Data sparsity poses a significant challenge for recommendation systems, prompting the research of Cross-Domain Recommendation ( CDR ). CDR aims to leverage more user-item interaction information from source domains to improve the recommendation performance in the target domain. However, a major challenge in CDR is the identification of transferable features. Traditional CDR methods struggle to distinguish between the various features of users, including domain-invariant features that are effective for feature transfer and domain-specific features that are detrimental to cross-domain information transfer. In this article, we aim to disentangle domain-invariant features and domain-specific features and effectively utilize these different features. This enables effective domain-to-domain information transfer by only transferring domain-invariant features while still considering the role of domain-specific features within their respective domains. Based on the superiority of graph structural feature learning and disentangled represent learning, we propose \(\mathbf{DMGCDR}\) —a model that learns D isentangled user feature representations and constructs a M ulti- G raph network for bidirectional knowledge transfer of shared features for CDR . Specifically, we designed two regularization terms to disentangle domain-invariant features and domain-specific features. Subsequently, we established a multi-graph convolutional network to enhance domain-specific features within single-domain graphs and transfer domain-invariant features across cross-domain graphs. Our approach also includes designing feature constraints to enhance the combination of features derived from different graphs and to uncover potential correlations among them. Extensive experiments on real-world datasets have demonstrated that our model significantly outperforms state-of-the-art CDR approaches.

Segmentation is a critical step in analyzing the developing human fetal brain. There have been vast improvements in automatic segmentation methods in the past several years, and the Fetal Brain Tissue Annotation (FeTA) Challenge 2021 helped to establish an excellent standard of fetal brain segmentation. However, FeTA 2021 was a single center study, limiting real-world clinical applicability and acceptance. The multi-center FeTA Challenge 2022 focused on advancing the generalizability of fetal brain segmentation algorithms for magnetic resonance imaging (MRI). In FeTA 2022, the training dataset contained images and corresponding manually annotated multi-class labels from two imaging centers, and the testing data contained images from these two centers as well as two additional unseen centers. The multi-center data included different MR scanners, imaging parameters, and fetal brain super-resolution algorithms applied. 16 teams participated and 17 algorithms were evaluated. Here, the challenge results are presented, focusing on the generalizability of the submissions. Both in- and out-of-domain, the white matter and ventricles were segmented with the highest accuracy (Top Dice scores: 0.89, 0.87 respectively), while the most challenging structure remains the grey matter (Top Dice score: 0.75) due to anatomical complexity. The top 5 average Dices scores ranged from 0.81-0.82, the top 5 average percentile Hausdorff distance values ranged from 2.3-2.5mm, and the top 5 volumetric similarity scores ranged from 0.90-0.92. The FeTA Challenge 2022 was able to successfully evaluate and advance generalizability of multi-class fetal brain tissue segmentation algorithms for MRI and it continues to benchmark new algorithms.

The recommendation system is subject to various biases, resulting in different training and testing data distribution. Most previous work either relies on a part of uniform data to guide model training which is difficult to obtain, or trains without any use of uniform data. However, not using uniform data may result in the inability to observe user’s real behaviors, leading to the presence of confounding factors, which harms the performance of unbiased recommendations. In this work, we proposed a novel method IML to perform debiasing recommendations by leveraging only the statistical characteristics of uniform dataset and training data. We use Invariant Meta-Learning(IML) to learn invariant features that remain insensitive to distributional changes. Finally, we propose a sample hard-aware weighted method to enhance training. Extensive experiments on real-world datasets demonstrate the effectiveness of IML.

In recommender systems, learning high-quality user and item representations is crucial for predicting user preferences. However, there are various confounding factors in observational data, resulting in data bias, which hinders the learning of user and item representations. Recent work proposed to use uniform data to alleviate bias problem. However, these methods fail to learn pure representations for unbiased prediction, which are not affected by confounding factors. This paper introduces a novel disentangled framework, named CDLRec, for learning unbiased representations, leveraging uniform data as supervisory signal for disentangling. Furthermore, to address the scarcity problem of uniform data, the contrastive learning is utilized to implement disentanglement by providing augmented samples. Specifically, two contrastive strategies are designed based on different sampling ways for positives and negatives. Extensive experiments are conducted over two real-world datasets and the results demonstrate the superior performance of our proposed method.

Location-based service recommendations usually need to collect and analyze the location information of trajectories generated by users with smart phones or wearable devices. It is easy to cause the location privacy leaks. The current location privacy protection methods usually confuse the adversary by adding fake locations into real trajectories to achieve the goal of privacy protection. However, these methods fail to consider the utility of user trajectories in service recommendations. In this paper, we propose a privacy-preserving trajectory generation algorithm based on service semantic similarity. To improve the utility of privacy-preserving trajectories for service recommendations, the algorithm constructs a series of service semantic grid maps and generates fake individuals with privacy-preserving trajectories considering both utility and privacy. Simulation results show that the proposed algorithm can effectively hide the locations in real trajectories of individuals and obtain higher effectiveness for service recommendations.

Cross-domain recommendation(CDR) is an effective method to alleviate the data sparsity problem in the recommendation system. How to learn common and domain-specific feature embeddings of users and items is a challenge especially where the two domains do not completely share users. Based on the superiority of graph structure feature learning, we propose a method of multi-graph convolutional feature transfer for cross-domain recommendation (MGCDR). For the common feature learning, MGCDR applies the feature similarity method to construct the residual network with the overlapping users, and then pre-training the feature embedding of the overlapping users. For the specific feature learning, MGCDR transfers the pre-trained embedding to initialize the spcific-domain graph, and obtain specific features through domain-specific users information propagation. Moreover, the attention mechanism is adopted to adaptively fuse the feature embedding. MGCDR is a dual-target recommendation model that fully exploits cross-domain mutual enhancement and joint training. Experiments on two pairs of real-world cross-domain datasets show the effectiveness of MGCDR.

Deciphering microRNA (miRNA) targets is important for understanding the function of miRNAs as well as miRNA-based diagnostics and therapeutics. Given the highly cell-specific nature of miRNA regulation, recent computational approaches typically exploit expression data to identify the most physiologically relevant target messenger RNAs (mRNAs). Although effective, those methods usually require a large sample size to infer miRNA-mRNA interactions, thus limiting their applications in personalized medicine. In this study, we developed a novel miRNA target prediction algorithm called miRACLe (miRNA Analysis by a Contact modeL). It integrates sequence characteristics and RNA expression profiles into a random contact model, and determines the target preferences by relative probability of effective contacts in an individual-specific manner. Evaluation by a variety of measures shows that fitting TargetScan, a frequently used prediction tool, into the framework of miRACLe can improve its predictive power with a significant margin and consistently outperform other state-of-the-art methods in prediction accuracy, regulatory potential and biological relevance. Notably, the superiority of miRACLe is robust to various biological contexts, types of expression data and validation datasets, and the computation process is fast and efficient. Additionally, we show that the model can be readily applied to other sequence-based algorithms to improve their predictive power, such as DIANA-microT-CDS, miRanda-mirSVR and MirTarget4. MiRACLe is publicly available at https://github.com/PANWANG2014/miRACLe.

In TCM theory, the syndrome is crucial to diagnose diseases and treat patients. In syndrome identification, the relation of symptoms usually correlates with syndrome and represents the pattern of syndrome at symptomatic level. Hence, we learn models for classifying syndromes in depression using 4 different algorithms, which are naive Bayes, Bayes network, SVM and C4.5. From the results of classification, we find that the dependence of symptoms has something to do with the accuracies of syndrome classification. Then, 8 symptom networks corresponding to depression and 7 syndromes are constructed to explore the interaction profile of symptoms under syndrome. By comparing syndrome-specific symptom network to the base network of depression, we discover the enriched edges and different nodes to represent the pattern of each syndrome. Literature and symptom ranking by Fisher score demonstrate the correctness of the different nodes selected through network comparison. After all, the enriched edges and different nodes associated with a given syndrome reveal the pattern of that syndrome at symptomatic level.