We introduce a subalgebra $overline F$ of the Clifford vertex superalgebra ($bc$ system) which is completely reducible as a $L^{Vir} (-2,0)$-module, $C_2$-cofinite, but it is not conformal and it is not isomorphic to the symplectic fermion algebra $m
athcal{SF}(1)$. We show that $mathcal{SF}(1)$ and $overline{F}$ are in an interesting duality, since $overline{F}$ can be equipped with the structure of a $mathcal{SF}(1)$-module and vice versa. Using the decomposition of $overline F$ and a free-field realization from arXiv:1711.11342, we decompose $L_k(mathfrak{osp}(1vert 2))$ at the critical level $k=-3/2$ as a module for $L_k(mathfrak{sl}(2))$. The decomposition of $L_k(mathfrak{osp}(1vert 2))$ is exactly the same as of the $N=4$ superconformal vertex algebra with central charge $c=-9$, denoted by $mathcal V^{(2)}$. Using the duality between $overline{F}$ and $mathcal{SF}(1)$, we prove that $L_k(mathfrak{osp}(1vert 2))$ and $mathcal V^{(2)}$ are in the duality of the same type. As an application, we construct and classify all irreducible $L_k(mathfrak{osp}(1vert 2))$-modules in the category $mathcal O$ and the category $mathcal R$ which includes relaxed highest weight modules. We also describe the structure of the parafermion algebra $N_{-3/2}(mathfrak{osp}(1vert 2))$ as a $N_{-3/2}(mathfrak{sl}(2))$-module. We extend this example, and for each $p ge 2$, we introduce a non-conformal vertex algebra $mathcal A^{(p)}_{new}$ and show that $mathcal A^{(p)}_{new} $ is isomorphic to the doublet vertex algebra as a module for the Virasoro algebra. We also construct the vertex algebra $ mathcal V^{(p)} _{new}$ which is isomorphic to the logarithmic vertex algebra $mathcal V^{(p)}$ as a module for $widehat{mathfrak{sl}}(2)$.
In this work, we study the problem of named entity recognition (NER) in a low resource scenario, focusing on few-shot and zero-shot settings. Built upon large-scale pre-trained language models, we propose a novel NER framework, namely SpanNER, which
learns from natural language supervision and enables the identification of never-seen entity classes without using in-domain labeled data. We perform extensive experiments on 5 benchmark datasets and evaluate the proposed method in the few-shot learning, domain transfer and zero-shot learning settings. The experimental results show that the proposed method can bring 10%, 23% and 26% improvements in average over the best baselines in few-shot learning, domain transfer and zero-shot learning settings respectively.
Federated Semi-Supervised Learning (FedSSL) has gained rising attention from both academic and industrial researchers, due to its unique characteristics of co-training machine learning models with isolated yet unlabeled data. Most existing FedSSL met
hods focus on the classical scenario, i.e, the labeled and unlabeled data are stored at the client side. However, in real world applications, client users may not provide labels without any incentive. Thus, the scenario of labels at the server side is more practical. Since unlabeled data and labeled data are decoupled, most existing FedSSL approaches may fail to deal with such a scenario. To overcome this problem, in this paper, we propose FedCon, which introduces a new learning paradigm, i.e., contractive learning, to FedSSL. Experimental results on three datasets show that FedCon achieves the best performance with the contractive framework compared with state-of-the-art baselines under both IID and Non-IID settings. Besides, ablation studies demonstrate the characteristics of the proposed FedCon framework.
This report describes the submission of the DKU-DukeECE-Lenovo team to the VoxCeleb Speaker Recognition Challenge (VoxSRC) 2021 track 4. Our system including a voice activity detection (VAD) model, a speaker embedding model, two clustering-based spea
ker diarization systems with different similarity measurements, two different overlapped speech detection (OSD) models, and a target-speaker voice activity detection (TS-VAD) model. Our final submission, consisting of 5 independent systems, achieves a DER of 5.07% on the challenge test set.
A computational fluid dynamics (CFD) simulation framework for predicting complex flows is developed on the Tensor Processing Unit (TPU) platform. The TPU architecture is featured with accelerated performance of dense matrix multiplication, large high
bandwidth memory, and a fast inter-chip interconnect, which makes it attractive for high-performance scientific computing. The CFD framework solves the variable-density Navier-Stokes equation using a Low-Mach approximation, and the governing equations are discretized by a finite difference method on a collocated structured mesh. It uses the graph-based TensorFlow as the programming paradigm. The accuracy and performance of this framework is studied both numerically and analytically, specifically focusing on effects of TPU-native single precision floating point arithmetic on solution accuracy. The algorithm and implementation are validated with canonical 2D and 3D Taylor Green vortex simulations. To demonstrate the capability for simulating turbulent flows, simulations are conducted for two configurations, namely the decaying homogeneous isotropic turbulence and a turbulent planar jet. Both simulations show good statistical agreement with reference solutions. The performance analysis shows a linear weak scaling and a super-linear strong scaling up to a full TPU v3 pod with 2048 cores.
Borophene, a monoatomic layer of boron atoms, stands out among two-dimensional (2D) materials, with its versatile properties of polymorphism, metallicity, plasmonics, superconductivity, tantalizing for physics exploration and next-generation devices.
Yet its phases are all synthesized on and stay bound to metal substrates, hampering both characterization and use. The growth on the inert insulator would allow post-synthesis exfoliation of borophene, but its weak adhesion to such substrate results in a very high 2D-nucleation barrier preventing clean borophene growth. This challenge can be circumvented in a devised and demonstrated here, with ab initio calculations, strategy. Naturally present 1D-defects, the step-edges on h-BN substrate surface, enable boron epitaxial assembly, reduce the nucleation dimensionality and lower the barrier by an order of magnitude (to 1.1 eV or less), yielding v1/9 phase. Weak borophene adhesion to the insulator makes it readily accessible for comprehensive property tests or transfer into the device setting.
In cell-free massive MIMO networks, an efficient distributed detection algorithm is of significant importance. In this paper, we propose a distributed expectation propagation (EP) detector for cell-free massive MIMO. The detector is composed of two m
odules, a nonlinear module at the central processing unit (CPU) and a linear module at the access point (AP). The turbo principle in iterative decoding is utilized to compute and pass the extrinsic information between modules. An analytical framework is then provided to characterize the asymptotic performance of the proposed EP detector with a large number of antennas. Simulation results will show that the proposed method outperforms the distributed detectors in terms of bit-error-rate.
In this paper, the structure of the parafermion vertex operator algebra associated to an integrable highest weight module for simple affine Lie superalgebra $osp(1|2n)$ is studied. Particularly, we determine the generators for this algebra.
Moire superlattices in van der Waals heterostructures have emerged as a powerful tool for engineering novel quantum phenomena. Here we report the observation of a correlated interlayer exciton insulator in a double-layer heterostructure composed of a
WSe2 monolayer and a WS2/WSe2 moire bilayer that are separated by an ultrathin hexagonal boron nitride (hBN). The moire WS2/WSe2 bilayer features a Mott insulator state at hole density p/p0 = 1, where p0 corresponds to one hole per moire lattice site. When electrons are added to the Mott insulator in the WS2/WSe2 moire bilayer and an equal number of holes are injected into the WSe2 monolayer, a new interlayer exciton insulator emerges with the holes in the WSe2 monolayer and the electrons in the doped Mott insulator bound together through interlayer Coulomb interactions. The excitonic insulator is stable up to a critical hole density of ~ 0.5p0 in the WSe2 monolayer, beyond which the system becomes metallic. Our study highlights the opportunities for realizing novel quantum phases in double-layer moire systems due to the interplay between the moire flat band and strong interlayer electron interactions.
By considering the spatial correspondence, dense self-supervised representation learning has achieved superior performance on various dense prediction tasks. However, the pixel-level correspondence tends to be noisy because of many similar misleading
pixels, e.g., backgrounds. To address this issue, in this paper, we propose to explore textbf{set} textbf{sim}ilarity (SetSim) for dense self-supervised representation learning. We generalize pixel-wise similarity learning to set-wise one to improve the robustness because sets contain more semantic and structure information. Specifically, by resorting to attentional features of views, we establish corresponding sets, thus filtering out noisy backgrounds that may cause incorrect correspondences. Meanwhile, these attentional features can keep the coherence of the same image across different views to alleviate semantic inconsistency. We further search the cross-view nearest neighbours of sets and employ the structured neighbourhood information to enhance the robustness. Empirical evaluations demonstrate that SetSim is superior to state-of-the-art methods on object detection, keypoint detection, instance segmentation, and semantic segmentation.
