Diffusion-based imputation methods, which impute missing features through the iterative propagation of observed features, have shown impressive performance in homogeneous graphs. However, these methods are not directly applicable to heterogeneous graphs, which have multiple types of nodes and edges, due to two key issues: (1) the presence of nodes with undefined features hinders diffusion-based imputation; (2) treating various edge types equally during diffusion does not fully utilize information contained in heterogeneous graphs. To address these challenges, this paper presents a novel imputation scheme that enables diffusion-based imputation in heterogeneous graphs. Our key idea involves (1) assigning a {\it virtual feature} to an undefined node feature and (2) determining the importance of each edge type during diffusion according to a new criterion. Through experiments, we demonstrate that our virtual feature scheme effectively serves as a bridge between existing diffusion-based methods and heterogeneous graphs, maintaining the advantages of these methods. Furthermore, we confirm that adjusting the importance of each edge type leads to significant performance gains on heterogeneous graphs. Extensive experimental results demonstrate the superiority of our scheme in both semi-supervised node classification and link prediction tasks on heterogeneous graphs with missing rates ranging from low to exceedingly high. The source code is available at https://github.com/daehoum1/hetgfd.

In this paper, we address learning tasks on graphs with missing features, enhancing the applicability of graph neural networks to real-world graph-structured data. We identify a critical limitation of existing imputation methods based on feature propagation: they produce channels with nearly identical values within each channel, and these low-variance channels contribute very little to performance in graph learning tasks. To overcome this issue, we introduce synthetic features that target the root cause of low-variance channel production, thereby increasing variance in these channels. By preventing propagation-based imputation methods from generating meaningless feature values shared across all nodes, our synthetic feature propagation scheme mitigates significant performance degradation, even under extreme missing rates. Extensive experiments demonstrate the effectiveness of our approach across various graph learning tasks with missing features, ranging from low to extremely high missing rates. Additionally, we provide both empirical evidence and theoretical proof to validate the low-variance problem. The source code is available at https://github.com/daehoum1/fisf.

Abstract Graph neural networks (GNNs) have seen significant growth in analyzing heterogeneous graphs, which contain different types of nodes and connections. Recent studies have attempted to analyze heterogeneous graphs without relying on manually selected metapaths. However, existing methods often struggle to learn long-range dependencies between distant nodes because of over-squashing and over-smoothing. This paper presents a new approach, type-aligning graph diffusion network (TAGDN), that effectively addresses these challenges. Our main idea is to boost the power of graph diffusion techniques in heterogeneous graphs with a type-adaptive alignment scheme. This approach allows us to exploit long-range dependencies between nodes in heterogeneous graphs without suffering over-squashing and over-smoothing. Through extensive experiments, we show that TAGDN outperforms state-of-the-art heterogeneous GNNs on diverse graph analysis tasks. Ablation studies show that our alignment scheme enhances the capacity of graph diffusion networks in heterogeneous graph analysis, yielding an average improvement of 5.02% in node classification and 25.43% in node clustering. Our implementation is available at https://github.com/SeongJinAhn/TAGDN .

Single-cell RNA sequencing (scRNA-seq) technologies enable the exploration of cellular heterogeneity and facilitate the construction of cell atlases. However, scRNA-seq data often contain a large portion of missing values (false zeros) or noisy values, hindering downstream analyses. To recover these false zeros, propagation-based imputation methods have been proposed using $k$-NN graphs. However they model only associating relationships among genes within a cell, while, according to well-known genetic evidence, there are both associating and dissociating relationships among genes. To apply this genetic evidence to gene-gene relationship modeling, this paper proposes a novel imputation method that newly employs dissociating relationships in addition to associating relationships. Our method constructs a $k$-NN graph to additionally model dissociating relationships via the negation of a given cell-gene matrix. Moreover, our method standardizes the value distribution (mean and variance) of each gene to have standard distributions regardless of the gene. Through extensive experiments, we demonstrate that the proposed method achieves exceptional performance gains over state-of-the-art methods in both cell clustering and gene expression recovery across six scRNA-seq datasets, validating the significance of using complete gene-gene relationships in accordance with genetic evidence. The source code is available at https://github.com/daehoum1/scCR.

Most Knowledge Graph(KG) embedding models require negative sampling to learn the representations of KG by discriminating the differences between positive and negative triples. Knowledge representation learning tasks such as link prediction are heavily influenced by the quality of negative samples. Despite many attempts, generating high-quality negative samples remains a challenge. In this paper, we propose a novel framework, Bootstrapped Knowledge graph Embedding based on Neighbor Expansion (BKENE), which learns representations of KG without using negative samples. Our model avoids using augmentation methods that can alter the semantic information when creating the two semantically similar views of KG. In particular, we generate an alternative view of KG by aggregating the information of the expanded neighbor of each node with multi-hop relation. Experimental results show that our BKENE outperforms the state-of-the-art methods for link prediction tasks.

Link prediction is one of the key problems for graph-structured data. With the advancement of graph neural networks, graph autoencoders (GAEs) and variational graph autoencoders (VGAEs) have been proposed to learn graph embeddings in an unsupervised way. It has been shown that these methods are effective for link prediction tasks. However, they do not work well in link predictions when a node whose degree is zero (i.g., isolated node) is involved. We have found that GAEs/VGAEs make embeddings of isolated nodes close to zero regardless of their content features. In this paper, we propose a novel Variational Graph Normalized AutoEncoder (VGNAE) that utilize L2-normalization to derive better embeddings for isolated nodes. We show that our VGNAEs outperform the existing state-of-the-art models for link prediction tasks. The code is available at https://github.com/SeongJinAhn/VGNAE.