Search

Graph Convolutional Network (GCN) is a critical method to capture non-sequential information of sentences and recognize long-distance syntactic information. However, the adjacency matrix of GCN has two problems: redundant syntactic information and wrong dependency parsing results. Because the syntactic information is represented by unweighted adjacency matrices in most existing GCN methods. Toward this end, we propose a novel model, PGCN-EA, using Piecewise Graph Convolutional Network with Edge-level Attention to address these two problems. In specific, we first employ the piecewise adjacency matrix based on entity pair, which aims to dynamically reduce the sentence's redundant features. Second, we propose Edge-level Attention to assign the different weights among nodes based on GCN's input and create the weight adjacency matrix, emphasizing the importance of child words with the target word and alleviating the influence of wrong dependency parsing. Our model on a benchmark dataset has carried out extensive experiments and achieved the best PR curve as compared to seven baseline models, which are at least more than 2.3 % . [ABSTRACT FROM AUTHOR]

As of present, the progress of conversational AI research has been greatly propelled by large-scale pre-trained language models. In particular, task-oriented dialogue systems have gained widespread attention owing to their immense potential in helping individuals accomplish diverse objectives, such as booking hotels, making restaurant reservations, and purchasing train tickets. In the past, task-oriented dialogue systems were typically viewed as a multi-step process that included spoken language understanding, dialogue state tracking, dialogue policy learning, and natural language generation. More recently, large-scale pre-trained language models enables the development of end-to-end neural pipeline task-oriented dialogue systems, which combine multiple steps into a single model, allowing for joint optimization and preventing error propagation. However, in order to explicitly retrieve information from databases to ensure the interpretability of the system, almost all end-to-end neural pipeline methods inevitably require predicting dialogue state as an intermediate result specialized for the domain or task, which results in significant challenges for generalization. To solve the problem above, we propose O ne- S tep T ask- O riented D ialogue (OSTOD) in this paper, which models task-oriented dialogue by synchronously generating activated states and retelling responses, where activated states refer to slot values that contribute to database access, and retelling responses are system responses that contain activated state information. Specifically, first, automatic methods are designed to build data containing activated states and retelling responses. Then, a joint generation model that synchronously predicts activated states and retelling responses in a single step is proposed for task-oriented dialogue modelling. Based on empirical results obtained from the MultiWOZ 2.0 and MultiWOZ 2.1 datasets, our OSTOD model demonstrates comparable performance to state-of-the-art baselines. Moreover, our model exhibits exceptional generalization capabilities in few-shot learning and domain transfer scenarios. [ABSTRACT FROM AUTHOR]

Machine reading comprehension (MRC) is a fundamental task of evaluating the natural language understanding ability of model, which requires complicated reasoning about the knowledge involved in the context as well as world knowledge. However, most existing approaches ignore the complicated reasoning process and solve it with a one-step "black box" model and massive data augmentation. Therefore, in this paper, we propose a modular knowledge reasoning approach based on neural network modules that explicitly model each reasoning process step. Five reasoning modules are designed and learned in an end-to-end manner, which leads to a more interpretable model. Experiments using the reasoning over paragraph effects in situations (ROPES) dataset, a challenging dataset that requires reasoning over paragraph effects in a situation, demonstrate the effectiveness and explainability of our proposed approach. Moreover, the transfer of our reasoning modules to the WinoGrande dataset under the zero-shot setting achieved competitive results compared with the data augmented model, proving the generalization capability. [ABSTRACT FROM AUTHOR]

Video visual relation inference refers to the task of automatically detecting the relation triplets between the observed objects in videos with the form of ${ < subject, predicate, object>}$ , which requires correctly labeling each detected object and their interaction predicates. Despite the recent advances in image visual relation detection using deep learning techniques, relation inference in videos remains a challenging topic. On one hand, since the introduction of temporal information, it needs to model the rich spatio-temporal visual information for objects and videos. On the other hand, wild videos are often annotated with incomplete relation triplet tags and a few of them are semantically overlapped. However, previous methods adopt hand-crafted visual features extracted from the trajectories, describing local appearance characteristics of isolated objects. And they treat the problem as a multi-class classification task, which makes the relation tags mutually exclusive. To address the above issues, we propose a novel model, termed Visual-Semantic Relation Network (VSRN). In this network, we leverage three-dimensional convolution kernel to capture spatio-temporal features, and encode global visual features in videos through pooling operation on each time slice. Moreover, the semantic collocations between objects are also incorporated so as to obtain comprehensive representations of the relationships. For relation classification, we treat the problem as a multi-label classification task and regard each tag to be independent to predict various relationships. Additionally, we modify commonly used evaluation metric, video-wise recall, to a pair-wise metric (Roop) for testing the performance of models in predicting multiple relationships for the object pairs, Extensive experimental results on two large-scale datasets demonstrate the effectiveness of our proposed model which significantly outperforms the previous works. [ABSTRACT FROM AUTHOR]

The knowledge-grounded dialogue task aims to generate responses that convey information from given knowledge documents. However, it is a challenge for the current sequence-based model to acquire knowledge from complex documents and integrate it to perform correct responses without the aid of an explicit semantic structure. To address these issues, we propose a novel graph structure, Grounded Graph (G 2), that models the semantic structure of both dialogue and knowledge to facilitate knowledge selection and integration for knowledge-grounded dialogue generation. We also propose a Grounded Graph Aware Transformer ( G 2 A T) model that fuses multi-forms knowledge (both sequential and graphic) to enhance knowledge-grounded response generation. Our experiments results show that our proposed model outperforms the previous state-of-the-art methods with more than 10% gains in response generation and nearly 20% improvement in factual consistency. Further, our model reveals good generalization ability and robustness. By incorporating semantic structures as prior knowledge in deep neural networks, our model provides an effective way to aid language generation. • An explicit semantic graph structure of knowledge-grounded dialogue is explored for the first time. • A novel method of syntax-based structure extraction is explicitly developed for knowledge-grounded dialogue. • A cutting-edge framework that integrates sequence and graph structure for knowledge-grounded dialogue is proposed. • State-of-the-art results of response generation and factual consistency for knowledge-grounded dialogue are reported. [ABSTRACT FROM AUTHOR]

As one of the pivotal technologies leading towards embodied intelligence, audio-visual segmentation is geared towards achieving precise segmentation of sounding objects, offering vast application prospects in scenarios such as emergency rescue and natural exploration. Nevertheless, the performance of audio-visual segmentation technology encounters limitations stemming from challenges related to the adaptation and fusion of cross-modal information encoding, as well as the decoding and generation of masks. To address these issues, this paper explores the adaptation of multi-modal information based on a shared encoder by employing a neural architecture search method to design a hierarchical encoder cooperation module for enhanced information interaction. An intermediate loss is leveraged to help the encoder to keep spatial knowledge reserved. Furthermore, an audio-guided class-aware decoder is devised to guide the generation of masks. Our approach has yielded competitive experimental results across multiple datasets, thus substantiating its effectiveness. [ABSTRACT FROM AUTHOR]

News-driven stock prediction investigates the correlation between news events and stock price movements. Previous work has considered effective ways for representing news events and their sequences, but rarely exploited the representation of underlying equity states. We address this issue by making use of a recurrent neural network to represent an equity state transition sequence, integrating news representation using contextualized representations as inputs to the state transition mechanism. Thanks to the separation of news and equity representations, our model can accommodate additional input factors. We design a novel random noise factor for modeling influencing factors beyond news events, and a future event factor to address the delay of news information (e.g., insider trading) and reduce the learning difficulties. Results show that the proposed model outperforms strong baselines in the literature. [ABSTRACT FROM AUTHOR]

Cross-domain word representation aims to learn high-quality semantic representations in an under-resourced domain by leveraging information in a resourceful domain. However, most existing methods mainly transfer the semantics of common words across domains, ignoring the semantic relations among domain-specific words. In this paper, we propose a domain structure-based transfer learning method to learn cross-domain representations by leveraging the relations among domain-specific words. To accomplish this, we first construct a semantic graph to capture the latent domain structure using domain-specific co-occurrence information. Then, in the domain adaptation process, beyond domain alignment, we employ Laplacian Eigenmaps to ensure the domain structure is consistently distributed in the learned embedding space. As such, the learned cross-domain word representations not only capture shared semantics across domains, but also maintain the latent domain structure. We performed extensive experiments on two tasks, namely sentiment analysis and query expansion. The experiment results show the effectiveness of our method for tasks in under-resourced domains. • A cross-domain word representation method for tasks with limited resources. • Fusing the latent semantics of the source and target domains. • Propagating common semantic knowledge from source domain to target domain. • Preserving domain structure in the domain adaptation process. • Effective and efficient for text-related and low-resource tasks. [ABSTRACT FROM AUTHOR]

Knowledge has long been cosnsidered a crucial part of natural language understanding. Many knowledge bases have been constructed, but none of them will ever be complete. Nevertheless, we argue that complete knowledge already exists in natural language. Most previous work on question answering retrieved such knowledge using traditional statistical methods, and consequently, the knowledge retrieved contained co-occurring phrases and could not provide guidance for model understanding and reasoning. Therefore, in addition to demonstrating the effectiveness of natural language knowledge in machine understanding, this study presents a novel knowledge retrieval approach that evaluates the importance of knowledge from a semantic perspective. Furthermore, we propose a knowledge revise mechanism that allows the model to revise the retrieved knowledge from local and global perspectives. We demonstrate our Semantic-rank-and-Knowledge-Revise-based Question Answering (SKR-QA) approach on two challenging multi-choice question and answering tasks: ARC–Challenge and OpenbookQA. Compared with the previous State-of-the-Art (SOTA) models, our work achieves consistent improvements. Moreover, the knowledge obtained by our method is more conducive to machine understanding, thus providing certain interpretability. [ABSTRACT FROM AUTHOR]

The Graph Convolutional Network (GCN) is a universal relation extraction method that can predict relations of entity pairs by capturing sentences' syntactic features. However, existing GCN methods often use dependency parsing to generate graph matrices and learn syntactic features. The quality of the dependency parsing will directly affect the accuracy of the graph matrix and change the whole GCN's performance. Because of the influence of noisy words and sentence length in the distant supervised dataset, using dependency parsing on sentences causes errors and leads to unreliable information. Therefore, it is difficult to obtain credible graph matrices and relational features for some special sentences. In this article, we present a Multi-Graph Cooperative Learning model (MGCL), which focuses on extracting the reliable syntactic features of relations by different graphs and harnessing them to improve the representations of sentences. We conduct experiments on a widely used real-world dataset, and the experimental results show that our model achieves the state-of-the-art performance of relation extraction. [ABSTRACT FROM AUTHOR]

In natural language, a group of words constitute a phrase and several phrases constitute a sentence. However, existing transformer-based models for sentence-level tasks abstract sentence-level semantics from word-level semantics directly, which override phrase-level semantics so that they may be not favorable for capturing more precise semantics. In order to resolve this problem, we propose a novel multi-granularity semantic representation (MGSR) model for relation extraction. This model can bridge the semantic gap between low-level semantic abstraction and high-level semantic abstraction by learning word-level, phrase-level, and sentence-level multi-granularity semantic representations successively. We segment a sentence into entity chunks and context chunks according to an entity pair. Thus, the sentence is represented as a non-empty segmentation set. The entity chunks are noun phrases, and the context chunks contain the key phrases expressing semantic relations. Then, the MGSR model utilizes inter-word, inner-chunk and inter-chunk three kinds of different self-attention mechanisms, respectively, to learn the multi-granularity semantic representations. The experiments on two standard datasets demonstrate our model outperforms the previous models. [ABSTRACT FROM AUTHOR]

Video visual relation recognition aims at mining the dynamic relation instances between objects in the form of 〈 subject , predicate , object 〉 , such as "person1-towards-person2" and "person-ride-bicycle". Existing solutions treat the problem as several independent sub-tasks, i.e., image object detection, video object tracking and trajectory-based relation prediction. We argue that such separation results in the lack of information flow between different sub-models, which creates redundant representation while each sub-task cannot share a common set of task-specific features. Toward this end, we connect these three sub-tasks in an end-to-end manner by proposing the 3-D relation proposal that serves as a bridge for relation feature learning. Specifically, we put forward a novel deep neural network, named 3DRN, to fuse the spatio-temporal visual characteristics, object label features, and spatial interactive features for learning the relation instances with multi-modal cues. In addition, a three-staged training strategy is also provided to facilitate large-scale parameter optimization. We conduct extensive experiments on two public datasets with different emphasis to demonstrate the effectiveness of the proposed end-to-end feature learning method for visual relation recognition in videos. Furthermore, we verify the potential of our approach by tackling the video relation detection task. [ABSTRACT FROM AUTHOR]

Since the length of microblog texts, such as tweets, is strictly limited to 140 characters, traditional Information Retrieval techniques suffer from the vocabulary mismatch problem severely and cannot yield good performance in the context of microblogosphere. To address this critical challenge, in this paper, we focus on the use of local conceptual word embeddings for enhance microblog retrieval effectiveness. In particular, we propose a novel k-Nearest Neighbor (k NN) based Query Expansion (QE) algorithm to generate words from local word embeddings to expand the original query, which leads to better understanding of the information need. Besides, in order to further satisfy users’ real-time information need, we incorporate temporal evidences into the expansion algorithm, which can boost recent tweets in the retrieval results with respect to a given topic. Experimental results on the official TREC Twitter corpora demonstrate the significant superiority of our approach over baseline methods. [ABSTRACT FROM AUTHOR]

In practical dialogue systems, it is crucial to avoid undesired responses and poor user experiences by detecting Out-Of-Scope (OOS) intents from user utterances. Currently, to detect OOS intents, limited-supervised methods are more potential due to using limited true OOS data to better model the OOS data distribution. However, existing limited-supervised methods often yield low-quality OOS augmentations, i.e., noisy OOS and long-distant OOS, due to the use of the retrieval-based data augmentation mechanism, which will damage the OOS intent detection performance. To tackle this problem, in this paper, we propose a novel OOS intent detection method by leveraging intent-invariant data augmentation, called InInOOS, which can generate high-quality pseudo-OOS utterances with invariant OOS intents but various slot values and different expressions to further enhance the OOS intent detection performance. Specifically, we first generate pseudo-OOS candidates using our pre-trained intent-invariant utterance generation model, and then the most beneficial candidates are selected to train an OOS intent detection model. Extensive experiments on two public datasets show that our method performs better than state-of-the-art baselines. • Enhance Out-Of-Scope (OOS) intent detection with intent-invariant data augmentation. • Our method effectively reduces the two kinds of low-quality augmentations. • Our method balances the validity and the diversity of the augmented OOS data. • Extensive experiments show that our method outperforms state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

• We present a hypergraph network model for nested entity mention recognition. • We recognize nested entities by tagging hyperedges instead of nodes. • We propose a theorem to make hyperedges be easily denoted in the program. • We solve the data imbalance problem and reduce the computing cost. We propose a hypergraph network (HGN) model to recognize the nested entity mentions in texts. This model can learn the representations for the sequence structures of natural languages and the representations for the hypergraph structures of nested entity mentions. Mainstream methods recognize an entity mention by separately tagging the words or the gaps between words, which may complicate the problem and not be favorable for capturing the overall features of the mention. To solve these issues, the HGN model treats each entity mention as a whole and tags it with one label. We represent each sentence as a hypergraph, in which nodes represent words and hyperedges represent entity mentions. Thus, entity mention recognition (EMR) is transformed into a task of classifying the hyperedges. The HGN model firstly uses encoders to extract the features and learn a hypergraph representation, and then recognizes entity mentions by tagging every hyperedge. The experiments on three standard datasets demonstrate our model outperforms the previous models for nested EMR. We openly release the source code at https://github.com/nlplab-ie/HGN. [ABSTRACT FROM AUTHOR]

Linguistic knowledge is useful for various NLP tasks, but the difficulty lies in the representation and application. We consider that linguistic knowledge is implied in a large-scale corpus, while classification knowledge, the knowledge related to the definitions of entity and relation types, is implied in the labeled training data. Therefore, a corpus subgraph is proposed to mine more linguistic knowledge from the easily accessible unlabeled data, and sentence subgraphs are used to acquire classification knowledge. They jointly constitute a relation knowledge graph (RKG) to extract relations from sentences in this paper. On RKG, entity recognition can be regarded as a property value filling problem and relation classification can be regarded as a link prediction problem. Thus, the multiple relation extraction can be treated as a reasoning process for knowledge completion. We combine statistical reasoning and neural network reasoning to segment sentences into entity chunks and non-entity chunks, then propose a novel Chunk Graph LSTM network to learn the representations of entity chunks and infer the relations among them. The experiments on two standard datasets demonstrate our model outperforms the previous models for multiple relation extraction. [ABSTRACT FROM AUTHOR]

Neural machine translation (NMT) optimized by maximum likelihood estimation (MLE) usually lacks the guarantee of translation adequacy. To alleviate this problem, we propose an NMT approach that heightens the adequacy in machine translation by transferring the semantic knowledge from bilingual sentence alignment learning. Specifically, we first design a discriminator that learns to estimate sentence aligning score over translation candidates. The discriminator is constructed by gated self-attention based sentence encoders and trained with an N -pair loss for better capturing lexical evidences from bilingual sentence pairs. Then we propose an adversarial training framework as well as a sentence alignment-aware decoding method for NMT to transfer the discriminator's learned semantic knowledge to NMT models. We conduct our experiments on Chinese → English, Uyghur → Chinese and English → German translation tasks. Experimental results show that our proposed methods outperform baseline NMT models on all these three translation tasks. Further analysis also indicates the characteristics of our approaches and details the semantic knowledge that transfered from the discriminator to the NMT model. [ABSTRACT FROM AUTHOR]

In computer-aided translation tasks, reducing the time of reviewing and post-editing on translations is meaningful for human translators. However, existing studies mainly aim to improve overall translation quality, which only reduces post-editing time. In this work, we firstly identify testing sentences which are highly similar to training set (high-potential sentences) to reduce reviewing time, then we focus on improving corresponding translation quality greatly to reduce post-editing time. From this point, we firstly propose two novel translation memory methods to characterize similarity between sentences on syntactic and template dimensions separately. Based on that, we propose a similarity-aware neural machine translation (similarity-NMT) which consists of two independent modules: (1) Identification Module, which can identify high-potential sentences of testing set according to multi-dimensional similarity information; (2) Translation Module, which can integrate multi-dimensional similarity information of parallel training sentence pairs into an attention-based NMT model by leveraging posterior regularization. Experiments on two Chinese ⇒ English domains have well-validated the effectiveness and universality of the proposed method of reducing human translator efforts. [ABSTRACT FROM AUTHOR]

We present a novel end-to-end model to jointly extract semantic relations and argument entities from sentence texts. This model does not require any handcrafted feature set or auxiliary toolkit, and hence it could be easily extended to a wide range of sequence tagging tasks. A new method of using the word morphology feature for relation extraction is studied in this paper. We combine the word morphology feature and the semantic feature to enrich the representing capacity of input vectors. Then, an input information enhanced unit is developed for the bidirectional long short-term memory network (Bi-LSTM) to overcome the information loss caused by the gate operations and the concatenation operations in the LSTM memory unit. A new tagging scheme using uncertain labels and a corresponding objective function are exploited to reduce the interference information from non-entity words. Experiments are performed on three datasets: The New York Times (NYT) and ACE2005 datasets for relation extraction and the SemEval 2010 task 8 dataset for relation classification. The results demonstrate that our model achieves a significant improvement over the state-of-the-art model for relation extraction on the NYT dataset and achieves a competitive performance on the ACE2005 dataset. [ABSTRACT FROM AUTHOR]

Yunnan plays a pivotal role in transmitting electricity from west to east within China's Southern Power Grid. During 7–13 January 2021, a large-scale continuous ice-covering event of ultra-high voltage (UHV) transmission lines occurred in the Qujing area of eastern Yunnan Province. Based on ERA5 reanalysis data and meteorological observation data of UHV transmission line icing in China's Southern Power Grid, the synoptic causes of the icing are comprehensively analyzed from various perspectives, including weather situations, vertical stratification of temperature and humidity, local meteorological elements, and atmospheric circulation indices. The results indicate a strong East Asian trough and a blocking high directing northern airflow southward ahead of the ridge. Cold air enters the Qujing area and combines with warm and moist air from the subtropical high pressure of 50–110° E. As warm and cold air masses form a quasi-stationary front over the northern mountainous area of Qujing due to topographic uplift, the mechanism of "supercooling and warm rain" caused by the "warm–cold" temperature profile structure leads to freezing rain events. Large-scale circulation indices in the Siberian High, East Asian Trough, and 50–110° E Subtropical High regions provided clear precursor signals within 0–2 days before the icing events. [ABSTRACT FROM AUTHOR]

In recent years, significant progress has been made in the field of non-autoregressive machine translations. However, the accuracy of non-autoregressive models still lags behind their autoregressive counterparts. This discrepancy can be attributed to the abundance of repetitive tokens in the target sequences generated by non-autoregressive models. In this study, we delve into this phenomenon and propose a novel approach to train a non-autoregressive model using unlikelihood loss. We evaluate our method on three widely used benchmark tasks. The experimental results demonstrating that our proposed approach significantly reduces the number of repetitive tokens while improving the overall performance of non-autoregressive machine translations. Compared to the baseline model "Mask-Predict", the average number of repetitions on IWSLT 14 DE → EN valid set is reduced from 0.48 to 0.17, resulting in a remarkable 62% decrease. [ABSTRACT FROM AUTHOR]

• Hydroxylated modification and structural characterization of lecithin. • The ability to reduce surface tension and viscosity is improved. • A more obvious effect on reducing the viscosity of the HMX-based PBX slurry. Reducing viscosity is an important factor affecting the dispersion and forming process of polymer bonded explosive (PBX). In this study, lecithin was modified by hydroxyl reaction in order to enhance its viscosity reduction effect. The structure of hydroxylated lecithin was confirmed by nuclear magnetic resonance spectroscopy (NMR), infrared spectrum (IR) and iodine values. The decrease of surface tension of styrene and the decrease of viscosity of HMX-based PBX slurry before and after modification were analyzed. The results showed that hydroxylated lecithin had better effect on reducing surface tension and viscosity. [ABSTRACT FROM AUTHOR]

Most short-text embedding models typically represent each short-text only using the literal meanings of the words, which makes these models indiscriminative for the ubiquitous polysemy. In order to enhance the semantic representation capability of the short-texts, we (i) propose a novel short-text conceptualization algorithm to assign the associated concepts for each short-text, and then (ii) introduce the conceptualization results into learning the conceptual short-text embeddings. Hence, this semantic representation is more expressive than some widely-used text representation models such as the latent topic model. Wherein, the short-text conceptualization algorithm used here is based on a novel co-ranking framework, enabling the signals (i.e., the words and the concepts) to fully interplay to derive the solid conceptualization for the short-texts. Afterwards, we further extend the conceptual short-text embedding models by utilizing an attention-based model that selects the relevant words within the context to make more efficient prediction. The experiments on the real-world datasets demonstrate that the proposed conceptual short-text embedding model and short-text conceptualization algorithm are more effective than the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

We propose a high-quality compressive ghost imaging method based on projected Landweber regularization and guided filter, which effectively reduce the undersampling noise and improve the resolution. In our scheme, the original object is reconstructed by decomposing of regularization and denoising steps instead of solving a minimization problem in compressive reconstruction process. The simulation and experimental results show that our method can obtain high ghost imaging quality in terms of PSNR and visual observation. [ABSTRACT FROM AUTHOR]

Magnetic resonance imaging (MRI) reconstruction from the smallest possible set of Fourier samples has been a difficult problem in medical imaging field. In our paper, we present a new approach based on a guided filter for efficient MRI recovery algorithm. The guided filter is an edge-preserving smoothing operator and has better behaviors near edges than the bilateral filter. Our reconstruction method is consist of two steps. First, we propose two cost functions which could be computed efficiently and thus obtain two different images. Second, the guided filter is used with these two obtained images for efficient edge-preserving filtering, and one image is used as the guidance image, the other one is used as a filtered image in the guided filter. In our reconstruction algorithm, we can obtain more details by introducing guided filter. We compare our reconstruction algorithm with some competitive MRI reconstruction techniques in terms of PSNR and visual quality. Simulation results are given to show the performance of our new method. [ABSTRACT FROM AUTHOR]

• Ghost imaging under low sampling: the proposed method can obtain high quality imaging results with fewer sampling numbers; • Accelerated iteration: the accelerated projected Landweber iteration reduces the number of iterations, improves imaging efficiency; • High quality imaging results: the proposed scheme can extract the joint structures of image, and improve the resolution of imaging results. The realization of high-quality imaging under low sampling is an effective way to solve the practical application of ghost imaging. In this paper, we present an advanced framework of compressed ghost imaging under low sampling. During the imaging process, the regularization and mutual structure filtering operations are performed alternately, which we call mutual structure ghost imaging (MSGI). In the joint filtering of our proposed scheme, the mutual-structural information contained in both the reference image and the target one is applied to enhance the capability of important edge preserving. Thus, high-resolution ghost imaging results can be obtained under low sampling. Moreover, we have adopted a fast-converging iterative format to obtain better imaging results with fewer iterations. Simulation and experimental results show that the proposed method can achieve high-quality imaging results from the random speckle patterns under low sampling and promote the practical process of ghost imaging. [ABSTRACT FROM AUTHOR]

Sentence auto-completion is an important feature that saves users many keystrokes in typing the entire sentence by providing suggestions as they type. Despite its value, the existing sentence auto-completion methods, such as query completion models, can hardly be applied to solving the object completion problem in sentences with the form of (subject, verb, object), due to the complex natural language description and the data deficiency problem. Towards this goal, we treat an SVO sentence as a three-element triple (subject, sentence pattern, object), and cast the sentence object completion problem as an element inference problem. These elements in all triples are encoded into a unified low-dimensional embedding space by our proposed TRANSFER model, which leverages the external knowledge base to strengthen the representation learning performance. With such representations, we can provide reliable candidates for the desired missing element by a linear model. Extensive experiments on a real-world dataset have well-validated our model. Meanwhile, we have successfully applied our proposed model to factoid question answering systems for answer candidate selection, which further demonstrates the applicability of the TRANSFER model. [ABSTRACT FROM PUBLISHER]

In non-autoregressive machine translation, the target tokens are generated by the decoder in one shot. Although this decoding process can significantly reduce the decoding latency, non-autoregressive machine translation still suffers from the sacrifice of translation accuracy. We argue that the reason for such decrease is the lack of the target dependencies, history and future information, between target tokens. So, in this work, we propose a novel method to address this problem. We suppose the hidden representation of a target token from the decoder should consist of three parts: history, present, and future information. And we dynamically aggregate such parts-to-whole information with capsule network for the decoder to improve the performance of non-autoregressive machine translation. In addition, to ensure the capsules learn the information as we expect, we introduce an autoregressive decoder. Several experiments on benchmark tasks demonstrate that the explicit modeling of history and future information can significantly improve performance of NAT model. Extensive analyses show that our model is able to learn history and future information as we expect. • In order to solve the problem that the NAT model cannot model target dependency, we adapt the capsule network to assist the model to learn future and history information. • To ensure the capsule network learn as we expect, we design an autoregressive model to extract information. • We conduct a series of experiments on the benchmark tasks, and the results shows that our model can improve the performance. [ABSTRACT FROM AUTHOR]

Abstract: A novel nunchaku-like organic second order nonlinear optical chromophore containing a D–π–A backbone and a rigid biphenylene unit was prepared. The UV–vis absorption spectra studies demonstrated that the large rigid structure could suppress dipole–dipole interactions effectively. The electro-optic coefficient of a poled film containing 20% wt of the new dye doped in amorphous polycarbonate afforded a value of 39 pm/V at 1310 nm. Compared with results obtained from a similar D–π–A chromophore without the rigid isolated group, the new dye gave better performance. Moreover, the electro-optic film prepared with the new dye gave greater temporal stability, compared to a similar D–π–A doped poled film that was reported previously. These results indicated that the introduction of a biphenylene group could effectively reduce the intermolecular electrostatic interactions and aid the alignment of the chromophores. [Copyright &y& Elsevier]

Two series of novel electro-optic (EO) polycarbonates incorporating two kinds of tricyanopyrroline-based nonlinear optical (NLO) chromophores were designed and synthesized. These new polycarbonates were prepared through the facile copolymerization of diol NLO chromophores and bisphenol A bis(chloroformate), and the successful preparations were demonstrated by 1H NMR and Fourier transform infrared (FT-IR) spectra. These polycarbonates possessed good thermal stabilities and also showed higher glass transition temperatures (T g) in the range of 156–165 °C. After corona poling, the EO coefficients (r 33) of two poled polycarbonates films were up to 52 pm/V (PC-DTCPC-Ph-2) and 46 pm/V (PC-DTCPC-FPh-2) at the wavelength of 1310 nm. The higher T gs endow the polycarbonates' poled films with good temporal stability of poling-induced dipole alignment, and the resulting poled films of PC-DTCPC-Ph-2 and PC-DTCPC-FPh-2 could retain 95% and 93% of the initial EO activities at 85 °C for more than 500 h respectively. [ABSTRACT FROM AUTHOR]

Joint entity and relation extraction is an important task in natural language processing and knowledge graph construction. Existing studies mainly focus on three issues: redundant predictions, overlapping triples and relation connections. However, as far as we know, none of them is able to solve the three problems simultaneously in a unified architecture. To address this issue, in this paper, we propose a novel translation based unified framework. Specifically, the proposed framework contains two components: an entity tagger and a relation extractor. The former is used to recognize all candidate head entities and tail entities respectively. The latter predicts relations for every entity pair dynamically through ranking with translation mechanism. To show the superiority of the proposed framework, we instantiate it through the simplest binary entity tagger and TransE algorithm. Extensive experiments over two widely used datasets demonstrate that, even with the simplest components, the proposed framework can still achieve competitive performance with most previous baselines. Moreover, the framework is flexible. It enjoys further performance boost when employing more powerful entity tagger and knowledge graph embedding algorithm. [ABSTRACT FROM AUTHOR]

Traditional fusion algorithm is prone to skipping the multimodal attribute of the magnetic resonance image (MRI), lack of consideration of the spatial information of electroencephalogram (EEG) signal, resulting in the ineffective fusion of magnetic resonance image and EEG signal. Thus, constraint fusion algorithm of the multimodal magnetic resonance image and EEG signal using deep learning is proposed. First, the magnetic resonance image and EEG signal are simultaneously acquired for preprocessing. Secondly, the three modal magnetic resonance image features of functional Magnetic Resonance Imaging(fMRI), structural Magnetic Resonance Imaging(sMRI) and Diffusion Tensor Imaging are extracted, respectively, to analyze different modal images by multi-index features fusion. Then, different neural network channels use different coding methods, focusing on the construction of local channel convolutional neural network(CNN) model based on Cross Stage Partial Network transformation, coding network and recognition network design. Finally, the inverse problem of EEG signal is solved by the twomey standardized algorithm, completing the fusion algorithm design. The results show that the CNN model of local channel renders good training effect, the signal preprocessing of the proposed algorithm has obvious effect, the extraction accuracy of magnetic resonance image features of different modals is high, the MI value of image fusion is as high as 0.92, and the image fusion precision is high, and the time consumption is much lower than other algorithms. [ABSTRACT FROM AUTHOR]

Diabetic ulcers represent a chronic condition characterized by prolonged hyperglycemia and delayed wound healing, accompanied by endocrine disorders, inflammatory responses, and microvascular damage in the epidermal tissue, demanding effective clinical treatment approaches. For thousands of years, ancient Chinese ethnopharmacological studies have documented the use of Poria cocos (Schw.) Wolf in treating diabetic ulcers. Recent research has substantiated the diverse pharmacological effects of Poria cocos (Schw.) Wolf, including its potential to alleviate hyperglycemia and exhibit anti-inflammatory, antioxidant, and immune regulatory properties, which could effectively mitigate diabetic ulcer symptoms. Furthermore, being a natural medicine, Poria cocos (Schw.) Wolf has demonstrated promising therapeutic effects and safety in the management of diabetic ulcers, holding significant clinical value. Despite its potential clinical efficacy and applications in diabetic ulcer treatment, the primary active components and underlying pharmacological mechanisms of Poria cocos (Schw.) Wolf remains unclear. Further investigations are imperative to establish a solid foundation for drug development in this domain. In this study, we aimed to identify the active compounds and potential targets of Poria cocos (Schw.) Wolf using UHPLC-Q-TOF-MS and TCMSP databases. Additionally, we attempt to identify targets related to diabetic ulcers. Following enrichment analysis, a network of protein-protein interactions was constructed to identify hub genes based on the common elements between the two datasets. To gain insights into the binding activities of the hub genes and active ingredients, molecular docking analysis was employed. Furthermore, to further validate the therapeutic effect of Poria cocos (Schw.) Wolf, we exerted in vitro experiments using human umbilical vein vascular endothelial cells and human myeloid leukemia monocytes (THP-1). The active ingredient of Poria cocos (Schw.) Wolf was applied in these experiments. Our investigations included various assays, such as CCK-8, scratch test, immunofluorescence, western blotting, RT-PCR, and flow cytometry, to explore the potential of Poria cocos (Schw.) Wolf triterpenoid extract (PTE) in treating diabetic ulcers. The findings here highlighted PTE as the primary active ingredient in Poria cocos (Schw.) Wolf. Utilizing network pharmacology, we identified 74 potential targets associated with diabetic ulcer treatment for Poria cocos (Schw.) Wolf, with five hub genes (JUN, MAPK1, STAT3, AKT1, and CTNNB1). Enrichment analysis revealed the involvement of multiple pathways in the therapeutic process, with the PI3K-AKT signaling pathway showing significant enrichment. Through molecular docking, we discovered that relevant targets within this pathway exhibited strong binding with the active components of Poria cocos (Schw.) Wolf. In vitro experiments unveiled that PTE (10 mg/L) facilitated the migration of human umbilical vein vascular endothelial cells (P < 0.05). PTE also increased the expression of CD31 and VEGF mRNA (P < 0.05) while activating the expressions of p-PI3K and p-AKT (P < 0.05). Moreover, PTE demonstrated its potential by reducing the expression of IL-1β, IL-6, TNF-α, and NF-κB mRNA in THP-1 (P < 0.05) and fostering M2 macrophage polarization. These results signify the potential therapeutic effects of PTE in treating diabetic ulcers, with its beneficial actions mediated through the PI3K-AKT signaling pathway. PTE is the main active ingredient in Poria cocos (Schw.) Wolf that exerts therapeutic effects. Through PI3K-AKT signaling pathway activation and inflammatory response reduction, PTE promotes angiogenesis, thereby healing diabetic ulcers. [ABSTRACT FROM AUTHOR]

The article focuses on exploring dense retrieval for dialogue response selection, addressing challenges in ranking response candidates efficiently and improving response quality. Topics include the development of an efficient response selection model with dense retrieval and interaction layers, the use of learning strategies for effective training, and the release of high-quality benchmarks for accurate evaluation of dialogue response selection models.

• An unsupervised tree-to-sequence neural machine translation method is proposed. • Learning latent trees for improving neural machine translation. • Comparisons between Gumbel Tree-based model and the baselines, RNMT and Transformer. • Analysis on the learned tree structures and attention scores on the structures. Neural machine translation has improved the translation accuracy greatly and received great attention of the machine translation community. Tree-based translation models aim to model the syntactic or semantic relation among long-distance words or phrases in a sentence. However, it faces the difficulties of expensive manual annotation cost and poor automatic annotation accuracy. In this paper, we focus on how to encode a source sentence into a vector in a unsupervised-tree way and then decode it into a target sentence. Our model incorporates Gumbel Tree-LSTM, which can learn how to compose tree structures from plain text without any tree annotation. We evaluate the proposed model on both spoken and news corpora, and show that the performance of our proposed model outperforms the attentional seq2seq model and the Transformer base model. [ABSTRACT FROM AUTHOR]

New energy sources such as solar energy have been developed and utilized due to the shortage of non-renewable energy, and photoelectric materials have attracted more attention, accordingly. Moreover, flexible optoelectronic devices are crucial in the fields of wearable artificial intelligence, portable energy, and smart medical care. At present, photoelectric devices based on traditional perovskite cobalt oxide rigid films have problems such as inability to bend and deformation, which limits their application environment. Here, natural mica with atomically flat surface, high thermal stability, high transparency and good mechanical flexibility was selected as the substrate for preparing a series of LaCoO 3 flexible photoelectric films. These films exhibit excellent photoelectric properties and flexible properties. More importantly, the flexible film still has fast response speed and reproducible light response after 500 bending cycles. It reveals that all-inorganic perovskite LaCoO 3 is a promising candidate for designing high-performance, flexible and wearable optoelectronic devices. [Display omitted] [ABSTRACT FROM AUTHOR]

Abstract: Novel organic second-order nonlinear optical (NLO) chromophore with a large flexible rigid isolated group was prepared. The NLO properties were measured in solution and PMMA polymer films. The first order hyperpolarizabilities were equal to 4×10−28 esu and 7×10−29 esu at 1064nm Nd:YAG laser wavelength in solvent and polymer, respectively and these results are in agreement with quantum chemical calculations. Moreover, this chromophore has high thermal stability when thermal decomposition temperature is 270°C. Electrostatic interactions of the chromophores were effectively reduced by isolated group. No apparent change of the shape of UV absorption spectra versus concentration was found. [Copyright &y& Elsevier]

Online rumors could have a great impact on public order, stock prices and even the presidential election. Therefore, the detection of online rumors is imperative. Despite the satisfactory performance achieved by the current methods, there are still some issues that need to be addressed. First, most of the current methods have not taken into account imposing attentional constraints on important related words in the sentences, resulting in inaccurate attention being paid to some irrelevant words. Second, most of the current methods for detecting rumors fail to effectively incorporate contextual information from words or sentences. In this paper, we propose a syntactic multi-level interaction network model which incorporates syntactic dependency relationships and multi-level interaction network for rumor detection. First, the SMNet model uses a syntactic dependency parser to extract the corresponding syntactic sentence structures and incorporates the extracted syntactic dependency relationships into the attention mechanism for language-driven word representation. Then, the multi-level interaction network is applied to obtain a richer semantic representation. After that, the global relation encoding capture the rich structural information and the rumor classification is performed to generate the verification result. We have conducted experiments on Weibo, Twitter15 and Twitter16 datasets for performance evaluation. Our SMNet model has achieved an accuracy of 95.9% on the Weibo dataset. In addition, our SMNet model has achieved an accuracy of 91.7% and 93.5% on Twitter 15 and Twitter 16, respectively. The experimental results show that our proposed SMNet model outperforms the baseline models and achieves the state-of-the-art performance for rumor detection. [ABSTRACT FROM AUTHOR]

Multi-resolution imaging is one of the key means to obtain the target scene information. The Hadamard matrix, which is orthogonal, is an important modulation matrix for single-pixel imaging. In particular, it can provide a good means for multi-resolution imaging. However, as far as we know, studies of high-efficiency multi-resolution single-pixel imaging are rare in the literature. In view of the practical application requirements of fast multi-resolution imaging, we propose a multi-resolution single-pixel imaging method based on Hadamard ‘pipeline’ coding, which can directly generate two-dimensional Hadamard basis patterns and multi-resolution Hadamard optimization sequences, whereby both the memory consumption and the complexity of coding implementation for multi-resolution imaging can be significantly reduced. The commonly used optimization method of Hadamard optimization sequence implementation and time consumption are also discussed. This method provides a new approach for Hadamard sequence optimization and multi-resolution single-pixel imaging applications. [ABSTRACT FROM AUTHOR]

Summary: Lithium‐sulfur (Li‐S) battery has been evoked increasing attention due to its creditable energy density and the abundant sulfur in nature. Nevertheless, its commercialization is still hampered in virtue of the poor conductivity of sulfur and grievous polysulfide shuttling. To address these issues, in this work, we report a cost‐efficient and facile tactic to fabricate frame‐structured neodymium‐doped bismuth vanadate (BiVO4) nanoarchitectures, which are reasonably designed as a polysulfide shield to alleviate the shuttling effects. The synthesized nanoarchitectures exhibit typical tetragonal phase with a unique frame structure. The frame structure of BiVO4 impedes polysulfide shuttling, and supplies electronic and ionic transmission routes. Furthermore, the rational doping of neodymium within the BiVO4 nanoarchitectures can improve electric conductivity, as well as efficiently alleviate the loss of sulfur and improve activity for sulfur reduction. Benefitting from the structural feature and doping strategy, the Li‐S battery performance with a neodymium‐doped BiVO4 nanoarchitecture is significantly improved. This strategy is easily adopted for fabricating other nanostructures, providing a feasible way to design cheap and efficient battery materials. This work is further available for synergistically uniting the virtues of the frame nanoarchitectures and doping protocol to develop advanced Li‐S batteries. [ABSTRACT FROM AUTHOR]

Summarization systems for various applications, such as opinion mining, online news services, and answering questions, have attracted increasing attention in recent years. These tasks are complicated, and a classic representation using bag-of-words does not adequately meet the comprehensive needs of applications that rely on sentence extraction. In this paper, we focus on representing sentences as continuous vectors as a basis for measuring relevance between user needs and candidate sentences in source documents. Embedding models based on distributed vector representations are often used in the summarization community because, through cosine similarity, they simplify sentence relevance when comparing two sentences or a sentence/query and a document. However, the vector-based embedding models do not typically account for the salience of a sentence, and this is a very necessary part of document summarization. To incorporate sentence salience, we developed a model, called CCTSenEmb, that learns latent discriminative Gaussian topics in the embedding space and extended the new framework by seamlessly incorporating both topic and sentence embedding into one summarization system. To facilitate the semantic coherence between sentences in the framework of prediction-based tasks for sentence embedding, the CCTSenEmb further considers the associations between neighboring sentences. As a result, this novel sentence embedding framework combines sentence representations, word-based content, and topic assignments to predict the representation of the next sentence. A series of experiments with the DUC datasets validate CCTSenEmb's efficacy in document summarization in a query-focused extraction-based setting and an unsupervised ILP-based setting. [ABSTRACT FROM AUTHOR]

Nd:BiVO 4 nanocrystals were synthesized by an effective and simple approach. Nd was incorporated into BiVO 4 host to enhance electronic conductivity and lithium-ion diffusion kinetics, thus promoting the electrochemical performance. When employed as anode materials in lithium-ion batteries, a good cycling stability and high capacity of ~611 mAh g−1 at 100 mA g−1 were delivered. The work can be extensively employed for fabricating other electrode materials and promoting the electrochemical performance in energy storage correlative domains. [ABSTRACT FROM AUTHOR]

Novel nonlinear optical (NLO) chromophore using benzo[b]furan moiety and trifluoromethyl-substituted tricyanofuran as the electron donor and acceptor respectively was designed and prepared. The structure of the novel chromophore was determined by NMR spectra, and the photophysical properties were studied via UV-vis spectra. The thermal decomposition temperature (T d)value of the obtained NLO chromophore reached to 212 °C, which endow the chromophore with good thermal stability during the materials processing. The nonlinearity of the chromophore was evaluated through theoretical calculation and electro-optic (EO) activity test, and the highest EO coefficient (r 33) was determined to 28 pm/V at the loading density of 25 wt% in polymethyl methacrylate. • Novel benzofuran-based nonlinear optical chromophore was designed and synthesized. • The acceptor was unexpected linked to the β -position of furan in the chromophore. • The present chromophore exhibited go0d thermal stability (T d reached to 212 °C). • The maximum EO coefficient of the reported chromophore reached to 28 p.m./V. [ABSTRACT FROM AUTHOR]

The modification and tuning features of nanostructured films are of great interest because of controllable and distinctive inherent properties in these materials. Here, nanocrystalline MoS 2 films were fabricated on the stainless steels by a radio frequency magnetron sputtering at ambient temperature. X-ray photoelectron spectroscopy, energy-dispersive X-ray spectroscopy, X-ray diffraction and Raman scattering spectroscopy were used to study the chemical state, chemical composition, crystal structure and vibrational properties of the fabricated MoS 2 films. The bias voltage dependent structural evolution and its influence on the optical properties of MoS 2 nanocrystalline films were systematically investigated. Besides, the residual stresses of MoS 2 nanocrystalline films were explored by employing sin2 ψ approach. X-ray diffraction demonstrates that the nanocrystalline MoS 2 films have single-phase hexagonal crystal structure. All MoS 2 films are polycrystalline in nature. The bandgap values are found to be intensively dependent on bias voltage. Our findings show that the nanocrystalline MoS 2 films with different physical properties and intense quantum confinement effect can be realized through adjusting bias voltages. This work may provide deep insight for realizing transitional metal dichalcogenide-based nanostructured film optoelectronic devices with tunable physical properties through a traditional, very cost-effective, and large-scale fabrication method. [ABSTRACT FROM AUTHOR]

Abstract Considerable attention on semiconducting nanocrystallines have been paid owing to their fascinating electrical and optical features, which can be utilized as luminophor and light absorber. Herein, we report a facile and ligand-free protocol to synthesize high-quality CdS nanocrystals on titanium plate. The crystal structural, micro-structural, local structural, surface chemistry and optical properties of CdS nanocrystals have been investigated. pH value dependent optical features of CdS nanocrystals are explored by photoluminescence and UV–vis diffuse reflectance spectroscopy. XRD results manifest that these products possess polycrystalline characteristics, cubic zinc-blende structure and nanomaterial features. The crystal defects and microstructure of these nanocrystals are researched by analysis of XRD line profiles. TEM images display nearly spherical nanocrystals with mean size about 4 nm, well crystalline feature and interplanar spacings of cubic zinc-blende CdS nanocrystals. Raman results also reveal the cubic sphalerite structure of CdS nanocrystals. The first three Raman peaks are assigned to the first, second, and third-order longitudinal optical phonon modes of cubic CdS, respectively. Interestingly, optical absorption spectra display that the absorbance shoulder shows a noticeable change with pH value of the precursor. The observations show that the nanocrystalline size is smaller at lower pH value and it changes as the pH value of the precursor alters, and CdS nanocrystals exhibit significant quantum confinement. More importantly, Photoluminescence spectra show that the first emission peaks can be attributed to band-edge luminescence, which is originated from the recombination of shallow trapped hole-electron pairs and/or excitons in cadmium sulfide nanomaterials, while the second emission peaks can be corresponded to trap-state luminescence, and PL intensity changes with the pH value of the precursor. The insights offered in this work may pave the way to devising of improved synthetic schemes for semiconductor nanocrystals with tailored microstructural properties, permitting precise adjusting of the optoelectronic performances of the prepared materials. Highlights • CdS nanocrystals were fabricated on titanium plate by a simple and ligand-free approach. • Crystal structure, local structure, surface chemistry and optical properties of the nanocrystals. • The insights may pave the way to devising of improved synthetic schemes for nanocrystals. [ABSTRACT FROM AUTHOR]

Herein, we demonstrated a well-rounded exploration of the universal structures, microstructures and physical properties of different rice grains. We show that these rice grains are nanocrystalline in nature, have the standard pattern of type A rice crystalline, and can be attributed to the hexagonal crystal structure with space group of P6. These rice grains have uneven surfaces, clusters of the granules, or schistose structure. Carbon, nitrogen and oxygen elements are presented in the rice grains. No significant difference is observed in FTIR spectra in these rice grains, indicating the fundamental chemical structures of these rice grains are analogical. Their microstructural and physical properties were investigated in detail. Moreover, chemical properties (e.g. thermal stability) of the rice grains were explored. This research provides an in-depth understanding on the physical and structural properties on the atomic and molecular level, as well as guidance for food and industry applications. [ABSTRACT FROM AUTHOR]

This paper proposes an extension of the non-local Bayesian denoising algorithm. The idea is to use elliptical patches instead of regular square patches in the grouping process. We calculate the elliptical patches by an iterative method. We then use an affine invariant patch similarity measure to calculate the distance between two elliptical patches. Since the elliptical patch is a shape-adaptive patch and this similarity measure performs a patch comparison by automatically adapting the size and shape of the patches, so more similar patches are found and used for image denoising. This algorithm denoising procedure goes through two identical iterations to further improve the denoising performance. Experimental results on test images demonstrate that this algorithm achieves state-of-the-art denoising performance in terms of numerical results and subjective visual quality, compared with the non-local Bayesian. [ABSTRACT FROM AUTHOR]