Few-shot named entity recognition (NER) aims to recognize novel named entities in low-resource domains utilizing existing knowledge. However, the present few-shot NER models assume that the labeled data are all clean without noise or outliers, and there are few works focusing on the robustness of the cross-domain transfer learning ability to textual adversarial attacks in Few-shot NER. In this work, we comprehensively explore and assess the robustness of few-shot NER models under textual adversarial attack scenario, and found the vulnerability of existing few-shot NER models. Furthermore, we propose a robust two-stage few-shot NER method with Boundary Discrimination and Correlation Purification (BDCP). Specifically, in the span detection stage, the entity boundary discriminative module is introduced to provide a highly distinguishing boundary representation space to detect entity spans. In the entity typing stage, the correlations between entities and contexts are purified by minimizing the interference information and facilitating correlation generalization to alleviate the perturbations caused by textual adversarial attacks. In addition, we construct adversarial examples for few-shot NER based on public datasets Few-NERD and Cross-Dataset. Comprehensive evaluations on those two groups of few-shot NER datasets containing adversarial examples demonstrate the robustness and superiority of the proposed method.

Graph self-supervised learning provides a powerful guarantee for learning high-quality representations in an unsupervised manner. Despite its early birth, the performance of generative graph self-supervised learning has long lagged behind that of up-and-coming contrastive learning, especially on node classification tasks. In this paper, we investigate potential issues in existing graph autoencoders and attribute their poor performance to three main aspects: complex decoder design, lack of desmoothing process in feature remap, and overemphasis on local topological proximity. To tackle these issues, we propose an effective and efficient graph autoencoder framework for unsupervised representation learning, which contains two key components: lightweight smoothness-aware feature reconstructor and global structural dependency catcher. After performing a desmoothing operation on encoded representations via a learnable high-pass filter, the feature decoder reconstructs the original features through a simple linear projection. The lightweight design liberates the decoder from self-supervised pretext tasks and puts the encoder more accountable for achieving optimization objectives, which promotes effective training of the encoder. Global structural dependency catcher utilizes graph diffusion to build a structural regularization to capture long-range topological dependency on a graph. The empirical studies demonstrate the effectiveness of our approach, which can surpass dominant contrastive learning methods.

Tuples extraction is a fundamental task for information extraction and knowledge graph construction.The extracted tuples are usually represented as knowledge triples consisting of subject, relation, and object.In practice, however, the validity of knowledge triples is associated with and changes with the spatial, temporal, or other kinds of constraints.Motivated by this observation, this paper proposes a constrained tuple extraction (CTE) task to guarantee the validity of knowledge tuples.Formally, the CTE task is to extract constrained tuples from unstructured text, which adds constraints to conventional triples.To this end, we propose an interaction-aware network.Combinatorial interactions among context-specific external features and distinct-granularity internal features are exploited to effectively mine the potential constraints.Moreover, we have built a new dataset containing totally 1,748,826 constrained tuples for training and 3656 ones for evaluation.Experiments on our dataset and the public CaRB dataset demonstrate the superiority of the proposed model.The constructed dataset and the codes are publicly available.

The abundant features and structure information on graphs provide a potential guarantee for learning high-quality representations without supervision. Feature attribute represents the inherent properties of nodes, while structure attribute describes their neighborhood relationship. These two types of attributes can be regarded as different modal forms of the same instance and should be consistent in identifying a member. We propose to directly regard feature and structure attributes as two separate views to embed this consistency into contrastive learning method, realizing graph information interaction on feature and structure in a cross-modal contrastive framework. Under this framework, node representations are learned in an unsupervised manner by maximizing the agreement between feature representation and structure representation. In terms of negative samples, instead of randomly sampling points from empirical distribution, a simple yet effective multi-sample mixing strategy is proposed to synthesize true negative samples with greater probability, alleviating the tricky false negative issue. Extensive experiments on multiple types of graphs demonstrate the effectiveness of the proposed method.

Current evidence indicates that the semantic representation of question and answer sentences is better generated by deep neural network-based sentence models than traditional methods in community answer selection tasks. In particular, as a widely recognized language model, the self-attention model computes the similarity between the specific word and the whole sets of words in the same sentence and generates new semantic representation through the similarity-weighted summation of semantic representations of the whole words. However, the self-attention operation entirely considers all the signals with a weighted sum operation, which disperses the distribution of attention, which may result in overlooking the relation of neighboring signals. This issue becomes serious when applying the self-attention model to online community question answering platforms because of the varied length of the user-generated questions and answers. To address this problem, we introduce an attention mechanism enhanced local self-attention (LSA), which restricts the range of original self-attention by a local window mechanism, thereby scaling linearly when increasing the sequence length. Furthermore, we propose stacking multiple LSA layers to model the relationship of multiscale$n$-gram features. It captures the word-to-word relationship in the first layer and then captures the chunk-to-chunk (such as lexical$n$-gram phrases) relationship in its deeper layers. We also test the effectiveness of the proposed model by applying the learned representation through the LSA model to a Siamese and a classification network in community question answer selection tasks. Experiments on the public datasets show that the proposed LSA achieves a good performance.

Recently the research of traffic flow prediction with deep learning framework has be largely developed, whereas most current methods are still faced with the following shortcomings. For spatial feature extraction, studies have shown that both local and non-local correlations exist on traffic networks. Considering the temporal dependencies, short-term impending and longer periodic components are two most critical patterns of traffic data, which further provide different information for the prediction task. Furthermore, multi-source heterogeneous external data, which naturally holds semantic gap with traffic data, also have impact on traffic flow. To solve the above problems, this paper proposes an AutoMSNet (Multi-Source Spatio-Temporal Network via Automatic neural architecture search). The AutoMSNet is composed of an encoder-decoder structure. The encoder takes neighboring data as inputs, while the decoder captures long-term periodic patterns. Thus, different functions of two temporal features are simultaneously extracted. Moreover, a neural architecture search space is designed for spatial feature extraction. Through architecture search technique, graph convolutions with different receptive fields are automatically selected and combined to form an optimal module structure. Therefore, both local and non-local spatial features can be adaptively captured. Besides, a meta learning feature fusion strategy is proposed to integrate external data, which can alleviate the semantic gap between different data sources. Extensive experiments on three real-world traffic datasets evaluate the superiority of the proposed model.

Multimodal sentiment analysis (MSA) is a very challenging task due to its complex and complementary interactions between multiple modalities, which can be widely applied into areas of product marketing, public opinion monitoring, and so on. However, previous works directly utilized the features extracted from multimodal data, in which the noise reduction within and among multiple modalities has been largely ignored before multimodal fusion. This paper proposes a multi-level attention map network (MAMN) to filter noise before multimodal fusion and capture the consistent and heterogeneous correlations among multi-granularity features for multimodal sentiment analysis. Architecturally, MAMN is comprised of three modules: multi-granularity feature extraction module, multi-level attention map generation module, and attention map fusion module. The first module is designed to sufficiently extract multi-granularity features from multimodal data. The second module is constructed to filter noise and enhance the representation ability for multi-granularity features before multimodal fusion. And the third module is built to extensibly mine the interactions among multi-level attention maps by the proposed extensible co-attention fusion method. Extensive experimental results on three public datasets show the proposed model is significantly superior to the state-of-the-art methods, and demonstrate its effectiveness on two tasks of document-based and aspect-based MSA tasks.

Graph neural networks (GNNs) have been widely used to accomplish graph classification tasks such as predicting molecular properties and classifying the labels of proteins. Discovering the latent discriminative substructures (e.g., functional groups in molecules) is a vital task to enhance the classification performance. In this paper, this task is addressed as a problem of discriminative graph representation learning. Specifically, a novel node sampling strategy is developed to achieve this goal. To this end, graph-dependent sampling vectors are first learned by a mini-network to exploit various informative substructures on graphs and sample some representative nodes, which could be regarded as performing a distributed sampling on graphs. Then, the sampled nodes are organized together topologically as a subgraph with landing probabilities of random walks. Moreover, a self-adaptive pooling ratio of nodes is obtained via feature smoothness of graphs, eliminating the trouble of manual selection of subgraph size. As a result, these treatments are equivalent to performing the difficult step of down-pooling operation on non-grid graph data. Extensive experiments and ablation studies on multiple benchmark datasets demonstrate the effectiveness and superiority of our proposed approach. Additionally, interpretability studies illustrate the ability of our model to extract discriminative substructures.

Predicting urban traffic flow is a challenging task, due to the complicated spatio-temporal dependencies on traffic networks. Urban traffic flow usually has both short-term neighboring and long-term periodic temporal dependencies. It is also noticed that the spatial correlations over different traffic nodes are both local and non-local. What’s more, the traffic flow is affected by various external factors. To capture the non-local spatial correlations, we propose a Dilated Attentional Graph Convolution (DAGC). The DAGC utilizes a dilated graph convolution kernel to expand the nodes’ receptive field and exploit multi-order neighborhood. Technically, the lower-order neighborhood corresponds to local spatial dependencies, while the higher-order neighborhood corresponds to non-local spatial dependencies between nodes. Based on DAGC, a Multi-Source Spatio-Temporal Network (MS-Net) is designed, which suffices to integrate long-range historical traffic data as well as multi-modal external information. MS-Net consists of four components: a spatial feature extraction module, a temporal feature fusion module, an external factors embedding module, and a multi-source data fusion module. Extensive experiments on three real traffic datasets demonstrates that the proposed model performs well on both the public transportation networks, road networks, and can handle large-scale traffic networks in particular the Beijing bus network which has more than 4,000 traffic nodes.

The task of visual question answering (VQA) is to generate an answer for a question according to the content of an image being asked. In this process, the critical problems of effectively embedding the question feature and image feature as well as transforming the features to the prediction of answer are still faithfully unresolved. In this paper, depending on these problems, a semantic-associated attention method and a reinforcement stacked learning mechanism are proposed. Firstly, within the associations of high-level semantics, a visual spatial attention model (VSA) and a multi-semantic attention model (MSA) are proposed to extract the low-level image feature and high-level semantic feature, respectively. Furthermore, we develop a reinforcement stacked learning architecture, which splits the transformation process into multiple stages, to gradually approach the answers. At each stage, a new reinforcement learning (RL) method is introduced to directly criticize inappropriate answers to optimize the model. The extensive experiments on the VQA task show that our method can achieve state-of-the-art performance.

Opinion mining is a challenging task to identify the opinions or sentiments underlying user generated contents, such as online product reviews, blogs, discussion forums, etc. Previous studies that adopt machine learning algorithms mainly focus on designing effective features for this complex task. This paper presents our approach based on tree kernels for opinion mining of online product reviews. Tree kernels alleviate the complexity of feature selection and generate effective features to satisfy the special requirements in opinion mining. In this paper, we define several tree kernels for sentiment expression extraction and sentiment classification, which are subtasks of opinion mining. Our proposed tree kernels encode not only syntactic structure information, but also sentiment related information, such as sentiment boundary and sentiment polarity, which are important features to opinion mining. Experimental results on a benchmark data set indicate that tree kernels can significantly improve the performance of both sentiment expression extraction and sentiment classification. Besides, a linear combination of our proposed tree kernels and traditional feature vector kernel achieves the best performances using the benchmark data set.

This paper presents an interactive algorithm for segmentation of natural images. The task is formulated as a problem of spline regression, in which the spline is derived in Sobolev space and has a form of a combination of linear and Green's functions. Besides its nonlinear representation capability, one advantage of this spline in usage is that, once it has been constructed, no parameters need to be tuned to data. We define this spline on the user specified foreground and background pixels, and solve its parameters (the combination coefficients of functions) from a group of linear equations. To speed up spline construction, K-means clustering algorithm is employed to cluster the user specified pixels. By taking the cluster centers as representatives, this spline can be easily constructed. The foreground object is finally cut out from its background via spline interpolation. The computational complexity of the proposed algorithm is linear in the number of the pixels to be segmented. Experiments on diverse natural images, with comparison to existing algorithms, illustrate the validity of our method.

This paper presents a new algorithm for Nonlinear Dimensionality Reduction (NLDR). Our algorithm is developed under the conceptual framework of compatible mapping. Each such mapping is a compound of a tangent space projection and a group of splines. Tangent space projection is estimated at each data point on the manifold, through which the data point itself and its neighbors are represented in tangent space with local coordinates. Splines are then constructed to guarantee that each of the local coordinates can be mapped to its own single global coordinate with respect to the underlying manifold. Thus, the compatibility between local alignments is ensured. In such a work setting, we develop an optimization framework based on reconstruction error analysis, which can yield a global optimum. The proposed algorithm is also extended to embed out of samples via spline interpolation. Experiments on toy data sets and real-world data sets illustrate the validity of our method.