arXiv subjects

Jun Yang

Publications and source records attributed to Jun Yang.

At least 325 records · Page 18Linked to original sources

arXiv2021

State-domain Change Point Detection for Nonlinear Time Series Regression

Change point detection in time series has attracted substantial interest, but most of the existing results have been focused on detecting change points in the time domain. This paper considers the situation where nonlinear time series have potential change points in the state domain. We apply a density-weighted anti-symmetric kernel function to the state domain and therefore propose a nonparametric procedure to test the existence of change points. When the existence of change points is affirmative, we further introduce an algorithm to estimate the number of change points together with their locations. Theoretical results of the proposed detection and estimation procedures are given and a real dataset is used to illustrate our methods.

stat.ME↗

arXiv2021

Planetary climate under extremely high vertical diffusivity

Planets with large moon(s) or those in the habitable zone of low-mass stars may experience much stronger tidal force and tide-induced ocean mixing than that on Earth. Thus, the vertical diffusivity (or, more precisely, diapycnal diffusivity) on such planets, which represents the strength of vertical mixing in the ocean, would be greater than that on Earth. In this study, we explore the effects of extremely high diffusivity on the ocean circulation and surface climate of Earth-like planets in one asynchronous rotation orbit. The response of planetary climate to 10 and 100 times greater vertical diffusivity than that found on Earth is investigated using a fully coupled atmosphere-ocean general circulation model. In order to perform a clear comparison with the climate of modern Earth, Earth's orbit, land-sea configuration, and present levels of greenhouse gases are included in the simulations. We find that a larger vertical diffusivity intensifies the meridional overturning circulation (MOC) in the ocean, which transports more heat to polar regions and melts sea ice there. Feedback associated with sea ice, clouds, and water vapor act to further amplify surface warming. When the vertical diffusivity is 10 (100) times the present-day value, the magnitude of MOC increases by $\approx$3 (18) times, and the global-mean surface temperature increases by $\approx$4$^{\circ}$C (10$^{\circ}$C). This study quantifies the climatic effect of an extremely strong vertical diffusivity and confirms an indirect link between planetary orbit, tidal mixing, ocean circulation, and surface climate. Our results suggest a moderate effect of varying vertical ocean mixing on planetary climate.

astro-ph.EP↗

arXiv2021

A long-lived compact jet in the black hole X-ray binary candidate AT2019wey

AT2019wey is a transient discovered by the Asteroid Terrestrial-impact Last Alert System (ATLAS) survey in December of 2019. Follow-up optical, radio, and X-ray observations led to classification of this source as a Galactic black hole X-ray binary. We carried out one-epoch 6.7 GHz European VLBI (Very Long Baseline Interferometry) Network (EVN) and two-epoch multiple-frequency (1.6, 4.5, 6.7 GHz) Very Long Baseline Array (VLBA) observations within a year after its discovery. These observations reveal a fading and flat-spectrum radio source with no discernible motion. These features suggest the detection of a compact jet. The source appears resolved at milliarcsecond scales, and the source angular size versus frequency trend is consistent with scatter broadening. This allows us to constrain the lower limit of the source distance to 6 kpc, if the scattering medium is in a Galactic spiral arm. For a source location larger than 3 kpc, the estimated upper limit of the peculiar velocity, relative to the local standard of rest, suggests the asymmetric natal kick may have occurred during the black hole formation stage.

astro-ph.HE↗

arXiv2021

Constraints on the number of X-ray Pulsars in IC 10 from a deep XMM-Newton Observation

We report the most sensitive search yet for X-ray pulsars in the dwarf starburst galaxy IC 10, which is known to contain a population of young high mass X-ray binaries. We searched for pulsations in 207 point-like X-ray sources in the direction of IC 10 by a 2012 \xmm~observation with a total exposure time of 134.5 ks. Pulsation searches in faint objects can be sensitive to the energy bands of the light curves, and the source and background extraction areas. We analyzed separately the PN and MOS barycenter corrected 0.2-12 keV data, with good time interval filtering. Different schemes for source and background extraction were compared, and the search was repeated in the narrower 0.5-8 keV energy band to increase the signal-to-noise ratio. For the most conservative parameters, 5 point sources produced significant peaks in the Lomb-Scargle periodogram (99\% significance, single trial, assuming white noise). A similar number of different candidates result from alternative analyses. A $\sim$4100 s period seen in all 3 instruments for the black hole (BH) + Wolf-Rayet (WR) binary IC 10 X-1 is probably due to red noise of astrophysical origin. Considering the periods, luminosities, and spatial distribution of the pulsar candidates in the direction of IC 10, they do not belong to the same distribution as the ones in the Magellanic Clouds and Milky Way. This result holds even if the candidates are spurious, since if the Small Magellanic Cloud were placed at the distance of IC 10, we would expect to see $\sim$5 pulsars at $L_x>10^{36}$ \lx ~inside the $D_{25}$ contour, and their periods would be of order 100 seconds, rather than the mostly $\sim$1 s periods for the candidates reported here, which lie outside the main body of the galaxy.

astro-ph.HE↗

arXiv2021

Celebrating Diversity in Shared Multi-Agent Reinforcement Learning

Recently, deep multi-agent reinforcement learning (MARL) has shown the promise to solve complex cooperative tasks. Its success is partly because of parameter sharing among agents. However, such sharing may lead agents to behave similarly and limit their coordination capacity. In this paper, we aim to introduce diversity in both optimization and representation of shared multi-agent reinforcement learning. Specifically, we propose an information-theoretical regularization to maximize the mutual information between agents' identities and their trajectories, encouraging extensive exploration and diverse individualized behaviors. In representation, we incorporate agent-specific modules in the shared neural network architecture, which are regularized by L1-norm to promote learning sharing among agents while keeping necessary diversity. Empirical results show that our method achieves state-of-the-art performance on Google Research Football and super hard StarCraft II micromanagement tasks.

cs.LG↗

arXiv2021

Average-Reward Reinforcement Learning with Trust Region Methods

Most of reinforcement learning algorithms optimize the discounted criterion which is beneficial to accelerate the convergence and reduce the variance of estimates. Although the discounted criterion is appropriate for certain tasks such as financial related problems, many engineering problems treat future rewards equally and prefer a long-run average criterion. In this paper, we study the reinforcement learning problem with the long-run average criterion. Firstly, we develop a unified trust region theory with discounted and average criteria and derive a novel performance bound within the trust region with the Perturbation Analysis (PA) theory. Secondly, we propose a practical algorithm named Average Policy Optimization (APO), which improves the value estimation with a novel technique named Average Value Constraint. Finally, experiments are conducted in the continuous control environment MuJoCo. In most tasks, APO performs better than the discounted PPO, which demonstrates the effectiveness of our approach. Our work provides a unified framework of the trust region approach including both the discounted and average criteria, which may complement the framework of reinforcement learning beyond the discounted objectives.

cs.LG↗

arXiv2021

Exposing Deepfake with Pixel-wise AR and PPG Correlation from Faint Signals

Deepfake poses a serious threat to the reliability of judicial evidence and intellectual property protection. In spite of an urgent need for Deepfake identification, existing pixel-level detection methods are increasingly unable to resist the growing realism of fake videos and lack generalization. In this paper, we propose a scheme to expose Deepfake through faint signals hidden in face videos. This scheme extracts two types of minute information hidden between face pixels-photoplethysmography (PPG) features and auto-regressive (AR) features, which are used as the basis for forensics in the temporal and spatial domains, respectively. According to the principle of PPG, tracking the absorption of light by blood cells allows remote estimation of the temporal domains heart rate (HR) of face video, and irregular HR fluctuations can be seen as traces of tampering. On the other hand, AR coefficients are able to reflect the inter-pixel correlation, and can also reflect the traces of smoothing caused by up-sampling in the process of generating fake faces. Furthermore, the scheme combines asymmetric convolution block (ACBlock)-based improved densely connected networks (DenseNets) to achieve face video authenticity forensics. Its asymmetric convolutional structure enhances the robustness of network to the input feature image upside-down and left-right flipping, so that the sequence of feature stitching does not affect detection results. Simulation results show that our proposed scheme provides more accurate authenticity detection results on multiple deep forgery datasets and has better generalization compared to the benchmark strategy.

cs.CV↗

arXiv2021

Believe What You See: Implicit Constraint Approach for Offline Multi-Agent Reinforcement Learning

Learning from datasets without interaction with environments (Offline Learning) is an essential step to apply Reinforcement Learning (RL) algorithms in real-world scenarios. However, compared with the single-agent counterpart, offline multi-agent RL introduces more agents with the larger state and action space, which is more challenging but attracts little attention. We demonstrate current offline RL algorithms are ineffective in multi-agent systems due to the accumulated extrapolation error. In this paper, we propose a novel offline RL algorithm, named Implicit Constraint Q-learning (ICQ), which effectively alleviates the extrapolation error by only trusting the state-action pairs given in the dataset for value estimation. Moreover, we extend ICQ to multi-agent tasks by decomposing the joint-policy under the implicit constraint. Experimental results demonstrate that the extrapolation error is successfully controlled within a reasonable range and insensitive to the number of agents. We further show that ICQ achieves the state-of-the-art performance in the challenging multi-agent offline tasks (StarCraft II). Our code is public online at https://github.com/YiqinYang/ICQ.

cs.AI↗

arXiv2021

Offline Reinforcement Learning with Value-based Episodic Memory

Offline reinforcement learning (RL) shows promise of applying RL to real-world problems by effectively utilizing previously collected data. Most existing offline RL algorithms use regularization or constraints to suppress extrapolation error for actions outside the dataset. In this paper, we adopt a different framework, which learns the V-function instead of the Q-function to naturally keep the learning procedure within the support of an offline dataset. To enable effective generalization while maintaining proper conservatism in offline learning, we propose Expectile V-Learning (EVL), which smoothly interpolates between the optimal value learning and behavior cloning. Further, we introduce implicit planning along offline trajectories to enhance learned V-values and accelerate convergence. Together, we present a new offline method called Value-based Episodic Memory (VEM). We provide theoretical analysis for the convergence properties of our proposed VEM method, and empirical results in the D4RL benchmark show that our method achieves superior performance in most tasks, particularly in sparse-reward tasks.

cs.LG↗

arXiv2021

Proposal of Analog In-Memory Computing with Magnified Tunnel Magnetoresistance Ratio and Universal STT-MRAM Cell

In-memory computing (IMC) is an effectual solution for energy-efficient artificial intelligence applications. Analog IMC amortizes the power consumption of multiple sensing amplifiers with analog-to-digital converter (ADC), and simultaneously completes the calculation of multi-line data with high parallelism degree. Based on a universal one-transistor one-magnetic tunnel junction (MTJ) spin transfer torque magnetic RAM (STT-MRAM) cell, this paper demonstrates a novel tunneling magnetoresistance (TMR) ratio magnifying method to realize analog IMC. Previous concerns include low TMR ratio and analog calculation nonlinearity are addressed using device-circuit interaction. Peripheral circuits are minimally modified to enable in-memory matrix-vector multiplication. A current mirror with feedback structure is implemented to enhance analog computing linearity and calculation accuracy. The proposed design maximumly supports 1024 2-bit input and 1-bit weight multiply-and-accumulate (MAC) computations simultaneously. The 2-bit input is represented by the width of the input (IN) pulses, while the 1-bit weight is stored in STT-MRAM and the x7500 magnified TMR (m-TMR) ratio is obtained by latching. The proposal is simulated using 28-nm CMOS process and MTJ compact model. The integral nonlinearity is reduced by 57.6% compared with the conventional structure. 9.47-25.4 TOPS/W is realized with 2-bit input, 1-bit weight and 4-bit output convolution neural network (CNN).

cs.ET↗

arXiv2021

Compact Dual-Polarization Silicon Integrated Couplers for Multicore Fibers

Compact fiber-to-chip couplers play an important role in optical interconnections, especially in data centers. However, the development of couplers has been mostly limited to standard single mode fibers, with few devices compatible with multicore and multimode fibers. Through the use of state-of-the-art optimization algorithms, we designed a compact dual-polarization coupler to interface chips and dense multicore fibers, demonstrating, for the first time, coupling to both polarizations of all the cores, with measured coupling efficiency of $-$4.3 dB and with a 3 dB bandwidth of 48 nm. The dual-polarization coupler has footprint of 200 $μm^2$ per core, which makes it the smallest fiber-to-chip coupler experimentally demonstrated on a standard silicon-on-insulator platform.

physics.optics↗

arXiv2021

ROBI: A Multi-View Dataset for Reflective Objects in Robotic Bin-Picking

In robotic bin-picking applications, the perception of texture-less, highly reflective parts is a valuable but challenging task. The high glossiness can introduce fake edges in RGB images and inaccurate depth measurements especially in heavily cluttered bin scenario. In this paper, we present the ROBI (Reflective Objects in BIns) dataset, a public dataset for 6D object pose estimation and multi-view depth fusion in robotic bin-picking scenarios. The ROBI dataset includes a total of 63 bin-picking scenes captured with two active stereo camera: a high-cost Ensenso sensor and a low-cost RealSense sensor. For each scene, the monochrome/RGB images and depth maps are captured from sampled view spheres around the scene, and are annotated with accurate 6D poses of visible objects and an associated visibility score. For evaluating the performance of depth fusion, we captured the ground truth depth maps by high-cost Ensenso camera with objects coated in anti-reflective scanning spray. To show the utility of the dataset, we evaluated the representative algorithms of 6D object pose estimation and multi-view depth fusion on the full dataset. Evaluation results demonstrate the difficulty of highly reflective objects, especially in difficult cases due to the degradation of depth data quality, severe occlusions and cluttered scene. The ROBI dataset is available online at https://www.trailab.utias.utoronto.ca/robi.

cs.RO↗

arXiv2021

GRB 210121A: A Typical Fireball Burst Detected by Two Small Missions

The Chinese CubeSat Mission, Gamma Ray Integrated Detectors (GRID), recently detected its first gamma-ray burst, GRB 210121A, which was jointly observed by the Gravitational wave high-energy Electromagnetic Counterpart All-sky Monitor (GECAM). This burst is confirmed by several other missions, including \fermi and \textit{Insight}-HXMT. We combined multi-mission observational data and performed a comprehensive analysis of the burst's temporal and spectral properties. Our results show that the burst is relatively special in its high peak energy, thermal-like low energy indices, and large fluence. By putting it to the $E_{\rm p}$-$E_{\rmγ, iso}$ relation diagram with assumed distance, we found this burst can be constrained at the redshift range of [0.3,3.0]. The thermal spectral component is also confirmed by the direct fit of the physical models to the observed spectra. Interestingly, the physical photosphere model also constrained a redshift of $z\sim$ 0.3 for this burst, which help us to identify a host galaxy candidate at such a distance within the location error box. Assuming the host galaxy is real, we found the burst can be best explained by the photosphere emission of a typical fireball with an initial radius of $r_0\sim$ 3.2 $\times 10^7$ cm.

astro-ph.HE↗

arXiv2021

Nucleus-electron correlation revising molecular bonding fingerprints from the exact wavefunction factorization

We present a novel theory and implementation for computing coupled electronic and quantal nuclear subsystems on a single potential energy surface, moving beyond the standard Born-Oppenheimer (BO) separation of nuclei and electrons. We formulate an exact self-consistent nucleus-electron embedding potential from the single product molecular wavefunction, and demonstrate that the fundamental behavior of correlated nucleus-electron can be computed for mean-field electrons that are responsive to a quantal anharmonic vibration of selected nuclei in a discrete variable representation. Geometric gauge choices are discussed and necessary for formulating energy invariant biorthogonal electronic equations. Our method is further applied to characterize vibrationally averaged molecular bonding properties of molecular energetics, bond length, protonic and electron density. Moreover, post-Hartree-Fock electron correlation can be conveniently computed on the basis of nucleus-electron coupled molecular orbitals, as demonstrated to correlated models of second-order Møllet-Plesset perturbation and full configuration interaction theories. Our approach not only accurately quantifies non-classical nucleus-electron couplings for revising molecular bonding properties, but also provides an alternative time-independent approach for deploying non-BO molecular quantum chemistry.

physics.chem-ph↗

arXiv2021

Asymmetry and Variability in the Transmission Spectra of Tidally Locked Habitable Planets

Spatial heterogeneity and temporal variability are general features in planetary weather and climate, due to the effects of planetary rotation, uneven stellar flux distribution, fluid motion instability, etc. In this study, we investigate the asymmetry and variability in the transmission spectra of 1:1 spin--orbit tidally locked (or called synchronously rotating) planets around low-mass stars. We find that for rapidly rotating planets, the transit atmospheric thickness on the evening terminator (east of the substellar region) is significantly larger than that of the morning terminator (west of the substellar region). The asymmetry is mainly related to the spatial heterogeneity in ice clouds, as the contributions of liquid clouds and water vapor are smaller. The underlying mechanism is that there are always more ice clouds on the evening terminator, due to the combined effect of coupled Rossby--Kelvin waves and equatorial superrotation that advect vapor and clouds to the east, especially at high levels of the atmosphere. For slowly rotating planets, the asymmetry reverses (the morning terminator has a larger transmission depth than the evening terminator) but the magnitude is small or even negligible. For both rapidly and slowly rotating planets, there is strong variability in the transmission spectra. The asymmetry signal is nearly impossible to be observed by the James Webb Space Telescope (JWST), because the magnitude of the asymmetry (about 10 ppm) is smaller than the instrumental noise and the high variability further increases the challenge.

astro-ph.EP↗

arXiv2021

Boosting the Convergence of Reinforcement Learning-based Auto-pruning Using Historical Data

Recently, neural network compression schemes like channel pruning have been widely used to reduce the model size and computational complexity of deep neural network (DNN) for applications in power-constrained scenarios such as embedded systems. Reinforcement learning (RL)-based auto-pruning has been further proposed to automate the DNN pruning process to avoid expensive hand-crafted work. However, the RL-based pruner involves a time-consuming training process and the high expense of each sample further exacerbates this problem. These impediments have greatly restricted the real-world application of RL-based auto-pruning. Thus, in this paper, we propose an efficient auto-pruning framework which solves this problem by taking advantage of the historical data from the previous auto-pruning process. In our framework, we first boost the convergence of the RL-pruner by transfer learning. Then, an augmented transfer learning scheme is proposed to further speed up the training process by improving the transferability. Finally, an assistant learning process is proposed to improve the sample efficiency of the RL agent. The experiments have shown that our framework can accelerate the auto-pruning process by 1.5-2.5 times for ResNet20, and 1.81-2.375 times for other neural networks like ResNet56, ResNet18, and MobileNet v1.

cs.LG↗

arXiv2021

Micro/nanoliter droplet extraction by controlling acoustic vortex with miniwatt

Micro/nanoliter droplet is capable of achieving versatile applications with tiny volume and substantial surface energy, which is a big plus over bulk liquid. Yet, the contradiction of elaborate manipulation and enough power is still a challenge. Here, we unleash the potential of our miniwatt aspirators pumping up liquid and creating droplets with the help of acoustic vortex beams, inspired by the power mechanism that spirals are significant for most mollusks that live in water. These droplet aspirators produce very large interface deformations by small radiation pressures with orbit angular momentum from spiral-electrode transducers. The precisely contactless manipulation of physical, chemical and biological objects at micrometric down to nanometric scales, promises tremendous development in fields as diverse as microrobotics, nanoreactors, or nanoassemblies.

physics.app-ph↗

arXiv2021

Extremely Low Footprint End-to-End ASR System for Smart Device

Recently, end-to-end (E2E) speech recognition has become popular, since it can integrate the acoustic, pronunciation and language models into a single neural network, which outperforms conventional models. Among E2E approaches, attention-based models, e.g. Transformer, have emerged as being superior. Such models have opened the door to deployment of ASR on smart devices, however they still suffer from requiring a large number of model parameters. We propose an extremely low footprint E2E ASR system for smart devices, to achieve the goal of satisfying resource constraints without sacrificing recognition accuracy. We design cross-layer weight sharing to improve parameter efficiency and further exploit model compression methods including sparsification and quantization, to reduce memory storage and boost decoding efficiency. We evaluate our approaches on the public AISHELL-1 and AISHELL-2 benchmarks. On the AISHELL-2 task, the proposed method achieves more than 10x compression (model size reduces from 248 to 24MB), at the cost of only minor performance loss (CER reduces from 6.49% to 6.92%).

cs.SD↗