IJCAI 2021 Papers

Most sequential recommender systems (SRSs) predict next-item as target for each user given its preceding items as input, assuming that each input is related to its target. However, users may unintentionally click on items that are inconsistent with their preference. We empirically verify that SRSs can be misguided with such unreliable instances (i.e. targets mismatch inputs). This inspires us to design a novel SRS By Eliminating unReliable Data (BERD) guided with two observations: (1) unreliable instances generally have high training loss; and (2) high-loss instances are not necessarily unreliable but uncertain ones caused by blurry sequential pattern. Accordingly, BERD models both loss and uncertainty of each instance via a Gaussian distribution to better distinguish unreliable instances; meanwhile an uncertainty-aware graph convolution network is exploited to assist in mining unreliable instances by lowering uncertainty. Extensive experiments on four real-world datasets demonstrate the superiority of our proposed BERD.

Document-level relation extraction aims to extract relations among multiple entity pairs from a document. Previously proposed graph-based or transformer-based models utilize the entities independently, regardless of global information among relational triples. This paper approaches the problem by predicting an entity-level relation matrix to capture local and global information, parallel to the semantic segmentation task in computer vision. Herein, we propose a Document U-shaped Network for document-level relation extraction. Specifically, we leverage an encoder module to capture the context information of entities and a U-shaped segmentation module over the image-style feature map to capture global interdependency among triples. Experimental results show that our approach can obtain state-of-the-art performance on three benchmark datasets DocRED, CDR, and GDA.

In its most traditional setting, the main concern of optimization theory is the search for optimal solutions for instances of a given computational problem. A recent trend of research in artificial intelligence, called solution diversity, has focused on the development of notions of optimality that may be more appropriate in settings where subjectivity is essential. The idea is that instead of aiming at the development of algorithms that output a single optimal solution, the goal is to investigate algorithms that output a small set of sufficiently good solutions that are sufficiently diverse from one another. In this way, the user has the opportunity to choose the solution that is most appropriate to the context at hand. It also displays the richness of the solution space. When combined with techniques from parameterized complexity theory, the paradigm of diversity of solutions offers a powerful algorithmic framework to address problems of practical relevance. In this work, we investigate the impact of this combination in the field of Kemeny Rank Aggregation, a well-studied class of problems lying in the intersection of order theory and social choice theory and also in the field of order theory itself. In particular, we show that KRA is fixed-parameter tractable with respect to natural parameters providing natural formalizations of the notions of diversity and of the notion of a sufficiently good solution. Our main results work both when considering the traditional setting of aggregation over linearly ordered votes, and in the more general setting where votes are partially ordered.

Heuristic search methods are widely used in many real-world autonomous systems. Yet, people always want to solve search problems that are larger than time allows. To address these challenging problems, even suboptimally, a planning agent should be smart enough to intelligently allocate its computational resources, to think carefully about where in the state space it should spend time searching. For finding optimal solutions, we must examine every node that is not provably too expensive. In contrast, to find suboptimal solutions when under time pressure, we need to be very selective about which nodes to examine. In this work, we will demonstrate that estimates of uncertainty, represented as belief distributions, can be used to drive search effectively. This type of algorithmic approach is known as metareasoning, which refers to reasoning about which reasoning to do. We will provide examples of improved algorithms for real-time search, bounded-cost search, and situated planning.

The notion of distortion in social choice problems has been defined to measure the loss in efficiency---typically measured by the utilitarian social welfare, the sum of utilities of the participating agents---due to having access only to limited information about the preferences of the agents. We survey the most significant results of the literature on distortion from the past 15 years, and highlight important open problems and the most promising avenues of ongoing and future work.

In polymatrix coordination games, each player x is a node of a graph and must select an action in her strategy set. Nodes are playing separate bimatrix games with their neighbors in the graph. Namely, the utility of x is given by the preference she has for her action plus, for each neighbor y, a payoff which strictly depends on the mutual actions played by x and y. We propose the new class of distance polymatrix coordination games, properly generalizing polymatrix coordination games, in which the overall utility of player x further depends on the payoffs arising by mutual actions of players v,z that are the endpoints of edges at any distance h

Recent works have shown that a rich set of semantic directions exist in the latent space of Generative Adversarial Networks (GANs), which enables various facial attribute editing applications. However, existing methods may suffer poor attribute variation disentanglement, leading to unwanted change of other attributes when altering the desired one. The semantic directions used by existing methods are at attribute level, which are difficult to model complex attribute correlations, especially in the presence of attribute distribution bias in GAN's training set. In this paper, we propose a novel framework (IALS) that performs Instance-Aware Latent-Space Search to find semantic directions for disentangled attribute editing. The instance information is injected by leveraging the supervision from a set of attribute classifiers evaluated on the input images. We further propose a Disentanglement-Transformation (DT) metric to quantify the attribute transformation and disentanglement efficacy and find the optimal control factor between attribute-level and instance-specific directions based on it. Experimental results on both GAN-generated and real-world images collectively show that our method outperforms state-of-the-art methods proposed recently by a wide margin. Code is available at https://github.com/yxuhan/IALS.

The multiple kernel k-means (MKKM) and its variants utilize complementary information from different kernels, achieving better performance than kernel k-means (KKM). However, the optimization procedures of previous works all comprise two stages, learning the continuous relaxed label matrix and obtaining the discrete one by extra discretization procedures. Such a two-stage strategy gives rise to a mismatched problem and severe information loss. To address this problem, we elaborate a novel Discrete Multiple Kernel k-means (DMKKM) model solved by an optimization algorithm that directly obtains the cluster indicator matrix without subsequent discretization procedures. Moreover, DMKKM can strictly measure the correlations among kernels, which is capable of enhancing kernel fusion by reducing redundancy and improving diversity. What’s more, DMKKM is parameter-free avoiding intractable hyperparameter tuning, which makes it feasible in practical applications. Extensive experiments illustrated the effectiveness and superiority of the proposed model.

Being an indispensable component in location-based social networks, next point-of-interest (POI) recommendation recommends users unexplored POIs based on their recent visiting histories. However, existing work mainly models check-in data as isolated POI sequences, neglecting the crucial collaborative signals from cross-sequence check-in information. Furthermore, the sparse POI-POI transitions restrict the ability of a model to learn effective sequential patterns for recommendation. In this paper, we propose Sequence-to-Graph (Seq2Graph) augmentation for each POI sequence, allowing collaborative signals to be propagated from correlated POIs belonging to other sequences. We then devise a novel Sequence-to-Graph POI Recommender (SGRec), which jointly learns POI embeddings and infers a user's temporal preferences from the graph-augmented POI sequence. To overcome the sparsity of POI-level interactions, we further infuse category-awareness into SGRec with a multi-task learning scheme that captures the denser category-wise transitions. As such, SGRec makes full use of the collaborative signals for learning expressive POI representations, and also comprehensively uncovers multi-level sequential patterns for user preference modelling. Extensive experiments on two real-world datasets demonstrate the superiority of SGRec against state-of-the-art methods in next POI recommendation.

Learning to order events at discourse-level is a crucial text understanding task. Despite many efforts for this task, the current state-of-the-art methods rely heavily on manually designed features, which are costly to produce and are often specific to tasks/domains/datasets. In this paper, we propose a new graph perspective on the task, which does not require complex feature engineering but can assimilate global features and learn inter-dependencies effectively. Specifically, in our approach, each document is considered as a temporal graph, in which the nodes and edges represent events and event-event relations respectively. In this sense, the temporal ordering task corresponds to constructing edges for an empty graph. To train our model, we design a graph mask pre-training mechanism, which can learn inter-dependencies of temporal relations by learning to recover a masked edge following graph topology. In the testing stage, we design an certain-first strategy based on model uncertainty, which can decide the prediction orders and reduce the risk of error propagation. The experimental results demonstrate that our approach outperforms previous methods consistently and can meanwhile maintain good global consistency.

We present a novel direction-aware feature-level frequency decomposition network for single image deraining. Compared with existing solutions, the proposed network has three compelling characteristics. First, unlike previous algorithms, we propose to perform frequency decomposition at feature-level instead of image-level, allowing both low-frequency maps containing structures and high-frequency maps containing details to be continuously refined during the training procedure. Second, we further establish communication channels between low-frequency maps and high-frequency maps to interactively capture structures from high-frequency maps and add them back to low-frequency maps and, simultaneously, extract details from low-frequency maps and send them back to high-frequency maps, thereby removing rain streaks while preserving more delicate features in the input image. Third, different from existing algorithms using convolutional filters consistent in all directions, we propose a direction-aware filter to capture the direction of rain streaks in order to more effectively and thoroughly purge the input images of rain streaks. We extensively evaluate the proposed approach in three representative datasets and experimental results corroborate our approach consistently outperforms state-of-the-art deraining algorithms.

Traditional crowd counting approaches usually use Gaussian assumption to generate pseudo density ground truth, which suffers from problems like inaccurate estimation of the Gaussian kernel sizes. In this paper, we propose a new measure-based counting approach to regress the predicted density maps to the scattered point-annotated ground truth directly. First, crowd counting is formulated as a measure matching problem. Second, we derive a semi-balanced form of Sinkhorn divergence, based on which a Sinkhorn counting loss is designed for measure matching. Third, we propose a self-supervised mechanism by devising a Sinkhorn scale consistency loss to resist scale changes. Finally, an efficient optimization method is provided to minimize the overall loss function. Extensive experiments on four challenging crowd counting datasets namely ShanghaiTech, UCF-QNRF, JHU++ and NWPU have validated the proposed method.

Multi-modal cues presented in videos are usually beneficial for the challenging video-text retrieval task on internet-scale datasets. Recent video retrieval methods take advantage of multi-modal cues by aggregating them to holistic high-level semantics for matching with text representations in a global view. In contrast to this global alignment, the local alignment of detailed semantics encoded within both multi-modal cues and distinct phrases is still not well conducted. Thus, in this paper, we leverage the hierarchical video-text alignment to fully explore the detailed diverse characteristics in multi-modal cues for fine-grained alignment with local semantics from phrases, as well as to capture a high-level semantic correspondence. Specifically, multi-step attention is learned for progressively comprehensive local alignment and a holistic transformer is utilized to summarize multi-modal cues for global alignment. With hierarchical alignment, our model outperforms state-of-the-art methods on three public video retrieval datasets.

Instead of learning with pointwise loss functions, learning with pairwise loss functions (pairwise learning) has received much attention recently as it is more capable of modeling the relative relationship between pairs of samples. However, most of the existing algorithms for pairwise learning fail to take into consideration the privacy issue in their design. To address this issue, previous work studied pairwise learning in the Differential Privacy (DP) model. However, their utilities (population errors) are far from optimal. To address the sub-optimal utility issue, in this paper, we proposed new pure or approximate DP algorithms for pairwise learning. Specifically, under the assumption that the loss functions are Lipschitz, our algorithms could achieve the optimal expected population risk for both strongly convex and general convex cases. We also conduct extensive experiments on real-world datasets to evaluate the proposed algorithms, experimental results support our theoretical analysis and show the priority of our algorithms.

Domain adaptation solves a learning problem in a target domain by utilizing the training data in a different but related source domain. As a simple and efficient method for domain adaptation, correlation alignment transforms the distribution of the source domain by utilizing the covariance matrix of the target domain, such that a model trained on the transformed source data can be applied to the target data. However, when source and target domains come from different institutes, exchanging information between the two domains might pose a potential privacy risk. In this paper, for the first time, we propose a differentially private correlation alignment approach for domain adaptation called PRIMA, which can provide privacy guarantees for both the source and target data. In PRIMA, to relieve the performance degradation caused by perturbing the covariance matrix in high dimensional setting, we present a random subspace ensemble based covariance estimation method which splits the feature spaces of source and target data into several low dimensional subspaces. Moreover, since perturbing the covariance matrix may destroy its positive semi-definiteness, we develop a shrinking based method for the recovery of positive semi-definiteness of the covariance matrix. Experimental results on standard benchmark datasets confirm the effectiveness of our approach.

Despite the valuable information contained in software chat messages, disentangling them into distinct conversations is an essential prerequisite for any in-depth analyses that utilize this information. To provide a better understanding of the current state-of-the-art, we evaluate five popular dialog disentanglement approaches on software-related chat. We find that existing approaches do not perform well on disentangling software-related dialogs that discuss technical and complex topics. Further investigation on how well the existing disentanglement measures reflect human satisfaction shows that existing measures cannot correctly indicate human satisfaction on disentanglement results. Therefore, in this paper, we introduce and evaluate a novel measure, named DLD. Using results of human satisfaction, we further summarize four most frequently appeared bad disentanglement cases on software-related chat to insight future improvements. These cases include (i) Ignoring Interaction Patterns, (ii) Ignoring Contextual Information, (iii) Mixing up Topics, and (iv) Ignoring User Relationships. We believe that our findings provide valuable insights on the effectiveness of existing dialog disentanglement approaches and these findings would promote a better application of dialog disentanglement in software engineering.

Meeting summarization is a challenging task due to its dynamic interaction nature among multiple speakers and lack of sufficient training data. Existing methods view the meeting as a linear sequence of utterances while ignoring the diverse relations between each utterance. Besides, the limited labeled data further hinders the ability of data-hungry neural models. In this paper, we try to mitigate the above challenges by introducing dialogue-discourse relations. First, we present a Dialogue Discourse-Dware Meeting Summarizer (DDAMS) to explicitly model the interaction between utterances in a meeting by modeling different discourse relations. The core module is a relational graph encoder, where the utterances and discourse relations are modeled in a graph interaction manner. Moreover, we devise a Dialogue Discourse-Aware Data Augmentation (DDADA) strategy to construct a pseudo-summarization corpus from existing input meetings, which is 20 times larger than the original dataset and can be used to pretrain DDAMS. Experimental results on AMI and ICSI meeting datasets show that our full system can achieve SOTA performance. Our codes and outputs are available at https://github.com/xcfcode/DDAMS/.

While the abuse of deepfake technology has brought about a serious impact on human society, the detection of deepfake videos is still very challenging due to their highly photorealistic synthesis on each frame. To address that, this paper aims to leverage the possible inconsistent cues among video frames and proposes a Temporal Dropout 3-Dimensional Convolutional Neural Network (TD-3DCNN) to detect deepfake videos. In the approach, the fixed-length frame volumes sampled from a video are fed into a 3-Dimensional Convolutional Neural Network (3DCNN) to extract features across different scales and identified whether they are real or fake. Especially, a temporal dropout operation is introduced to randomly sample frames in each batch. It serves as a simple yet effective data augmentation and can enhance the representation and generalization ability, avoiding model overfitting and improving detecting accuracy. In this way, the resulting video-level classifier is accurate and effective to identify deepfake videos. Extensive experiments on benchmarks including Celeb-DF(v2) and DFDC clearly demonstrate the effectiveness and generalization capacity of our approach.

Deep clustering techniques combine representation learning with clustering objectives to improve their performance. Among existing deep clustering techniques, autoencoder-based methods are the most prevalent ones. While they achieve promising clustering results, they suffer from an inherent conflict between preserving details, as expressed by the reconstruction loss, and finding similar groups by ignoring details, as expressed by the clustering loss. This conflict leads to brittle training procedures, dependence on trade-off hyperparameters and less interpretable results. We propose our framework, ACe/DeC, that is compatible with Autoencoder Centroid based Deep Clustering methods and automatically learns a latent representation consisting of two separate spaces. The clustering space captures all cluster-specific information and the shared space explains general variation in the data. This separation resolves the above mentioned conflict and allows our method to learn both detailed reconstructions and cluster specific abstractions. We evaluate our framework with extensive experiments to show several benefits: (1) cluster performance – on various data sets we outperform relevant baselines; (2) no hyperparameter tuning – this improved performance is achieved without introducing new clustering specific hyperparameters; (3) interpretability – isolating the cluster specific information in a separate space is advantageous for data exploration and interpreting the clustering results; and (4) dimensionality of the embedded space – we automatically learn a low dimensional space for clustering. Our ACe/DeC framework isolates cluster information, increases stability and interpretability, while improving cluster performance.