In this paper, we explore a novel model reusing task tailored for graph neural networks (GNNs), termed as “deep graph reprogramming”. We strive to reprogram a pretrained GNN, without amending raw node features nor model parameters, to handle a bunch of cross-level downstream tasks in various domains. To this end, we propose an innovative Data Reprogramming paradigm alongside a Model Reprogramming paradigm. The former one aims to address the challenge of diversified graph feature dimensions for various tasks on the input side, while the latter alleviates the dilemma of fixed per-task-per-model behavior on the model side. For data reprogramming, we specifically devise an elaborated Meta-FeatPadding method to deal with heterogeneous input dimensions, and also develop a transductive Edge-Slimming as well as an inductive Meta-GraPadding approach for diverse homogenous samples. Meanwhile, for model reprogramming, we propose a novel task-adaptive Reprogrammable-Aggregator, to endow the frozen model with larger expressive capacities in handling cross-domain tasks. Experiments on fourteen datasets across node/graph classification/regression, 3D object recognition, and distributed action recognition, demonstrate that the proposed methods yield gratifying results, on par with those by re-training from scratch.

The advancements of single-cell RNA sequencing (scRNA-seq) technologies have provided us unprecedented opportunities to characterize cellular states and investigate the mechanisms of complex diseases. Due to technical issues such as dropout events, scRNA-seq data contains excess of false zero counts, which has a substantial impact on the downstream analyses. Although several computational approaches have been proposed to impute dropout events in scRNA-seq data, there is no strong consensus on which is the best approach. In this study, we propose a novel weighted ensemble learning method, named EnTSSR, to impute dropout events in scRNA-seq data. By using a multi-view two-side sparse self-representation framework, our model can exploit the consensus similarities between genes and between cells based on the imputed results of various imputation methods. Moreover, we introduce a weighted ensemble strategy to leverage the information captured by various imputation methods effectively. Down-sampling experiments, clustering analysis, differential expression analysis and cell trajectory inference are carried out to evaluate the performance of our proposed model. Experiment results demonstrate that our EnTSSR can effectively recover the true expression pattern of scRNA-seq data.