Learning semantic-rich representations from raw unlabeled time series data is critical for downstream tasks such as classification and forecasting. Contrastive learning has recently shown its promising representation learning capability in the absence of expert annotations. However, existing contrastive approaches generally treat each instance independently, which leads to false negative pairs that share the same semantics. To tackle this problem, we propose MHCCL, a Masked Hierarchical Cluster-wise Contrastive Learning model, which exploits semantic information obtained from the hierarchical structure consisting of multiple latent partitions for multivariate time series. Motivated by the observation that fine-grained clustering preserves higher purity while coarse-grained one reflects higher-level semantics, we propose a novel downward masking strategy to filter out fake negatives and supplement positives by incorporating the multi-granularity information from the clustering hierarchy. In addition, a novel upward masking strategy is designed in MHCCL to remove outliers of clusters at each partition to refine prototypes, which helps speed up the hierarchical clustering process and improves the clustering quality. We conduct experimental evaluations on seven widely-used multivariate time series datasets. The results demonstrate the superiority of MHCCL over the state-of-the-art approaches for unsupervised time series representation learning.
Contrastive learning (CL) has recently been demonstrated critical in improving recommendation performance. The fundamental idea of CL-based recommendation models is to maximize the consistency between representations learned from different graph augmentations of the user-item bipartite graph. In such a self-supervised manner, CL-based recommendation models are expected to extract general features from the raw data to tackle the data sparsity issue. Despite the effectiveness of this paradigm, we still have no clue what underlies the performance gains. In this paper, we first reveal that CL enhances recommendation through endowing the model with the ability to learn more evenly distributed user/item representations, which can implicitly alleviate the pervasive popularity bias and promote long-tail items. Meanwhile, we find that the graph augmentations, which were considered a necessity in prior studies, are relatively unreliable and less significant in CL-based recommendation. On top of these findings, we put forward an eXtremely Simple Graph Contrastive Learning method (XSimGCL) for recommendation, which discards the ineffective graph augmentations and instead employs a simple yet effective noise-based embedding augmentation to create views for CL. A comprehensive experimental study on three large and highly sparse benchmark datasets demonstrates that, though the proposed method is extremely simple, it can smoothly adjust the uniformity of learned representations and outperforms its graph augmentation-based counterparts by a large margin in both recommendation accuracy and training efficiency. The code is released at https://github.com/Coder-Yu/SELFRec.
The sequential recommendation systems capture users' dynamic behavior patterns to predict their next interaction behaviors. Most existing sequential recommendation methods only exploit the local context information of an individual interaction sequence and learn model parameters solely based on the item prediction loss. Thus, they usually fail to learn appropriate sequence representations. This paper proposes a novel recommendation framework, namely Graph Contrastive Learning for Sequential Recommendation (GCL4SR). Specifically, GCL4SR employs a Weighted Item Transition Graph (WITG), built based on interaction sequences of all users, to provide global context information for each interaction and weaken the noise information in the sequence data. Moreover, GCL4SR uses subgraphs of WITG to augment the representation of each interaction sequence. Two auxiliary learning objectives have also been proposed to maximize the consistency between augmented representations induced by the same interaction sequence on WITG, and minimize the difference between the representations augmented by the global context on WITG and the local representation of the original sequence. Extensive experiments on real-world datasets demonstrate that GCL4SR consistently outperforms state-of-the-art sequential recommendation methods.
Question Generation (QG), as a challenging Natural Language Processing task, aims at generating questions based on given answers and context. Existing QG methods mainly focus on building or training models for specific QG datasets. These works are subject to two major limitations: (1) They are dedicated to specific QG formats (e.g., answer-extraction or multi-choice QG), therefore, if we want to address a new format of QG, a re-design of the QG model is required. (2) Optimal performance is only achieved on the dataset they were just trained on. As a result, we have to train and keep various QG models for different QG datasets, which is resource-intensive and ungeneralizable. To solve the problems, we propose a model named Unified-QG based on lifelong learning techniques, which can continually learn QG tasks across different datasets and formats. Specifically, we first build a format-convert encoding to transform different kinds of QG formats into a unified representation. Then, a method named \emph{STRIDER} (\emph{S}imilari\emph{T}y \emph{R}egular\emph{I}zed \emph{D}ifficult \emph{E}xample \emph{R}eplay) is built to alleviate catastrophic forgetting in continual QG learning. Extensive experiments were conducted on $8$ QG datasets across $4$ QG formats (answer-extraction, answer-abstraction, multi-choice, and boolean QG) to demonstrate the effectiveness of our approach. Experimental results demonstrate that our Unified-QG can effectively and continually adapt to QG tasks when datasets and formats vary. In addition, we verify the ability of a single trained Unified-QG model in improving $8$ Question Answering (QA) systems' performance through generating synthetic QA data.
Contrastive learning (CL) recently has received considerable attention in the field of recommendation, since it can greatly alleviate the data sparsity issue and improve recommendation performance in a self-supervised manner. A typical way to apply CL to recommendation is conducting edge/node dropout on the user-item bipartite graph to augment the graph data and then maximizing the correspondence between representations of the same user/item augmentations under a joint optimization setting. Despite the encouraging results brought by CL, however, what underlies the performance gains still remains unclear. In this paper, we first experimentally demystify that the uniformity of the learned user/item representation distributions on the unit hypersphere is closely related to the recommendation performance. Based on the experimental findings, we propose a graph augmentation-free CL method to simply adjust the uniformity by adding uniform noises to the original representations for data augmentations, and enhance recommendation from a geometric view. Specifically, the constant graph perturbation during training is not required in our method and hence the positive and negative samples for CL can be generated on-the-fly. The experimental results on three benchmark datasets demonstrate that the proposed method has distinct advantages over its graph augmentation-based counterparts in terms of both the ability to improve recommendation performance and the running/convergence speed. The code is released at https://github.com/Coder-Yu/QRec.
Cross-modal hashing (CMH) is one of the most promising methods in cross-modal approximate nearest neighbor search. Most CMH solutions ideally assume the labels of training and testing set are identical. However, the assumption is often violated, causing a zero-shot CMH problem. Recent efforts to address this issue focus on transferring knowledge from the seen classes to the unseen ones using label attributes. However, the attributes are isolated from the features of multi-modal data. To reduce the information gap, we introduce an approach called LAEH (Label Attributes Embedding for zero-shot cross-modal Hashing). LAEH first gets the initial semantic attribute vectors of labels by word2vec model and then uses a transformation network to transform them into a common subspace. Next, it leverages the hash vectors and the feature similarity matrix to guide the feature extraction network of different modalities. At the same time, LAEH uses the attribute similarity as the supplement of label similarity to rectify the label embedding and common subspace. Experiments show that LAEH outperforms related representative zero-shot and cross-modal hashing methods.
We raise and define a new crowdsourcing scenario, open set crowdsourcing, where we only know the general theme of an unfamiliar crowdsourcing project, and we don't know its label space, that is, the set of possible labels. This is still a task annotating problem, but the unfamiliarity with the tasks and the label space hampers the modelling of the task and of workers, and also the truth inference. We propose an intuitive solution, OSCrowd. First, OSCrowd integrates crowd theme related datasets into a large source domain to facilitate partial transfer learning to approximate the label space inference of these tasks. Next, it assigns weights to each source domain based on category correlation. After this, it uses multiple-source open set transfer learning to model crowd tasks and assign possible annotations. The label space and annotations given by transfer learning will be used to guide and standardize crowd workers' annotations. We validate OSCrowd in an online scenario, and prove that OSCrowd solves the open set crowdsourcing problem, works better than related crowdsourcing solutions.
Due to the unreliability of Internet workers, it's difficult to complete a crowdsourcing project satisfactorily, especially when the tasks are multiple and the budget is limited. Recently, meta learning has brought new vitality to few-shot learning, making it possible to obtain a classifier with a fair performance using only a few training samples. Here we introduce the concept of \emph{meta-worker}, a machine annotator trained by meta learning for types of tasks (i.e., image classification) that are well-fit for AI. Unlike regular crowd workers, meta-workers can be reliable, stable, and more importantly, tireless and free. We first cluster unlabeled data and ask crowd workers to repeatedly annotate the instances nearby the cluster centers; we then leverage the annotated data and meta-training datasets to build a cluster of meta-workers using different meta learning algorithms. Subsequently, meta-workers are asked to annotate the remaining crowdsourced tasks. The Jensen-Shannon divergence is used to measure the disagreement among the annotations provided by the meta-workers, which determines whether or not crowd workers should be invited for further annotation of the same task. Finally, we model meta-workers' preferences and compute the consensus annotation by weighted majority voting. Our empirical study confirms that, by combining machine and human intelligence, we can accomplish a crowdsourcing project with a lower budget than state-of-the-art task assignment methods, while achieving a superior or comparable quality.
On the increase of major depressive disorders (MDD), many researchers paid attention to their recognition and treatment. Existing MDD recognition algorithms always use a single time-frequency domain method method, but the single time-frequency domain method is too simple and is not conducive to simulating the complex link relationship between brain functions. To solve this problem, this paper proposes a recognition method based on multi-layer brain functional connectivity networks (MBFCN) for major depressive disorder and conducts cognitive analysis. Cognitive analysis based on the proposed MBFCN finds that the Alpha-Beta1 frequency band is the key sub-band for recognizing MDD. The connections between the right prefrontal lobe and the temporal lobe of the extremely depressed disorders (EDD) are deficient in the brain functional connectivity networks (BFCN) based on phase lag index (PLI). Furthermore, potential biomarkers by the significance analysis of depression features and PHQ-9 can be found.