Paraphrase generation is an important task in natural language processing. Previous works focus on sentence-level paraphrase generation, while ignoring document-level paraphrase generation, which is a more challenging and valuable task. In this paper
, we explore the task of document-level paraphrase generation for the first time and focus on the inter-sentence diversity by considering sentence rewriting and reordering. We propose CoRPG (Coherence Relationship guided Paraphrase Generation), which leverages graph GRU to encode the coherence relationship graph and get the coherence-aware representation for each sentence, which can be used for re-arranging the multiple (possibly modified) input sentences. We create a pseudo document-level paraphrase dataset for training CoRPG. Automatic evaluation results show CoRPG outperforms several strong baseline models on the BERTScore and diversity scores. Human evaluation also shows our model can generate document paraphrase with more diversity and semantic preservation.
We theoretically study potential unconventional superconductivity in doped AB-type IV-VI semi- conductors, based on a minimal effective model with interaction up to the next-nearest neighbors. According to the experimental implications, we focus on t
he spin-triplet channels and obtain the superconducting phase diagram with respect to the anisotropy of the Fermi surfaces and the inter- action strength. All the states in the phase diagram are time reversal invariant and are topologically nontrivial. Specifically, in the phase diagram there appear a mirror symmetry protected topological Dirac superconductor phase, a mirror symmetry protected second-order topological superconductor phase, and a full-gap topological superconductor phase with winding number 4. The point group symmetry breaking of each superconducting ground state is also discussed.
With the rapid increase of available digital data, DNA storage is identified as a storage media with high density and capability of long-term preservation, especially for archival storage systems. However, the encoding density (i.e., how many binary
bits can be encoded into one nucleotide) and error handling are two major factors intertwined in DNA storage. Considering encoding density, theoretically, one nucleotide can encode two binary bits (upper bound). However, due to biochemical constraints and other necessary information associated with payload, the encoding densities of various DNA storage systems are much less than this upper bound. Additionally, all existing studies of DNA encoding schemes are based on static analysis and really lack the awareness of dynamically changed digital patterns. Therefore, the gap between the static encoding and dynamic binary patterns prevents achieving a higher encoding density for DNA storage systems. In this paper, we propose a new Digital Pattern-Aware DNA storage system, called DP-DNA, which can efficiently store digital data in DNA storage with high encoding density. DP-DNA maintains a set of encoding codes and uses a digital pattern-aware code (DPAC) to analyze the patterns of a binary sequence for a DNA strand and selects an appropriate code for encoding the binary sequence to achieve a high encoding density. An additional encoding field is added to the DNA encoding format, which can distinguish the encoding scheme used for those DNA strands, and thus we can decode DNA data back to its original digital data. Moreover, to further improve the encoding density, a variable-length scheme is proposed to increase the feasibility of the coding scheme with a high encoding density. Finally, the experimental results indicate that the proposed DP-DNA achieves up to 103.5% higher encoding densities than prior work.
Transient XUV spectroscopy is growing in popularity for the measurement of solar fuel and photovoltaic materials as it can separately measure electron and hole energies for multiple elements at once. However, interpretation of transient XUV measureme
nts is complicated by changes in core-valence exciton and angular momentum effects after photoexcitation. Here, we report the photoexcited electron and hole dynamics for ZnTe, a promising material for CO2 reduction, following 400 nm excitation. We apply a newly developed, ab-initio theoretical approach based on density functional theory and the Bethe-Salpeter equation to accurately predict the excited state change in the measured transient XUV spectra. Electrons excited to the conduction band are measured with a thermalization rate of 70 $pm$ 40 fs. Holes are excited with an average excess energy of ~1 eV and thermalize in 1130 $pm$ 150 fs. The theoretical approach also allows an estimated assignment of inter- and intra-valley relaxation pathways in k-space using the relative amplitudes of the core-valence excitons.
Brain age estimation based on magnetic resonance imaging (MRI) is an active research area in early diagnosis of some neurodegenerative diseases (e.g. Alzheimer, Parkinson, Huntington, etc.) for elderly people or brain underdevelopment for the young g
roup. Deep learning methods have achieved the state-of-the-art performance in many medical image analysis tasks, including brain age estimation. However, the performance and generalisability of the deep learning model are highly dependent on the quantity and quality of the training data set. Both collecting and annotating brain MRI data are extremely time-consuming. In this paper, to overcome the data scarcity problem, we propose a generative adversarial network (GAN) based image synthesis method. Different from the existing GAN-based methods, we integrate a task-guided branch (a regression model for age estimation) to the end of the generator in GAN. By adding a task-guided loss to the conventional GAN loss, the learned low-dimensional latent space and the synthesised images are more task-specific. It helps to boost the performance of the down-stream task by combining the synthesised images and real images for model training. The proposed method was evaluated on a public brain MRI data set for age estimation. Our proposed method outperformed (statistically significant) a deep convolutional neural network based regression model and the GAN-based image synthesis method without the task-guided branch. More importantly, it enables the identification of age-related brain regions in the image space. The code is available on GitHub (https://github.com/ruizhe-l/tgb-gan).
The reliance on vision for tasks related to cooking and eating healthy can present barriers to cooking for oneself and achieving proper nutrition. There has been little research exploring cooking practices and challenges faced by people with visual i
mpairments. We present a content analysis of 122 YouTube videos to highlight the cooking practices of visually impaired people, and we describe detailed practices for 12 different cooking activities (e.g., cutting and chopping, measuring, testing food for doneness). Based on the cooking practices, we also conducted semi-structured interviews with 12 visually impaired people who have cooking experience and show existing challenges, concerns, and risks in cooking (e.g., tracking the status of tasks in progress, verifying whether things are peeled or cleaned thoroughly). We further discuss opportunities to support the current practices and improve the independence of people with visual impairments in cooking (e.g., zero-touch interactions for cooking). Overall, our findings provide guidance for future research exploring various assistive technologies to help people cook without relying on vision.
Graphical User Interface (GUI) provides visual bridges between software apps and end users. However, due to the compatibility of software or hardware, UI display issues such as text overlap, blurred screen, image missing always occur during GUI rende
ring on different devices. Because these UI display issues can be found directly by human eyes, in this paper, we implement an online UI display issue detection tool OwlEyes-Online, which provides a simple and easy-to-use platform for users to realize the automatic detection and localization of UI display issues. The OwlEyes-Online can automatically run the app and get its screenshots and XML files, and then detect the existence of issues by analyzing the screenshots. In addition, OwlEyes-Online can also find the detailed area of the issue in the given screenshots to further remind developers. Finally, OwlEyes-Online will automatically generate test reports with UI display issues detected in app screenshots and send them to users. The OwlEyes-Online was evaluated and proved to be able to accurately detect UI display issues. Tool Link: http://www.owleyes.online:7476 Github Link: https://github.com/franklinbill/owleyes Demo Video Link: https://youtu.be/002nHZBxtCY
Transient X-ray absorption techniques can measure ultrafast dynamics of the elemental edges in a material or multiple layer junction, giving them immense potential for deconvoluting concurrent processes. However, the interpretation of the photoexcite
d changes to an X-ray edge is not as simple as directly probing a transition with optical or infrared wavelengths. The core hole left by the core-level transition distorts the measured absorption and reflection spectra, both hiding and revealing different aspects of a photo-induced process. In this perspective, we describe the implementation and interpretation of transient X-ray experiments. This description includes a guide of how to choose the best wavelength and corresponding X-ray sources when designing an experiment. As an example, we focus on the rising use of extreme ultraviolet (XUV) spectroscopy for understanding performance limiting behaviors in solar energy materials, such as measurements of polaron formation, electron and hole kinetics, and charge transport in each layer of a metal-oxide-semiconductor junction. The ability of measuring photoexcited carriers in each layer of a multilayer junction could prove particularly impactful in the study of molecules, materials, and their combinations that lead to functional devices in photochemistry and photoelectrochemistry.
Direct optimization of IR metrics has often been adopted as an approach to devise and develop ranking-based recommender systems. Most methods following this approach aim at optimizing the same metric being used for evaluation, under the assumption th
at this will lead to the best performance. A number of studies of this practice bring this assumption, however, into question. In this paper, we dig deeper into this issue in order to learn more about the effects of the choice of the metric to optimize on the performance of a ranking-based recommender system. We present an extensive experimental study conducted on different datasets in both pairwise and listwise learning-to-rank scenarios, to compare the relative merit of four popular IR metrics, namely RR, AP, nDCG and RBP, when used for optimization and assessment of recommender systems in various combinations. For the first three, we follow the practice of loss function formulation available in literature. For the fourth one, we propose novel loss functions inspired by RBP for both the pairwise and listwise scenario. Our results confirm that the best performance is indeed not necessarily achieved when optimizing the same metric being used for evaluation. In fact, we find that RBP-inspired losses perform at least as well as other metrics in a consistent way, and offer clear benefits in several cases. Interesting to see is that RBP-inspired losses, while improving the recommendation performance for all uses, may lead to an individual performance gain that is correlated with the activity level of a user in interacting with items. The more active the users, the more they benefit. Overall, our results challenge the assumption behind the current research practice of optimizing and evaluating the same metric, and point to RBP-based optimization instead as a promising alternative when learning to rank in the recommendation context.
Although deep neural networks have been widely employed and proven effective in sentiment analysis tasks, it remains challenging for model developers to assess their models for erroneous predictions that might exist prior to deployment. Once deployed
, emergent errors can be hard to identify in prediction run-time and impossible to trace back to their sources. To address such gaps, in this paper we propose an error detection framework for sentiment analysis based on explainable features. We perform global-level feature validation with human-in-the-loop assessment, followed by an integration of global and local-level feature contribution analysis. Experimental results show that, given limited human-in-the-loop intervention, our method is able to identify erroneous model predictions on unseen data with high precision.
