Part of speech (POS) tagging is a familiar NLP task. State of the art taggers routinely achieve token-level accuracies of over 97% on news body text, evidence that the problem is well understood. However, the register of English news headlines, headl
inese, is very different from the register of long-form text, causing POS tagging models to underperform on headlines. In this work, we automatically annotate news headlines with POS tags by projecting predicted tags from corresponding sentences in news bodies. We train a multi-domain POS tagger on both long-form and headline text and show that joint training on both registers improves over training on just one or naively concatenating training sets. We evaluate on a newly-annotated corpus of over 5,248 English news headlines from the Google sentence compression corpus, and show that our model yields a 23% relative error reduction per token and 19% per headline. In addition, we demonstrate that better headline POS tags can improve the performance of a syntax-based open information extraction system. We make POSH, the POS-tagged Headline corpus, available to encourage research in improved NLP models for news headlines.
Growth-induced pattern formations in curved film-substrate structures have attracted extensive attentions recently. In most existing literature, the growth tensor is assumed to be homogeneous or piecewise homogeneous. In this paper, we aim at clarify
ing the influence of a growth gradient on pattern formation and pattern evolution in bilayered tubular tissues under plane-strain deformation. In the framework of finite elasticity, a bifurcation condition is derived for a general material model and a generic growth function. Then we suppose that both layers are composed of neo-Hookean materials. In particular, the growth function is assumed to decay linearly from the inner surface or from the outer surface. It is found that a gradient in the growth has a weak effect on the critical state, compared to the homogeneous growth type where both layers share the same growth factor. Furthermore, a finite element model is built to validate the theoretical model and to investigate the post-buckling behaviors. It is found that the associated pattern transition is not controlled by the growth gradient but by the ratio of the shear modulus between two layers. Different morphologies can occur when the modulus ratio is varied. The current analysis could provide useful insight into the influence of a growth gradient on surface instabilities and suggests that a homogeneous growth field may provide a good approximation on interpreting complicated morphological formations in multiple systems.
Task-adaptive pre-training (TAPT) and Self-training (ST) have emerged as the major semi-supervised approaches to improve natural language understanding (NLU) tasks with massive amount of unlabeled data. However, its unclear whether they learn similar
representations or they can be effectively combined. In this paper, we show that TAPT and ST can be complementary with simple TFS protocol by following TAPT -> Finetuning -> Self-training (TFS) process. Experimental results show that TFS protocol can effectively utilize unlabeled data to achieve strong combined gains consistently across six datasets covering sentiment classification, paraphrase identification, natural language inference, named entity recognition and dialogue slot classification. We investigate various semi-supervised settings and consistently show that gains from TAPT and ST can be strongly additive by following TFS procedure. We hope that TFS could serve as an important semi-supervised baseline for future NLP studies.
In order to minimize the impact of LC (lane-changing) maneuver, this research proposes a novel LC algorithm in mixed traffic. The LC maneuver is parsed into two stages: one is from the decision point to the execution point (finding a suitable gap), a
nd the other is from the execution point to the end point (performing the LC maneuver). Thereafter, a multiobjective optimization problem integrating these two stages is constructed, in which the comfort, efficiency and safety of the LC vehicle and the surrounding vehicles are simultaneously considered. Through introducing the NSGA-II (Non-dominated Sorting Genetic Algorithm), the pareto-optimal frontier and pareto-optimal solution of this problem is obtained. The nearest point of the frontier to the origin is used as the final solution. Through the micro-level analysis of the operating status of each vehicle, macro-level analysis of the traffic flow state within the LC area, and the sensitivity analysis of pareto-optimal frontier, we verify the performance of our proposed algorithm. Results demonstrate that compared with the existing algorithm, our algorithm could provide the optimal execution point and trajectory with the least impact on surroundings. The operation status of the traffic flow within the LC area has been significantly improved. We anticipate that this research could provide valuable insights into autonomous driving technology.
The measurement of transient optical fields has proven critical to understanding the dynamical mechanisms underlying ultrafast physical and chemical phenomena, and is key to realizing higher speeds in electronics and telecommunications. Complete char
acterization of optical waveforms, however, requires an optical oscilloscope capable of resolving the electric field oscillations with sub-femtosecond resolution and with single-shot operation. Here, we show that strong-field nonlinear excitation of photocurrents in a silicon-based image sensor chip can provide the sub-cycle optical gate necessary to characterize carrier-envelope phase-stable optical waveforms in the mid-infrared. By mapping the temporal delay between an intense excitation and weak perturbing pulse onto a transverse spatial coordinate of the image sensor, we show that the technique allows single-shot measurement of few-cycle waveforms.
This paper addresses the deep face recognition problem under an open-set protocol, where ideal face features are expected to have smaller maximal intra-class distance than minimal inter-class distance under a suitably chosen metric space. To this end
, hyperspherical face recognition, as a promising line of research, has attracted increasing attention and gradually become a major focus in face recognition research. As one of the earliest works in hyperspherical face recognition, SphereFace explicitly proposed to learn face embeddings with large inter-class angular margin. However, SphereFace still suffers from severe training instability which limits its application in practice. In order to address this problem, we introduce a unified framework to understand large angular margin in hyperspherical face recognition. Under this framework, we extend the study of SphereFace and propose an improved variant with substantially better training stability -- SphereFace-R. Specifically, we propose two novel ways to implement the multiplicative margin, and study SphereFace-R under three different feature normalization schemes (no feature normalization, hard feature normalization and soft feature normalization). We also propose an implementation strategy -- characteristic gradient detachment -- to stabilize training. Extensive experiments on SphereFace-R show that it is consistently better than or competitive with state-of-the-art methods.
Internet search affects peoples cognition of the world, so mitigating biases in search results and learning fair models is imperative for social good. We study a unique gender bias in image search in this work: the search images are often gender-imba
lanced for gender-neutral natural language queries. We diagnose two typical image search models, the specialized model trained on in-domain datasets and the generalized representation model pre-trained on massive image and text data across the internet. Both models suffer from severe gender bias. Therefore, we introduce two novel debiasing approaches: an in-processing fair sampling method to address the gender imbalance issue for training models, and a post-processing feature clipping method base on mutual information to debias multimodal representations of pre-trained models. Extensive experiments on MS-COCO and Flickr30K benchmarks show that our methods significantly reduce the gender bias in image search models.
Crystal-level strain rate sensitivity and temperature sensitivity are investigated in Zircaloy-4 using combined of bending creep test, digital image correlation, electron backscatter detection and thermo-mechanical tensile tests with crystal plastici
ty modelling. Crystal rate-sensitive properties are extracted from room temperature microscale creep, and temperature sensitivity from thermal polycrystalline responses. Crystal plasticity results show that large microscale creep strain is observed near notch tip increased up to 50% due to cross-slip activation. Grain-level microscale SRS is highly heterogeneous, and its crystallographic sensitivity is dependent on plastic deformation rate and underlying grain-based dislocation slip activation. Pyramidal <c+a> slip and total dislocation pileups contribute to temperature-sensitive texture effect on yielding and strength hardening. A faithful reconstruction of polycrystal and accurate rate-sensitive single-crystal properties are the key to capture multi-scale SRSs.
Using ab initio tight-binding approaches, we investigate Floquet band engineering of the 1T phase of transition metal dichalcogenides (MX2, M = W, Mo and X = Te, Se, S) monolayers under the irradiation with circularly polarized light. Our first princ
iples calculations demonstrate that light can induce important transitions in the topological phases of this emerging materials family. For example, upon irradiation, Te-based MX2 undergoes a phase transition from quantum spin Hall (QSH) semimetal to time-reversal symmetry broken QSH insulator with a nontrivial band gap of up to 92.5 meV. On the other hand, Se- and S-based MX2 undergoes the topological phase transition from the QSH effect to the quantum anomalous Hall (QAH) effect and into trivial phases with increasing light intensity. From a general perspective, our work brings further insight into non-equilibrium topological systems.
Advanced nuclear reactors often exhibit complex thermal-fluid phenomena during transients. To accurately capture such phenomena, a coarse-mesh three-dimensional (3-D) modeling capability is desired for modern nuclear-system code. In the coarse-mesh 3
-D modeling of advanced-reactor transients that involve flow and heat transfer, accurately predicting the turbulent viscosity is a challenging task that requires an accurate and computationally efficient model to capture the unresolved fine-scale turbulence. In this paper, we propose a data-driven coarse-mesh turbulence model based on local flow features for the transient analysis of thermal mixing and stratification in a sodium-cooled fast reactor. The model has a coarse-mesh setup to ensure computational efficiency, while it is trained by fine-mesh computational fluid dynamics (CFD) data to ensure accuracy. A novel neural network architecture, combining a densely connected convolutional network and a long-short-term-memory network, is developed that can efficiently learn from the spatial-temporal CFD transient simulation results. The neural network model was trained and optimized on a loss-of-flow transient and demonstrated high accuracy in predicting the turbulent viscosity field during the whole transient. The trained models generalization capability was also investigated on two other transients with different inlet conditions. The study demonstrates the potential of applying the proposed data-driven approach to support the coarse-mesh multi-dimensional modeling of advanced reactors.
