With the increasing computation of training graph neural networks (GNNs) on large-scale graphs, graph condensation (GC) has emerged as a promising solution to synthesize compact, substitute graphs of the large-scale original graphs for efficient GNN training. However, these condensed graphs are specifically designed for the node classification task, significantly limiting the versatility of the synthesized data across various downstream tasks. This limitation predominantly stems from the reliance of existing GC methods on classification as the surrogate task for optimization, which leads to an excessive dependence on node labels and restricts their utility in label-scarcity scenarios. More critically, this surrogate task tends to overfit class-specific information within the condensed graph, consequently restricting the generalization capabilities of GC for other downstream tasks. To address these challenges, we introduce Contrastive Graph Condensation (CTGC), which adopts a self-supervised surrogate task to extract critical, causal information from the original graph and enhance the cross-task generalizability of the condensed graph. Specifically, CTGC employs a dual-branch framework to disentangle the generation of the node attributes and graph structures, where a dedicated structural branch is designed to explicitly encode geometric information through nodes' positional embeddings. By implementing an alternating optimization scheme with contrastive loss terms, CTGC promotes the mutual enhancement of both branches and facilitates high-quality graph generation through the model inversion technique. Extensive experiments demonstrate that CTGC excels in handling various downstream tasks with a limited number of labels, consistently outperforming state-of-the-art GC methods.

Node classification with Graph Neural Networks (GNN) under a fixed set of labels is well studied, while Graph Few-Shot Class Incremental Learning (GFSCIL), which involves learning a GNN classifier as graph nodes and classes growing over time sporadically, has received much less attention despite its importance. We introduce inductive GFSCIL that continually learns novel classes with newly emerging nodes while maintaining performance on old classes without accessing previous data. This addresses the practical concern of transductive GFSCIL, which requires storing the entire graph with historical data. Compared to the transductive GFSCIL, the inductive setting exacerbates catastrophic forgetting due to inaccessible previous data during incremental training, in addition to the overfitting issue caused by label sparsity. Thus, we propose a novel method, called Topology-based class Augmentation and Prototype calibration (TAP). To be specific, it first performs a topology-based class augmentation method, helping replicate the setting of disjoint subgraphs with nodes of novel classes received in incremental sessions, to enhance backbone versatility. In incremental learning, given the limited number of novel class samples, we propose an iterative prototype calibration to improve the separation of class prototypes. Furthermore, as backbone fine-tuning poses the feature distribution drift, prototypes of old classes start failing over time, we propose the prototype shift method for old classes to compensate for the drift. We showcase the proposed method on four datasets.

The burgeoning volume of graph data presents significant computational challenges in training graph neural networks (GNNs), critically impeding their efficiency in various applications. To tackle this challenge, graph condensation (GC) has emerged as a promising acceleration solution, focusing on the synthesis of a compact yet representative graph for efficiently training GNNs while retaining performance. Despite the potential to promote scalable use of GNNs, existing GC methods are limited to aligning the condensed graph with merely the observed static graph distribution. This limitation significantly restricts the generalization capacity of condensed graphs, particularly in adapting to dynamic distribution changes. In real-world scenarios, however, graphs are dynamic and constantly evolving, with new nodes and edges being continually integrated. Consequently, due to the limited generalization capacity of condensed graphs, applications that employ GC for efficient GNN training end up with sub-optimal GNNs when confronted with evolving graph structures and distributions in dynamic real-world situations. To overcome this issue, we propose open-world graph condensation (OpenGC), a robust GC framework that integrates structure-aware distribution shift to simulate evolving graph patterns and exploit the temporal environments for invariance condensation. This approach is designed to extract temporal invariant patterns from the original graph, thereby enhancing the generalization capabilities of the condensed graph and, subsequently, the GNNs trained on it. Furthermore, to support the periodic re-condensation and expedite condensed graph updating in life-long graph learning, OpenGC reconstructs the sophisticated optimization scheme with kernel ridge regression and non-parametric graph convolution, significantly accelerating the condensation process while ensuring the exact solutions. Extensive experiments on both real-world and synthetic evolving graphs demonstrate that OpenGC outperforms state-of-the-art (SOTA) GC methods in adapting to dynamic changes in open-world graph environments.

Abstract Graph neural networks (GNNs) have achieved state-of-the-art results for semi-supervised node classification on graphs. Nevertheless, the challenge of how to effectively learn GNNs with very few labels is still under-explored. As one of the prevalent semi-supervised methods, pseudo-labeling has been proposed to explicitly address the label scarcity problem. It is the process of augmenting the training set with pseudo-labeled unlabeled nodes to retrain a model in a self-training cycle. However, the existing pseudo-labeling approaches often suffer from two major drawbacks. First, these methods conservatively expand the label set by selecting only high-confidence unlabeled nodes without assessing their informativeness. Second, these methods incorporate pseudo-labels to the same loss function with genuine labels, ignoring their distinct contributions to the classification task. In this paper, we propose a novel informative pseudo-labeling framework (InfoGNN) to facilitate learning of GNNs with very few labels. Our key idea is to pseudo-label the most informative nodes that can maximally represent the local neighborhoods via mutual information maximization. To mitigate the potential label noise and class-imbalance problem arising from pseudo-labeling, we also carefully devise a generalized cross entropy with a class-balanced regularization to incorporate pseudo-labels into model retraining. Extensive experiments on six real-world graph datasets validate that our proposed approach significantly outperforms state-of-the-art baselines and competitive self-supervised methods on graphs.

Credit Default Prediction (CDP) has received increasing attention with the prevalence of financial loaning services. Many research efforts have been dedicated to developing novel soft features (i.e. non-financial features), such that they can complement hard features (i.e. financial features) and assist to learn a better default predicting model. But most works combine those features from various sources by just concating them together, and ignore that inappropriate feature fusion methods would compromise model performances. Therefore, in this paper, we propose an Attentive Feature Fusion (AFF) framework for credit default prediction using deep neural networks (DNNs). According to distinct characteristics of the data features, we divide features into multiple groups, and learn their latent representations with separate DNNs, respectively. Then the attention mechanism is applied to integrate those representations together, which allows the important features to be always emphasized and contribute more to the final decision. Experiments on the Lending Club dataset demonstrate that the proposed method can effectively improve the default predicting performances.

Active learning (AL) on attributed graphs has received increasing attention with the prevalence of graph-structured data. Although AL has been widely studied for alleviating label sparsity issues with the conventional nonrelational data, how to make it effective over attributed graphs remains an open research question. Existing AL algorithms on node classification attempt to reuse the classic AL query strategies designed for nonrelational data. However, they suffer from two major limitations. First, different AL query strategies calculated in distinct scoring spaces are often naively combined to determine which nodes to be labeled. Second, the AL query engine and the learning of the classifier are treated as two separating processes, resulting in unsatisfactory performance. In this article, we propose a SEmisupervised Adversarial active Learning (SEAL) framework on attributed graphs, which fully leverages the representation power of deep neural networks and devises a novel AL query strategy for node classification in an adversarial way. Our framework learns two adversarial components; a graph embedding network that encodes both the unlabeled and labeled nodes into a common latent space, expecting to trick the discriminator to regard all nodes as already labeled, and a semisupervised discriminator network that distinguishes the unlabeled from the existing labeled nodes. The divergence score, generated by the discriminator in a unified latent space, serves as the informativeness measure to actively select the most informative node to be labeled by an oracle. The two adversarial components form a closed loop to mutually and simultaneously reinforce each other toward enhancing the AL performance. Extensive experiments on real-world networks validate the effectiveness of the SEAL framework with superior performance improvements to state-of-the-art baselines on node classification tasks.

High efficiency video coding (HEVC), which is the newest video coding standard currently, achieves the best coding efficiency compared with all the other existing video coding standards. However, the computational complexity of the typical HEVC encoder dramatically increases because of the recursive searching scheme for finding the best coding unit (CU) partitions. In this paper, an adaptive fast CU size decision algorithm for HEVC Intra prediction is proposed based on CU complexity classification (CC) by using machine learning (ML) technology. Firstly, certain image features are extracted to characterize the CU complexity, which has a strong relationship with CU partitions, and then, the support vector machine is employed to analyze and construct the classification model according to the CU complexity. Finally, the proposed adaptive fast CU size decision algorithm, named as CCML, is released based on the complexity classification. The experimental results show that the proposed algorithm could achieve around 60% encoding time reduction for various test video sequences on average with only 1.26% Bjontegaard delta bit rate increase compared with the reference test model HM15.0 of HEVC.

Because of the co-existing of H.264/AVC and high efficiency video coding standard (HEVC) in the coming long period, video transcoding technology has become an essential part of multimedia communication in the field of the Internet of Vehicles (IoV). However, due to the huge computational complexity of re-encoding processes, traditionally cascaded transcoders greatly increase the computing burden of the embedded devices and impact the real-time capability of transportation communication systems. In order to address this problem, a fast transcoding solution is proposed in this paper. First, we exploit the mapping relationship among H.264/AVC decoding information and HEVC coding unit (CU) depth decision and prediction unit (PU) mode decision. Then, a three-output classification model is built for CU depth decision processes, and a two-output classification model is built for PU mode selection processes by using support vector machine method. Finally, the models are applied into the cascaded transcoder to accelerate the re-encoding process. The experimental results show that our proposal averagely achieves up to 53.7% and 52.3% complexity reductions under Lowdelay_P_main and Randomaccess_main configurations, respectively, with the negligible rate-distortion degradation, which show a great potential in improving the transcoding efficiency in the real-time video communication system of IoV.