Recently, the character-word lattice structure has been proved to be effective for Chinese named entity recognition (NER) by incorporating word information. However, since is complex and dynamic, lattice-based models are hard fully utilize parallel computation of GPUs usually have a low inference speed. In this paper, we propose FLAT: Flat-LAttice Transformer NER, which converts into flat consisting spans. Each span corresponds character or latent its position in original lattice. With power...

The Bidirectional long short-term memory networks (BiLSTM) have been widely used as an encoder in models solving the named entity recognition (NER) task. Recently, Transformer is broadly adopted various Natural Language Processing (NLP) tasks owing to its parallelism and advantageous performance. Nevertheless, performance of NER not good it other NLP tasks. In this paper, we propose TENER, a architecture adopting adapted Encoder model character-level features word-level features. By...

Pre-Trained Models have been widely applied and recently proved vulnerable under backdoor attacks: the released pre-trained weights can be maliciously poisoned with certain triggers. When triggers are activated, even fine-tuned model will predict pre-defined labels, causing a security threat. These backdoors generated by poisoning methods erased changing hyper-parameters during fine-tuning or detected finding In this paper, we propose stronger weight-poisoning attack method that introduces...

Named entity recognition (NER) is the task to detect and classify spans in text. When overlap between each other, named as nested NER. Span-based methods have been widely used tackle Most of these get a score matrix, where entry corresponds span. However, previous work ignores spatial relations matrix. In this paper, we propose using Convolutional Neural Network (CNN) model relations. Despite being simple, experiments three commonly NER datasets show that our surpasses several recently...

The role of hydrogen in nitrogen-doped ZnO thin films was studied by Fourier transform infrared (FTIR) absorption and modeled first-principles calculations to understand the difficulty doping p-type with nitrogen. Nitrogen-doped were fabricated low-pressure metal-organic chemical vapor deposition (MOCVD). High levels nitrogen incorporation observed, but acceptor concentrations remained low. Theoretical analysis suggests there is a high probability that NO− H+ charged defects combine form...

First-principles calculations show that AB defects substituting on an O site in ZnO where A, B=N, O, or C are important class of whose physical properties cannot be described by the usual split interstitials but rather substitutional diatomic molecules. The molecular natures (AB)O reflected their vibrational frequencies which redshifted from those corresponding free molecules only about 10%. These calculated results agree with frequency range recently observed IR measurement N-doped ZnO....

Xiaonan Li, Kai Lv, Hang Yan, Tianyang Lin, Wei Zhu, Yuan Ni, Guotong Xie, Xiaoling Wang, Xipeng Qiu. Proceedings of the 61st Annual Meeting Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Purpose This study aims to use a bottom-up, inductive approach derive destination image attributes from large quantities of online consumer narratives and establish classification system based on relationships among places. Design/methodology/approach Content social network analyses were used explore the structure for destinations narratives. Cluster analysis was then group by attributes, ANOVA provided comparisons. Findings Twenty-two identified combined into three groups (core, expected,...

Two experiments designed to assist in understanding the physics of certain back contacts on p-type CdTe solar-cell devices are described. In first experiment, x-ray photoelectron and Auger electron spectroscopies used show that etching HNO3:H3PO4 results a Te layer surface. second photoemission spectroscopy is explore electronic properties evaporated deposited thin-film, polycrystalline p-CdTe an effort develop band diagram for Te/p-CdTe interface. The motivation developing derives from...

This article shows the development and testing of a microchip with integrated electrochemical sensors for measurement pH, temperature, dissolved oxygen viable biomass concentration under yeast cultivation conditions. Measurements were done both dynamic batch conditions as well prolonged continuous The response compared conventional techniques. sensor was based on impedance spectroscopy. results matched very dry weight measurements showed limit detection approximately 1 g/L. monitored...

In the past few years, satellite communications have greatly affected our daily lives, and integrated terrestrial-satellite network can combine advantages of base stations (BSs) to provide wider coverage lower cost. Because resources are limited, how allocate through effective methods has become a major challenge. This paper proposes framework for resource allocation based on non-orthogonal multiple access (NOMA). Then, deployment method local cache pools is given achieve time delay maximize...

In-context learning is a new paradigm where language model observes few examples and directly outputs the test input's prediction. Previous works have shown that it sensitive to provided randomly sampled probably cause inferior performance. In this paper, we propose finding "support examples" for in-context learning: Given training dataset, aims select one permutation of examples, which can well characterize task thus lead superior Although traditional gradient-based training, there are...

The induction of oxygen radicals and oxidative stress are major pathways through which nanomaterials cause adverse health effects. Dynamic monitoring redox processes in living cells exposed to is currently limited due the inadequacy conventional methods. Herein, we construct a Grx1-roGFP2 (glutaredoxin 1 fused with redox-sensitive Green Fluorescent Protein 2) protein sensor expressed Madin-Darby Canine Kidney (MDCK) that allows dynamic analysis metal oxide (MOx) nanoparticle-induced stress....

Forming a low-resistance contact to p-type CdTe is critical issue for successful commercialization of CdTe-based photovoltaic devices. One solution this problem has been incorporate surface pretreatments facilitate formation. In article, the effects nitric–phosphoric (NP) acid pretreatment on material properties and device performance are investigated polycrystalline We demonstrate that NP pretreatment, when applied thin films, forms thick, highly conductive Te layer back film exposed grain...

Oil-paper insulation of higher voltage winding and traction in on-board transformers frequently withstand large AC fluctuation voltages mainly combined with inter-harmonic components. This paper presents the characteristics partial discharge (PD) oil-paper under superimposed voltages. The experimental results show that addition to fundamental sinusoidal component variation peak dv/dt along phase. has a significant impact on PD including inception (PDIV), breakdown voltage, repetition rate,...