When reading a literary piece, readers often make inferences about various characters roles, personalities, relationships, intents, actions, etc. While humans can readily draw upon their past experiences to build such a character-centric view of the
narrative, understanding characters in narratives can be a challenging task for machines. To encourage research in this field of character-centric narrative understanding, we present LiSCU -- a new dataset of literary pieces and their summaries paired with descriptions of characters that appear in them. We also introduce two new tasks on LiSCU: Character Identification and Character Description Generation. Our experiments with several pre-trained language models adapted for these tasks demonstrate that there is a need for better models of narrative comprehension.
When applying a stochastic/incremental algorithm, one must choose the order to draw samples. Among the most popular approaches are cyclic sampling and random reshuffling, which are empirically faster and more cache-friendly than uniform-iid-sampling.
Cyclic sampling draws the samples in a fixed, cyclic order, which is less robust than reshuffling the samples periodically. Indeed, existing works have established worst case convergence rates for cyclic sampling, which are generally worse than that of random reshuffling. In this paper, however, we found a certain cyclic order can be much faster than reshuffling and one can discover it at a low cost! Studying and comparing different sampling orders typically require new analytic techniques. In this paper, we introduce a norm, which is defined based on the sampling order, to measure the distance to solution. Applying this technique on proximal Finito/MISO algorithm allows us to identify the optimal fixed ordering, which can beat random reshuffling by a factor up to log(n)/n in terms of the best-known upper bounds. We also propose a strategy to discover the optimal fixed ordering numerically. The established rates are state-of-the-art compared to previous works.
The scale of deep learning nowadays calls for efficient distributed training algorithms. Decentralized momentum SGD (DmSGD), in which each node averages only with its neighbors, is more communication efficient than vanilla Parallel momentum SGD that
incurs global average across all computing nodes. On the other hand, the large-batch training has been demonstrated critical to achieve runtime speedup. This motivates us to investigate how DmSGD performs in the large-batch scenario. In this work, we find the momentum term can amplify the inconsistency bias in DmSGD. Such bias becomes more evident as batch-size grows large and hence results in severe performance degradation. We next propose DecentLaM, a novel decentralized large-batch momentum SGD to remove the momentum-incurred bias. The convergence rate for both non-convex and strongly-convex scenarios is established. Our theoretical results justify the superiority of DecentLaM to DmSGD especially in the large-batch scenario. Experimental results on a variety of computer vision tasks and models demonstrate that DecentLaM promises both efficient and high-quality training.
Optimized for pixel fidelity metrics, images compressed by existing image codec are facing systematic challenges when used for visual analysis tasks, especially under low-bitrate coding. This paper proposes a visual analysis-motivated rate-distortion
model for Versatile Video Coding (VVC) intra compression. The proposed model has two major contributions, a novel rate allocation strategy and a new distortion measurement model. We first propose the region of interest for machine (ROIM) to evaluate the degree of importance for each coding tree unit (CTU) in visual analysis. Then, a novel CTU-level bit allocation model is proposed based on ROIM and the local texture characteristics of each CTU. After an in-depth analysis of multiple distortion models, a visual analysis friendly distortion criteria is subsequently proposed by extracting deep feature of each coding unit (CU). To alleviate the problem of lacking spatial context information when calculating the distortion of each CU, we finally propose a multi-scale feature distortion (MSFD) metric using different neighboring pixels by weighting the extracted deep features in each scale. Extensive experimental results show that the proposed scheme could achieve up to 28.17% bitrate saving under the same analysis performance among several typical visual analysis tasks such as image classification, object detection, and semantic segmentation.
A characteristic of a Fermi liquid is the T^2 dependence of its resistivity, sometimes referred to as the Baber law. However, for most metals, this behavior is only restricted to very low temperatures, usually below 20 K. Here, we experimentally demo
nstrate that for the single-crystal van der Waals layered material MoOCl2, the Baber law holds in a wide temperature range up to ~120 K, indicating that the electron-electron scattering plays a dominant role in this material. Combining with the specific heat measurement, we find that the modified Kadowaki-Woods ratio of the material agrees well with many other strongly correlated metals. Furthermore, in the magneto-transport measurement, a colossal magneto-resistance is observed, which reaches ~350% at 9 T and displays no sign of saturation. With the help of first-principles calculations, we attribute this behavior to the presence of open orbits on the Fermi surface. We also suggest that the dominance of electron-electron scattering is related to an incipient charge density wave state of the material. Our results establish MoOCl2 as a strongly correlated metal and shed light on the underlying physical mechanism, which may open a new path for exploring the effects of electron-electron interaction in van der Waals layered structures.
The Scaled Relative Graph (SRG) by Ryu, Hannah, and Yin (arXiv:1902.09788, 2019) is a geometric tool that maps the action of a multi-valued nonlinear operator onto the 2D plane, used to analyze the convergence of a wide range of iterative methods. As
the SRG includes the spectrum for linear operators, we can view the SRG as a generalization of the spectrum to multi-valued nonlinear operators. In this work, we further study the SRG of linear operators and characterize the SRG of block-diagonal and normal matrices.
Many iterative methods in optimization are fixed-point iterations with averaged operators. As such methods converge at an $mathcal{O}(1/k)$ rate with the constant determined by the averagedness coefficient, establishing small averagedness coefficient
s for operators is of broad interest. In this paper, we show that the averagedness coefficients of the composition of averaged operators by Ogura and Yamada (Numer Func Anal Opt 32(1--2):113--137, 2002) and the three-operator splitting by Davis and Yin (Set-Valued Var Anal 25(4):829--858, 2017) are tight. The analysis relies on the scaled relative graph, a geometric tool recently proposed by Ryu, Hannah, and Yin (arXiv:1902.09788, 2019).
A high-dimensional potential energy surface (PES) for CO interaction with the Au(111) surface is developed using a machine-learning algorithm. Including both molecular and surface coordinates, this PES enables the simulation of the recent experiment
on scattering of vibrationally excited CO from Au(111). Trapping in a physisorption well is observed to increase with decreasing incidence energy. While energy dissipation of physisorbed CO is slow, due to weak coupling with both the phonons and electron-hole pairs, its access to the chemisorption well facilitates fast vibrational relaxation of CO through nonadiabatic coupling with surface electron-hole pairs.
We study theoretically the spin and orbital angular momentum (OAM) Hall effect in a high mobility two-dimensional electron system with Rashba and Dresselhuas spin-orbit coupling by introducing both the spin and OAM torque corrections, respectively, t
o the spin and OAM currents. We find that when both bands are occupied, the spin Hall conductivity is still a constant (i.e., independent of the carrier density) which, however, has an opposite sign to the previous value. The spin Hall conductivity in general would not be cancelled by the OAM Hall conductivity. The OAM Hall conductivity is also independent of the carrier density but depends on the strength ratio of the Rashba to Dresselhaus spin-orbit coupling, suggesting that one can manipulate the total Hall current through tuning the Rashba coupling by a gate voltage. We note that in a pure Rashba system, though the spin Hall conductivity is exactly cancelled by the OAM Hall conductivity due to the angular momentum conservation, the spin Hall effect could still manifest itself as nonzero magnetization Hall current and finite magnetization at the sample edges because the magnetic dipole moment associated with the spin of an electron is twice as large as that of the OAM. We also evaluate the electric field-induced OAM and discuss the origin of the OAM Hall current. Finally, we find that the spin and OAM Hall conductivities are closely related to the Berry vector (or gauge) potential.
