The Fact Extraction and VERification (FEVER) shared task was launched to support the development of systems able to verify claims by extracting supporting or refuting facts from raw text. The shared task organizers provide a large-scale dataset for the consecutive steps involved in claim verification, in particular, document retrieval, fact extraction, and claim classification. In this paper, we present our claim verification pipeline approach, which, according to the preliminary results, scored third in the shared task, out of 23 competing systems. For the document retrieval, we implemented a new entity linking approach. In order to be able to rank candidate facts and classify a claim on the basis of several selected facts, we introduce two extensions to the Enhanced LSTM (ESIM).
Polarization optics plays a pivotal role in diffractive, refractive, and emerging flat optics, and has been widely employed in contemporary optical industries and daily life. Advanced polarization manipulation leads to robust control of the polarization direction of light. Nevertheless, polarization control has been studied largely independent of the phase or intensity of light. Here, we propose and experimentally validate a Malus-metasurface-assisted paradigm to enable simultaneous and independent control of the intensity and phase properties of light simply by polarization modulation. The orientation degeneracy of the classical Malus's law implies a new degree of freedom and enables us to establish a one-to-many mapping strategy for designing anisotropic plasmonic nanostructures to engineer the Pancharatnam-Berry phase profile, while keeping the continuous intensity modulation unchanged. The proposed Malus metadevice can thus generate a near-field greyscale pattern, and project an independent far-field holographic image using an ultrathin and single-sized metasurface. This concept opens up distinct dimensions for conventional polarization optics, which allows one to merge the functionality of phase manipulation into an amplitudemanipulation-assisted optical component to form a multifunctional nano-optical device without increasing the complexity of the nanostructures. It can empower advanced applications in information multiplexing and encryption, anti-counterfeiting, dual-channel display for virtual/augmented reality, and many other related fields.
Metasurfaces have recently been used for multichannel image displays with pixel-size lower than a wavelength, which indicates the potential application in ultracompact anticounterfeiting with high-density and hidden information. However, current multichannel metasurfaces applied in anticounterfeiting are based on the sophisticated nanostructure design or at the cost of giving up some controls on the optical transmission matrix to encode multiple information channels. That is, the overall degrees of freedom offered by these metasurfaces are a "zero-sum game". Here, inspired by the orientation degeneracy indicated in Malus law, we propose a multiplexed anticounterfeiting metasurface consisting of single-sized nanostructures, which provide a new degree of freedom to increase the information capacity of anticounterfeiting without burdening the nanostructure design and fabrication. Specifically, the proposed metasurfaces can record a continuous grayscale image (channel 1) multiplexed with a totally/partially independent, interrelated, or watermarked anticounterfeiting pattern (channel 2). The two channels can be readily switched by polarization control. All experimental metasurface-images (meta-images) with high fidelity agree well with our design. With advantages such as ultracompactness, high-density information, multichannel displays, and strong concealment, the anticounterfeiting metasurfaces can empower advanced research and applications of metasurfaces in high-end optical anticounterfeiting and many other related fields.
Efficient transmission-type meta-holograms have been demonstrated using high-index dielectric nanostructures based on Huygens' principle. It is crucial that the geometry size of building blocks be judiciously optimized individually for spectral overlap of electric and magnetic dipoles. In contrast, reflection-type meta-holograms using the metal/insulator/metal scheme and geometric phase can be readily achieved with high efficiency and small thickness. Here, we demonstrate a general platform for design of dual magnetic resonance based meta-holograms based on the geometric phase using silicon nanostructures that are quarter wavelength thick for visible light. Significantly, the projected holographic image can be unambiguously observed without a receiving screen even under the illumination of natural light. Within the well-developed semiconductor industry, our ultrathin magnetic resonance-based meta-holograms may have promising applications in anticounterfeiting and information security.
Optical metasurfaces can offer high-quality multichannel displays by modulating different degrees of freedom of light, demonstrating great potential in the next generation of optical encryption and anti-counterfeiting. Different from the direct imaging modality of metasurfaces, single-pixel imaging (SPI) as a typical computational imaging technique obtains the object image from a decryption-like computational process. Here, we propose an optical encryption scheme by introducing metasurface-images (meta-images) into the encoding and decoding processes as the keys of SPI encryption. Different high-quality meta-images generated by a dual-channel Malus metasurface play the role of keys to encode multiple target images and retrieve them following the principle of SPI. Our work eliminates the conventional digital transmission process of keys in SPI encryption, enables the reusability of a single metasurface in different encryption processes, and thereby paves the way toward a high-security optical encryption between direct and indirect imaging methods.
Automated fact-checking based on machine learning is a promising approach to identify false information distributed on the web. In order to achieve satisfactory performance, machine learning methods require a large corpus with reliable annotations for the different tasks in the fact-checking process. Having analyzed existing fact-checking corpora, we found that none of them meets these criteria in full. They are either too small in size, do not provide detailed annotations, or are limited to a single domain. Motivated by this gap, we present a new substantially sized mixed-domain corpus with annotations of good quality for the core fact-checking tasks: document retrieval, evidence extraction, stance detection, and claim validation. To aid future corpus construction, we describe our methodology for corpus creation and annotation, and demonstrate that it results in substantial inter-annotator agreement. As baselines for future research, we perform experiments on our corpus with a number of model architectures that reach high performance in similar problem settings. Finally, to support the development of future models, we provide a detailed error analysis for each of the tasks. Our results show that the realistic, multi-domain setting defined by our data poses new challenges for the existing models, providing opportunities for considerable improvement by future systems. 12 underlined words in the resolution are hyperlinks 13 https://www.mturk.com/
By virtue of the extraordinary capability of manipulating the polarization state, amplitude and phase of electromagnetic fields, metasurfaces can be employed to display holographic or nanoprinting images with unprecedented spatial resolution. Bringing holography and nanoprinting together is an effective way toward information multiplexing. However, current approaches mostly utilize interleaving or stacking nanostructures with different functionalities to construct multiplexed metasurfaces, hence they are equivalent to a combination of several metasurfaces and the information capacity of each metasurface remains unchanged. Here, by combining intensity modulation governed by Malus's law with phase manipulation based on both geometric and propagation phases, a single‐cell‐designed metasurface for three‐channel image displays is proposed. The new design strategy can significantly improve the information capacity since the extra phase modulation originates from the orientation degeneracy and dimension variation of nanostructures rather than multilayer or interleaving design. Specifically, a three‐channel metasurface is experimentally demonstrated, which can simultaneously record a continuous grayscale nanoprinting image in the near field and project two independent holographic images in the far field. With the advantages of crosstalk‐free and ultracompactness, the proposed three‐channel metasurfaces can empower the design of multifunctional nano‐optical elements for applications in image displays, optical anticounterfeiting, optical storage and many other related fields.
With the rapid progress in computer science, including artificial intelligence, big data and cloud computing, full-space spot generation can be pivotal to many practical applications, such as facial recognition, motion detection, augmented reality, etc. These opportunities may be achieved by using diffractive optical elements (DOEs) or light detection and ranging (LIDAR). However, DOEs suffer from intrinsic limitations, such as demanding depth-controlled fabrication techniques, large thicknesses (more than the wavelength), Lambertian operation only in half space, etc. LIDAR nevertheless relies on complex and bulky scanning systems, which hinders the miniaturization of the spot generator. Here, inspired by a Lambertian scatterer, we report a Hermitian-conjugate metasurface scrambling the incident light to a cloud of random points in full space with compressed information density, functioning in both transmission and reflection spaces. Over 4044 random spots are experimentally observed in the entire space, covering angles at nearly 90°. Our scrambling metasurface is made of amorphous silicon with a uniform subwavelength height, a nearly continuous phase coverage, a lightweight, flexible design, and low-heat dissipation. Thus, it may be mass produced by and integrated into existing semiconductor foundry designs. Our work opens important directions for emerging 3D recognition sensors, such as motion sensing, facial recognition, and other applications.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.