolyte with an anion-pair solvation structure based on dimethyl ether with a low melting point and low viscosity, leading to high ionic conductivity between a wide temperature range. Besides that, through systematic X-ray photoelectron spectroscopy integrated with transmission electron microscopy characterizations, we evaluate the interface of CFx for low-temperature performance. We conclude that the fast transport and anion-pairing solvation structure of the electrolyte bring about reduced charge transfer resistance at low temperatures, which resulted in significantly enhanced performance of Li/CFx cells. Utilizing 50 mg/cm2 loading electrodes, the Li/CFx still displayed 1530 Wh/kg at reduced temperature. This work provides insights into the electrolyte design that may overcome the operational limits of batteries in extreme environments.

cond-mat.mtrl-sci↗

arXiv2022

Measuring and Clustering Network Attackers using Medium-Interaction Honeypots

Network honeypots are often used by information security teams to measure the threat landscape in order to secure their networks. With the advancement of honeypot development, today's medium-interaction honeypots provide a way for security teams and researchers to deploy these active defense tools that require little maintenance on a variety of protocols. In this work, we deploy such honeypots on five different protocols on the public Internet and study the intent and sophistication of the attacks we observe. We then use the information gained to develop a clustering approach that identifies correlations in attacker behavior to discover IPs that are highly likely to be controlled by a single operator, illustrating the advantage of using these honeypots for data collection.

cs.CR↗

arXiv2021

Sparse deep computer-generated holography for optical microscopy

Computer-generated holography (CGH) has broad applications such as direct-view display, virtual and augmented reality, as well as optical microscopy. CGH usually utilizes a spatial light modulator that displays a computer-generated phase mask, modulating the phase of coherent light in order to generate customized patterns. The algorithm that computes the phase mask is the core of CGH and is usually tailored to meet different applications. CGH for optical microscopy usually requires 3D accessibility (i.e., generating overlapping patterns along the $z$-axis) and micron-scale spatial precision. Here, we propose a CGH algorithm using an unsupervised generative model designed for optical microscopy to synthesize 3D selected illumination. The algorithm, named sparse deep CGH, is able to generate sparsely distributed points in a large 3D volume with higher contrast than conventional CGH algorithms.

physics.optics↗

arXiv2020

A Tale of Evil Twins: Adversarial Inputs versus Poisoned Models

Despite their tremendous success in a range of domains, deep learning systems are inherently susceptible to two types of manipulations: adversarial inputs -- maliciously crafted samples that deceive target deep neural network (DNN) models, and poisoned models -- adversely forged DNNs that misbehave on pre-defined inputs. While prior work has intensively studied the two attack vectors in parallel, there is still a lack of understanding about their fundamental connections: what are the dynamic interactions between the two attack vectors? what are the implications of such interactions for optimizing existing attacks? what are the potential countermeasures against the enhanced attacks? Answering these key questions is crucial for assessing and mitigating the holistic vulnerabilities of DNNs deployed in realistic settings. Here we take a solid step towards this goal by conducting the first systematic study of the two attack vectors within a unified framework. Specifically, (i) we develop a new attack model that jointly optimizes adversarial inputs and poisoned models; (ii) with both analytical and empirical evidence, we reveal that there exist intriguing "mutual reinforcement" effects between the two attack vectors -- leveraging one vector significantly amplifies the effectiveness of the other; (iii) we demonstrate that such effects enable a large design spectrum for the adversary to enhance the existing attacks that exploit both vectors (e.g., backdoor attacks), such as maximizing the attack evasiveness with respect to various detection methods; (iv) finally, we discuss potential countermeasures against such optimized attacks and their technical challenges, pointing to several promising research directions.

cs.LG↗

arXiv2020

Secret Sharing based Secure Regressions with Applications

Nowadays, the utilization of the ever expanding amount of data has made a huge impact on web technologies while also causing various types of security concerns. On one hand, potential gains are highly anticipated if different organizations could somehow collaboratively share their data for technological improvements. On the other hand, data security concerns may arise for both data holders and data providers due to commercial or sociological concerns. To make a balance between technical improvements and security limitations, we implement secure and scalable protocols for multiple data holders to train linear regression and logistic regression models. We build our protocols based on the secret sharing scheme, which is scalable and efficient in applications. Moreover, our proposed paradigm can be generalized to any secure multiparty training scenarios where only matrix summation and matrix multiplications are used. We demonstrate our approach by experiments which shows the scalability and efficiency of our proposed protocols, and finally present its real-world applications.

cs.LG↗

arXiv2017

Dynamic Scaling of Virtualized, Distributed Service Chains: A Case Study of IMS

The emerging paradigm of network function virtualization advocates deploying virtualized network functions (VNF) on standard virtualization platforms for significant cost reduction and management flexibility. There have been system designs for managing dynamic deployment and scaling of VNF service chains within one cloud data center. Many real-world network services involve geo-distributed service chains, with prominent examples of mobile core networks and IMSs (IP Multimedia Subsystems). Virtualizing these service chains requires efficient coordination of VNF deployment across different geo-distributed data centers over time, calling for new management system design. This paper designs a dynamic scaling system for geo-distributed VNF service chains, using the case of an IMS. IMSs are widely used subsystems for delivering multimedia services among mobile users in a 3G/4G network, whose virtualization has been broadly advocated in the industry for reducing cost, improving network usage efficiency and enabling dynamic network topology reconfiguration for performance optimization. Our scaling system design caters to key control-plane and data-plane service chains in an IMS, combining proactive and reactive approaches for timely, cost-effective scaling of the service chains. We evaluate our system design using real-world experiments on both emulated platforms and geo-distributed clouds.

cs.NI↗

arXiv2016

Online Scaling of NFV Service Chains across Geo-distributed Datacenters

Network Function Virtualization (NFV) is an emerging paradigm that turns hardware-dependent implementation of network functions (i.e., middleboxes) into software modules running on virtualized platforms, for significant cost reduction and ease of management. Such virtual network functions (VNFs) commonly constitute service chains, to provide network services that traffic flows need to go through. Efficient deployment of VNFs for network service provisioning is key to realize the NFV goals. Existing efforts on VNF placement mostly deal with offline or one-time placement, ignoring the fundamental, dynamic deployment and scaling need of VNFs to handle practical time-varying traffic volumes. This work investigates dynamic placement of VNF service chains across geo-distributed datacenters to serve flows between dispersed source and destination pairs, for operational cost minimization of the service chain provider over the entire system span. An efficient online algorithm is proposed, which consists of two main components: (1) A regularization-based approach from online learning literature to convert the offline optimal deployment problem into a sequence of one-shot regularized problems, each to be efficiently solved in one time slot; (2) An online dependent rounding scheme to derive feasible integer solutions from the optimal fractional solutions of the one-shot problems, and to guarantee a good competitive ratio of the online algorithm over the entire time span. We verify our online algorithm with solid theoretical analysis and trace-driven simulations under realistic settings.

cs.NI↗

arXiv2016

Online VNF Scaling in Datacenters

Network Function Virtualization (NFV) is a promising technology that promises to significantly reduce the operational costs of network services by deploying virtualized network functions (VNFs) to commodity servers in place of dedicated hardware middleboxes. The VNFs are typically running on virtual machine instances in a cloud infrastructure, where the virtualization technology enables dynamic provisioning of VNF instances, to process the fluctuating traffic that needs to go through the network functions in a network service. In this paper, we target dynamic provisioning of enterprise network services - expressed as one or multiple service chains - in cloud datacenters, and design efficient online algorithms without requiring any information on future traffic rates. The key is to decide the number of instances of each VNF type to provision at each time, taking into consideration the server resource capacities and traffic rates between adjacent VNFs in a service chain. In the case of a single service chain, we discover an elegant structure of the problem and design an efficient randomized algorithm achieving a e/(e-1) competitive ratio. For multiple concurrent service chains, an online heuristic algorithm is proposed, which is O(1)-competitive. We demonstrate the effectiveness of our algorithms using solid theoretical analysis and trace-driven simulations.

cs.NI↗

arXiv2011

Two new accreting, pulsating white dwarfs: SDSS J1457+51 and BW Sculptoris

We report the discovery of rapid periodic signals in the light curves of two cataclysmic variables with prominent white-dwarf components in their spectra, SDSS J1457+51 and BW Sculptoris. These stars therefore appear to be new members of the GW Lib class of variable star, in which the fast periodic (and non-commensurate with the orbital period) signals are believed to arise from non-radial pulsations in the underlying white dwarf. The power spectra of both stars show complex signals with primary periods near 10 and 20 minutes. These signals change in frequency by a few percent on a timescale of weeks or less, and probably contain an internal fine structure unresolved by our observations. We also detect double-humped waves signifying the underlying orbital periods, near 78 minutes for both stars. In addition, BW Scl shows a transient but powerful signal with a period near 87 minutes, a quiescent superhump. The 11 % excess over the orbital period is difficult to understand, and may arise from an eccentric instability near the 2:1 resonance in the accretion disc.

astro-ph.SR↗

arXiv2006

V4745 Sgr -- a nova above the period gap and an intermediate polar candidate

A period analysis of CCD unfiltered photometry of V4745 Sgr (Nova Sgr 2003 #1) performed during 23 nights in the years 2003 - 2005 is presented. The photometric data are modulated with a period of 0.20782 +- 0.00003 d ($4.98768 +- 0.00072 h). Following the shape of the phased light curve and the presence of the periodicity in all data sets with no apparent change in its value, we interpret this periodicity as orbital in nature and this is consistent with a cataclysmic variable above the period gap. We found a probable short-term periodicity of 0.017238 +- 0.000037 d (24.82272 +- 0.05328 m) which we interpret as the probable spin period of the white dwarf or the beat period between the orbital and spin period. Therefore, we propose that nova V4745 Sgr should be classified as an intermediate polar candidate, supporting the proposed link between transition-oscillation novae and intermediate polars. The mass-period relation for cataclysmic variables yields a secondary mass of about 0.52 +- 0.05 M_solar.

astro-ph↗

arXiv2006

Nova V4743 Sagittarii 2002: An Intermediate Polar Candidate

We present the results of 11 nights of CCD unfiltered photometry of V4743 Sgr (Nova Sgr 2002 # 3) from 2003 and 2005. We find two periods of 0.2799 d ~ 6.7 h and 0.01642 d ~ 24 min in the 2005 data. The long period is also present in the 2003 data, but only weak evidence of the shorter period is found in this year. The 24-min period is somewhat longer than the 22-min period, which was detected from X-ray observations. We suggest that the 6.7-h periodicity represents the orbital period of the underlying binary system and that the 24-min period is the beat periodicity between the orbital period and the X-ray period, which is presumably the spin period of the white dwarf. Thus, V4743 Sgr should be classified as an intermediate polar (DQ Her star). About six months after the nova outburst, the optical light curve of V4743 Sgr seemed to show quasi-periodic oscillations, which are typical of the transient phase in classical nova. Therefore, our results support the previous suggestion that the trans ition phase in novae may be related to intermediate polars.

astro-ph↗

arXiv2006

The 2005 July Superoutburst of the Dwarf Nova 2QZ J021927.9-304545: the SU UMa Nature Confirmed

We report on time-resolved photometry of the 2005 July superoutburst of the dwarf nova, 2QZ J021927.9-304545. The resultant light curves showed conspicuous superhumps with a period of 0.081113(19) days, confirming the SU UMa nature of the object. Although we missed the maximum phase of the outburst, the amplitude of the superoutburst well exceeded 5 mag. This value is slightly larger than that of typical SU UMa-type dwarf novae. The superhump period decreased as time elapsed, as can be seen in most SU UMa-type dwarf novae. Based on the archive of ASAS-3, the recurrence time of a superoutburst of the variable turned out to be about 400 days. This value is typical of well known SU UMa stars. The distance to this system was roughly estimated as 370(+20, -60) pc using an empirical relation.

astro-ph↗

arXiv2005

Detection of orbital and superhump periods in Nova V2574 Ophiuchi (2004)

We present the results of 37 nights of CCD unfiltered photometry of nova V2574 Oph (2004) from 2004 and 2005. We find two periods of 0.14164 d (~3.40 h) and 0.14773 d (~3.55 h) in the 2005 data. The 2004 data show variability on a similar timescale, but no coherent periodicity was found. We suggest that the longer periodicity is the orbital period of the underlying binary system and that the shorter period represents a negative superhump. The 3.40 h period is about 4% shorter than the orbital period and obeys the relation between superhump period deficit and binary period. The detection of superhumps in the light curve is evidence of the presence of a precessing accretion disk in this binary system shortly after the nova outburst. From the maximum magnitude - rate of decline relation, we estimate the decay rate t_2 = 17+/-4 d and a maximum absolute visual magnitude of M_Vmax = -7.7+/-1.7 mag.

astro-ph↗