arXiv subjects

Jie Li

Publications and source records attributed to Jie Li.

At least 541 records · Page 30Linked to original sources

arXiv2021

Observation of the out-of-plane polarized spin current from CVD grown WTe2

Weyl semimetal Td-phase WTe2 possesses the spin-resolved band structure with strong spin-orbit coupling, holding promises as a useful spin source material. The noncentrosymmetric crystalline structure of Td-WTe2 endows the generation of the out-of-plane polarized spin, which is of great interest in magnetic memory applications. Previously, WTe2 was explored in spin devices based on mechanically exfoliated single crystal flakes with a size of micrometers. For practical spintronics applications, it is highly desirable to implement wafer-scale thin films. In this work, we utilize centimeter-scale chemical vapor deposition (CVD)-grown Td-WTe2 thin films and study the spin current generation by the spin torque ferromagnetic resonance technique. We find the in-plane and out-of-plane spin conductivities of 7.36 x 10^3 (h/2e) (ohm-m)^-1 and 1.76 x 10^3 (h/2e) (ohm-m)^-1, respectively, in CVD-growth 5 nm-WTe2. We further demonstrate the current-induced magnetization switching in WTe2/NiFe at room temperature in the domain wall motion regime, which may invigorate potential spintronic device innovations based on Weyl semimetals.

cond-mat.mtrl-sci↗

arXiv2021

Hierarchical Lovász Embeddings for Proposal-free Panoptic Segmentation

Panoptic segmentation brings together two separate tasks: instance and semantic segmentation. Although they are related, unifying them faces an apparent paradox: how to learn simultaneously instance-specific and category-specific (i.e. instance-agnostic) representations jointly. Hence, state-of-the-art panoptic segmentation methods use complex models with a distinct stream for each task. In contrast, we propose Hierarchical Lovász Embeddings, per pixel feature vectors that simultaneously encode instance- and category-level discriminative information. We use a hierarchical Lovász hinge loss to learn a low-dimensional embedding space structured into a unified semantic and instance hierarchy without requiring separate network branches or object proposals. Besides modeling instances precisely in a proposal-free manner, our Hierarchical Lovász Embeddings generalize to categories by using a simple Nearest-Class-Mean classifier, including for non-instance "stuff" classes where instance segmentation methods are not applicable. Our simple model achieves state-of-the-art results compared to existing proposal-free panoptic segmentation methods on Cityscapes, COCO, and Mapillary Vistas. Furthermore, our model demonstrates temporal stability between video frames.

cs.CV↗

arXiv2021

Semi-supervised Contrastive Learning with Similarity Co-calibration

Semi-supervised learning acts as an effective way to leverage massive unlabeled data. In this paper, we propose a novel training strategy, termed as Semi-supervised Contrastive Learning (SsCL), which combines the well-known contrastive loss in self-supervised learning with the cross entropy loss in semi-supervised learning, and jointly optimizes the two objectives in an end-to-end way. The highlight is that different from self-training based semi-supervised learning that conducts prediction and retraining over the same model weights, SsCL interchanges the predictions over the unlabeled data between the two branches, and thus formulates a co-calibration procedure, which we find is beneficial for better prediction and avoid being trapped in local minimum. Towards this goal, the contrastive loss branch models pairwise similarities among samples, using the nearest neighborhood generated from the cross entropy branch, and in turn calibrates the prediction distribution of the cross entropy branch with the contrastive similarity. We show that SsCL produces more discriminative representation and is beneficial to few shot learning. Notably, on ImageNet with ResNet50 as the backbone, SsCL achieves 60.2% and 72.1% top-1 accuracy with 1% and 10% labeled samples, respectively, which significantly outperforms the baseline, and is better than previous semi-supervised and self-supervised methods.

cs.CV↗

arXiv2021

Direct and indirect reinforcement learning

Reinforcement learning (RL) algorithms have been successfully applied to a range of challenging sequential decision making and control tasks. In this paper, we classify RL into direct and indirect RL according to how they seek the optimal policy of the Markov decision process problem. The former solves the optimal policy by directly maximizing an objective function using gradient descent methods, in which the objective function is usually the expectation of accumulative future rewards. The latter indirectly finds the optimal policy by solving the Bellman equation, which is the sufficient and necessary condition from Bellman's principle of optimality. We study policy gradient forms of direct and indirect RL and show that both of them can derive the actor-critic architecture and can be unified into a policy gradient with the approximate value function and the stationary state distribution, revealing the equivalence of direct and indirect RL. We employ a Gridworld task to verify the influence of different forms of policy gradient, suggesting their differences and relationships experimentally. Finally, we classify current mainstream RL algorithms using the direct and indirect taxonomy, together with other ones including value-based and policy-based, model-based and model-free.

cs.LG↗

arXiv2021

6 GHz hyperfast rotation of an optically levitated nanoparticle in vacuum

We report an experimental observation of a record-breaking ultra-high rotation frequency about 6 GHz in an optically levitated nanoparticle system. We optically trap a nanoparticle in the gravity direction with a high numerical aperture (NA) objective lens, which shows significant advantages in compensating the influences of the scattering force and the photophoretic force on the trap, especially at intermediate pressure (about 100 Pa). This allows us to trap a nanoparticle from atmospheric to low pressure ($10^{-3}$ Pa) without using feedback cooling. We measure a highest rotation frequency about 4.3 GHz of the trapped nanoparticle without feedback cooling and a 6 GHz rotation with feedback cooling, which is the fastest mechanical rotation ever reported to date. Our work provides useful guides for efficiently observing hyperfast rotation in the optical levitation system, and may find various applications such as in ultrasensitive torque detection, probing vacuum friction, and testing unconventional decoherence theories.

physics.optics↗

arXiv2021

Exploring Multi-dimensional Data via Subset Embedding

Multi-dimensional data exploration is a classic research topic in visualization. Most existing approaches are designed for identifying record patterns in dimensional space or subspace. In this paper, we propose a visual analytics approach to exploring subset patterns. The core of the approach is a subset embedding network (SEN) that represents a group of subsets as uniformly-formatted embeddings. We implement the SEN as multiple subnets with separate loss functions. The design enables to handle arbitrary subsets and capture the similarity of subsets on single features, thus achieving accurate pattern exploration, which in most cases is searching for subsets having similar values on few features. Moreover, each subnet is a fully-connected neural network with one hidden layer. The simple structure brings high training efficiency. We integrate the SEN into a visualization system that achieves a 3-step workflow. Specifically, analysts (1) partition the given dataset into subsets, (2) select portions in a projected latent space created using the SEN, and (3) determine the existence of patterns within selected subsets. Generally, the system combines visualizations, interactions, automatic methods, and quantitative measures to balance the exploration flexibility and operation efficiency, and improve the interpretability and faithfulness of the identified patterns. Case studies and quantitative experiments on multiple open datasets demonstrate the general applicability and effectiveness of our approach.

cs.LG↗

arXiv2021

Wireless Sensing With Deep Spectrogram Network and Primitive Based Autoregressive Hybrid Channel Model

Human motion recognition (HMR) based on wireless sensing is a low-cost technique for scene understanding. Current HMR systems adopt support vector machines (SVMs) and convolutional neural networks (CNNs) to classify radar signals. However, whether a deeper learning model could improve the system performance is currently not known. On the other hand, training a machine learning model requires a large dataset, but data gathering from experiment is cost-expensive and time-consuming. Although wireless channel models can be adopted for dataset generation, current channel models are mostly designed for communication rather than sensing. To address the above problems, this paper proposes a deep spectrogram network (DSN) by leveraging the residual mapping technique to enhance the HMR performance. Furthermore, a primitive based autoregressive hybrid (PBAH) channel model is developed, which facilitates efficient training and testing dataset generation for HMR in a virtual environment. Experimental results demonstrate that the proposed PBAH channel model matches the actual experimental data very well and the proposed DSN achieves significantly smaller recognition error than that of CNN.

cs.IT↗

arXiv2021

Drafting and Revision: Laplacian Pyramid Network for Fast High-Quality Artistic Style Transfer

Artistic style transfer aims at migrating the style from an example image to a content image. Currently, optimization-based methods have achieved great stylization quality, but expensive time cost restricts their practical applications. Meanwhile, feed-forward methods still fail to synthesize complex style, especially when holistic global and local patterns exist. Inspired by the common painting process of drawing a draft and revising the details, we introduce a novel feed-forward method named Laplacian Pyramid Network (LapStyle). LapStyle first transfers global style patterns in low-resolution via a Drafting Network. It then revises the local details in high-resolution via a Revision Network, which hallucinates a residual image according to the draft and the image textures extracted by Laplacian filtering. Higher resolution details can be easily generated by stacking Revision Networks with multiple Laplacian pyramid levels. The final stylized image is obtained by aggregating outputs of all pyramid levels. %We also introduce a patch discriminator to better learn local patterns adversarially. Experiments demonstrate that our method can synthesize high quality stylized images in real time, where holistic style patterns are properly transferred.

cs.CV↗

arXiv2021

Decoding electron tunnelling delay time by embracing wave-particle duality

Tunnelling lies at the heart of quantum mechanics and is a fundamental process in attosecond science, molecular biology, and quantum devices. Whether tunnelling takes time and how a microscopic particle transits through a barrier have been debated since the early days of quantum mechanics. The time required for an electron to tunnel through an atomic potential barrier has been measured with attosecond angular streaking (attoclock), and a recent work on the hydrogen atom claimed that electron tunnelling is instantaneous. However, the time required for Rb atoms to tunnel through an optical potential barrier has been measured to be on the order of milliseconds with a recent Larmor clock measurement. The essence of electron and atom tunnelling is identical, but the reason for the contradictory conclusions remains unknown. Here, we demonstrate that the sub-barrier potential interaction is the root of the nonzero tunnelling delay time. We reveal that the wave-particle duality of a tunnelling electron must be fully taken into account when decoding the tunnelling delay time from the attoclock measurements. Based on energy-resolved attoclock measurements, we show that the tunnelling delay time of an electron ranges from 24 to 58 attoseconds, and the counterintuitive result that an electron with a lower energy may spend less time in the barrier is consistent with the velocity-dependent tunnelling time of atoms in Larmor clock measurements. Our results unify the tunnelling time of microscopic particles by highlighting the classically forbidden sub-barrier potential interactions of matter waves.

physics.atom-ph↗

arXiv2021

Towards SocialVR: Evaluating a Novel Technology for Watching Videos Together

Social VR enables people to interact over distance with others in real-time. It allows remote people, typically represented as avatars, to communicate and perform activities together in a join shared virtual environment, extending the capabilities of traditional social platforms like Facebook and Netflix. This paper explores the benefits and drawbacks provided by a lightweight and low-cost Social VR platform (SocialVR), in which users are captured by several cameras and reconstructed in real-time. In particular, the paper contributes with (1) the design and evaluation of an experimental protocol for Social VR experiences; (2) the report of a production workflow for this new type of media experiences; and (3) the results of experiments with both end-users (N=15 pairs) and professionals (N=25) to evaluate the potential of the SocialVR platform. Results from the questionnaires and semi-structured interviews show that end-users rated positively towards the experiences provided by the SocialVR platform, which enabled them to sense emotions and communicate effortlessly. End-users perceived the photo-realistic experience of SocialVR similar to face-to-face scenarios and appreciated this new creative medium. From a commercial perspective, professionals confirmed the potential of this communication medium and encourage further research for the adoption of the platform in the commercial landscape

cs.MM↗

arXiv2021

Earth's carbon deficit caused by early loss through irreversible sublimation

Carbon is an essential element for life but its behavior during Earth's accretion is not well understood. Carbonaceous grains in meteoritic and cometary materials suggest that irreversible sublimation, and not condensation, governs carbon acquisition by terrestrial worlds. Through astronomical observations and modeling we show that the sublimation front of carbon carriers in the solar nebula, or the soot line, moved inward quickly so that carbon-rich ingredients would be available for accretion at 1 au after the first million years. On the other hand, geological constraints firmly establish a severe carbon deficit in Earth, requiring the destruction of inherited carbonaceous organics in the majority of its building blocks. The carbon-poor nature of the Earth thus implies carbon loss in its precursor material through sublimation within the first million years.

astro-ph.EP↗

arXiv2021

Early volatile depletion on planetesimals inferred from C-S systematics of iron meteorite parent bodies

During the formation of terrestrial planets, volatile loss may occur through nebular processing, planetesimal differentiation, and planetary accretion. We investigate iron meteorites as an archive of volatile loss during planetesimal processing. The carbon contents of the parent bodies of magmatic iron meteorites are reconstructed by thermodynamic modelling. Calculated solid/molten alloy partitioning of C increases greatly with liquid S concentration and inferred parent body C concentrations range from 0.0004 to 0.11 wt.%. Parent bodies fall into 2 compositional clusters characterized by cores with medium, and low C/S. Both of these require significant planetesimal degassing, as metamorphic devolatilization on chondrite-like precursors is insufficient to account for their C depletions. Planetesimal core formation models, ranging from closed system extraction to degassing of a wholly molten body, show that significant open system silicate melting and volatile loss is required to match medium and low C/S parent body core compositions. Greater depletion in C relative to S is the hallmark of silicate degassing, indicating that parent body core compositions record processes that affect composite silicate/iron planetesimals. Degassing of bare cores stripped of their silicate mantles would deplete S with negligible C loss, and could not account for inferred parent body core compositions. Devolatilization during small-body differentiation is thus a key process in shaping the volatile inventory of terrestrial planets derived from planetesimals and planetary embryos.

astro-ph.EP↗

arXiv2021

Performance Analysis of Multipath BGP

Multipath BGP (M-BGP) allows a BGP router to install multiple 'equally-good' paths, via parallel inter-domain border links, to a destination prefix. M-BGP differs from the multipath routing techniques in many ways, e.g. M-BGP is only implemented at border routers of Autonomous Systems (ASes); and while it shares traffic to different IP addresses in a destination prefix via different border links, any traffic to a given destination IP always follows the same border link. Recently we studied Looking Glass data and reported the wide deployment of M-BGP in the Internet; in particular, Hurricane Electric (AS6939) has implemented over 1,000 cases of M-BGP to hundreds of its peering ASes. In this paper, we analyzed the performance of M-BGP. We used RIPE Atlas to send traceroute probes to a series of destination prefixes through Hurricane Electric's border routers implemented with M-BGP. We examined the distribution of Round Trip Time to each probed IP address in a destination prefix and their variation during the measurement. We observed that the deployment of M-BGP can guarantee stable routing between ASes and enhance a network's resilience to traffic changes. Our work provides insights into the unique characteristics of M-BGP as an effective technique for load balancing.

cs.NI↗

arXiv2021

Association: Remind Your GAN not to Forget

Neural networks are susceptible to catastrophic forgetting. They fail to preserve previously acquired knowledge when adapting to new tasks. Inspired by human associative memory system, we propose a brain-like approach that imitates the associative learning process to achieve continual learning. We design a heuristics mechanism to potentiatively stimulate the model, which guides the model to recall the historical episodes based on the current circumstance and obtained association experience. Besides, a distillation measure is added to depressively alter the efficacy of synaptic transmission, which dampens the feature reconstruction learning for new task. The framework is mediated by potentiation and depression stimulation that play opposing roles in directing synaptic and behavioral plasticity. It requires no access to the original data and is more similar to human cognitive process. Experiments demonstrate the effectiveness of our method in alleviating catastrophic forgetting on image-to-image translation tasks.

cs.CV↗

arXiv2021

Two-Dimensional Multifunctional Materials from Endohedral Fullerenes

A new multifunctional 2D material is theoretically predicted based on systematic ab-initio calculations and model simulations for the honeycomb lattice of endohedral fullerene W@C28 molecules. It has structural bistability, ferroelectricity, multiple magnetic phases, and excellent valley characters and can be easily functionalized by the proximity effect with magnetic isolators such as MnTiO3. Furthermore, we may also manipulate the valley Hall and spin transport properties by selectively switch a few W@C28 molecules to the metastable phase. These findings pave a new way in integrate different functions in a single 2D material for technological innovations.

cond-mat.mtrl-sci↗

arXiv2021

Sympathetic cooling of a radio-frequency LC circuit to its ground state in an optoelectromechanical system

We present a complete theory for laser cooling of a macroscopic radio-frequency LC electrical circuit by means of an optoelectromechanical system, consisting of an optical cavity dispersively coupled to a nanomechanical oscillator, which is in turn capacitively coupled to the LC circuit of interest. We determine the optimal parameter regime where the LC resonator can be cooled down to its quantum ground state, which requires a large optomechanical cooperativity, and a larger electromechanical cooperativity. Moreover, comparable optomechanical and electromechanical coupling rates are preferable for reaching the quantum ground state.

quant-ph↗

arXiv2021

Approximate Optimal Filter for Linear Gaussian Time-invariant Systems

State estimation is critical to control systems, especially when the states cannot be directly measured. This paper presents an approximate optimal filter, which enables to use policy iteration technique to obtain the steady-state gain in linear Gaussian time-invariant systems. This design transforms the optimal filtering problem with minimum mean square error into an optimal control problem, called Approximate Optimal Filtering (AOF) problem. The equivalence holds given certain conditions about initial state distributions and policy formats, in which the system state is the estimation error, control input is the filter gain, and control objective function is the accumulated estimation error. We present a policy iteration algorithm to solve the AOF problem in steady-state. A classic vehicle state estimation problem finally evaluates the approximate filter. The results show that the policy converges to the steady-state Kalman gain, and its accuracy is within 2 %.

eess.SY↗

arXiv2021

Entangling the vibrational modes of two massive ferromagnetic spheres using cavity magnomechanics

We present a scheme to entangle the vibrational phonon modes of two massive ferromagnetic spheres in a dual-cavity magnomechanical system. In each cavity, a microwave cavity mode couples to a magnon mode (spin wave) via the magnetic dipole interaction, and the latter further couples to a deformation phonon mode of the ferromagnetic sphere via a nonlinear magnetostrictive interaction. We show that by directly driving the magnon mode with a red-detuned microwave field to activate the magnomechanical anti-Stokes process a cavity-magnon-phonon state-swap interaction can be realized. Therefore, if the two cavities are further driven by a two-mode squeezed vacuum field, the quantum correlation of the driving fields is successively transferred to the two magnon modes and subsequently to the two phonon modes, i.e., the two ferromagnetic spheres become remotely entangled. Our work demonstrates that cavity magnomechanical systems allow to prepare quantum entangled states at a more massive scale than currently possible with other schemes.

quant-ph↗