Graph neural networks have shown significant success in the field of graph representation learning. Graph convolutions perform neighborhood aggregation and represent one of the most important graph operations. Nevertheless, one layer of these neighborhood aggregation methods only consider immediate neighbors, and the performance decreases when going deeper to enable larger receptive fields. Several recent studies attribute this performance deterioration to the over-smoothing issue, which states that repeated propagation makes node representations of different classes indistinguishable. In this work, we study this observation systematically and develop new insights towards deeper graph neural networks. First, we provide a systematical analysis on this issue and argue that the key factor compromising the performance significantly is the entanglement of representation transformation and propagation in current graph convolution operations. After decoupling these two operations, deeper graph neural networks can be used to learn graph node representations from larger receptive fields. We further provide a theoretical analysis of the above observation when building very deep models, which can serve as a rigorous and gentle description of the over-smoothing issue. Based on our theoretical and empirical analysis, we propose Deep Adaptive Graph Neural Network (DAGNN) to adaptively incorporate information from large receptive fields. A set of experiments on citation, co-authorship, and co-purchase datasets have confirmed our analysis and insights and demonstrated the superiority of our proposed methods.
Texas A&M University（德克萨斯农工大学）；The Hong Kong Polytechnic University（香港理工大学）；Samsung Research America（三星美国研究院）
Graph neural networks (GNNs), which learn the representation of a node by aggregating its neighbors, have become an effective computational tool in downstream applications. Over-smoothing is one of the key issues which limit the performance of GNNs as the number of layers increases. It is because the stacked aggregators would make node representations converge to indistinguishable vectors. Several attempts have been made to tackle the issue by bringing linked node pairs close and unlinked pairs distinct. However, they often ignore the intrinsic community structures and would result in sub-opti…
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Arizona State University（亚利桑那州立大学）
Recently, motivated by real applications, a major research direction in graph neural networks (GNNs) is to explore deeper structures. For instance, the graph connectivity is not always consistent with the label distribution (e.g., the closest neighbors of some nodes are not from the same category). In this case, GNNs need to stack more layers, in order to find the same categorical neighbors in a longer path for capturing the class-discriminative information. However, two major problems hinder the deeper GNNs to obtain satisfactory performance, i.e., vanishing gradient and over-smoothing. On on…
Tsinghua University（清华大学）；Johns Hopkins University（约翰斯·霍普金斯大学）；Microsoft Research Asia（微软亚洲研究院）
The invariance to permutations of the adjacency matrix, i.e., graph isomorphism, is an overarching requirement for Graph Neural Networks (GNNs). Conventionally, this prerequisite can be satisfied by the invariant operations over node permutations when aggregating messages. However, such an invariant manner may ignore the relationships among neighboring nodes, thereby hindering the expressivity of GNNs. In this work, we devise an efficient permutation-sensitive aggregation mechanism via permutation groups, capturing pairwise correlations between neighboring nodes. We prove that our approach is …
University of South Carolina（南卡罗来纳大学）；Department of Computer Science and Engineering（计算机科学与工程系）
Machine learning (ML) based materials discovery has emerged as one of the most promising approaches for breakthroughs in materials science. While heuristic knowledge based descriptors have been combined with ML algorithms to achieve good performance, the complexity of the physicochemical mechanisms makes it urgently needed to exploit representation learning from either compositions or structures for building highly effective materials machine learning models. Among these methods, the graph neural networks have shown the best performance by its capability to learn high-level features from cryst…
Molecular property prediction is gaining increasing attention due to its diverse applications. One task of particular interests and importance is to predict quantum chemical properties without 3D equilibrium structures. This is practically favorable since obtaining 3D equilibrium structures requires extremely expensive calculations. In this work, we design a deep graph neural network to predict quantum properties by directly learning from 2D molecular graphs. In addition, we propose a 3D graph neural network to learn from low-cost conformer sets, which can be obtained with open-source tools us…
The University of Texas at Austin（德克萨斯大学奥斯汀分校）；Rice University（莱斯大学）
Training deep graph neural networks (GNNs) is notoriously hard. Besides the standard plights in training deep architectures such as vanishing gradients and overfitting, it also uniquely suffers from over-smoothing, information squashing, and so on, which limits their potential power for encoding the high-order neighbor structure in large-scale graphs. Although numerous efforts are proposed to address these limitations, such as various forms of skip connections, graph normalization, and random dropping, it is difficult to disentangle the advantages brought by a deep GNN architecture from those …
DenseGNN: universal and scalable deeper graph neural networks for high-performance property prediction in crystals and molecules
School of Materials Science and Engineering, Shanghai Jiao Tong University（上海交通大学材料科学与工程学院）；Zhangjiang Institute for Advanced Study, Shanghai Jiao Tong University（上海交通大学张江高等研究院）；Materials Genome Initiative Center, Shanghai Jiao Tong University（上海交通大学材料基因组计划中心）
Convolutional graph networks are used in particle physics for effective event reconstructions and classifications. However, their performances can be limited by the considerable amount of sensors used in modern particle detectors if applied to sensor-level data. We present a pooling scheme that uses partitioning to create pooling kernels on graphs, similar to pooling on images. Partition pooling can be used to adopt successful image recognition architectures for graph neural network applications in particle physics. The reduced computational resources allow for deeper networks and more extensi…
Knowledge distillation (KD) has demonstrated its effectiveness to boost the performance of graph neural networks (GNNs), where its goal is to distill knowledge from a deeper teacher GNN into a shallower student GNN. However, it is actually difficult to train a satisfactory teacher GNN due to the well-known over-parametrized and over-smoothing issues, leading to invalid knowledge transfer in practical applications. In this paper, we propose the first Free-direction Knowledge Distillation framework via Reinforcement learning for GNNs, called FreeKD, which is no longer required to provide a deepe…
Michigan State University（密歇根州立大学）；New Jersey Institute of Technology（新泽西理工学院）；IBM T.J. Watson Research Center（IBM托马斯·J·沃森研究中心）
Recent years have witnessed remarkable success achieved by graph neural networks (GNNs) in many real-world applications such as recommendation and drug discovery. Despite the success, oversmoothing has been identified as one of the key issues which limit the performance of deep GNNs. It indicates that the learned node representations are highly indistinguishable due to the stacked aggregators. In this paper, we propose a new perspective to look at the performance degradation of deep GNNs, i.e., feature overcorrelation. Through empirical and theoretical study on this matter, we demonstrate the …
Graph Convolutional Networks (GCNs) have been drawing significant attention with the power of representation learning on graphs. Unlike Convolutional Neural Networks (CNNs), which are able to take advantage of stacking very deep layers, GCNs suffer from vanishing gradient, over-smoothing and over-fitting issues when going deeper. These challenges limit the representation power of GCNs on large-scale graphs. This paper proposes DeeperGCN that is capable of successfully and reliably training very deep GCNs. We define differentiable generalized aggregation functions to unify different message agg…
Graph Neural Networks (GNNs) have achieved significant success in learning better representations by performing feature propagation and transformation iteratively to leverage neighborhood information. Nevertheless, iterative propagation restricts the information of higher-layer neighborhoods to be transported through and fused with the lower-layer neighborhoods', which unavoidably results in feature smoothing between neighborhoods in different layers and can thus compromise the performance, especially on heterophily networks. Furthermore, most deep GNNs only recognize the importance of higher-…
Graph Neural Networks (GNNs) are widely used to perform different machine learning tasks on graphs. As the size of the graphs grows, and the GNNs get deeper, training and inference time become costly in addition to the memory requirement. Thus, without sacrificing accuracy, graph sparsification, or model compression becomes a viable approach for graph learning tasks. A few existing techniques only study the sparsification of graphs and GNN models. In this paper, we develop a SparseGCN pipeline to study all possible sparsification in GNN. We provide a theoretical analysis and empirically show t…
Understanding the effect of depth in deep learning is a critical problem. In this work, we utilize the Fourier analysis to empirically provide a promising mechanism to understand why feedforward deeper learning is faster. To this end, we separate a deep neural network, trained by normal stochastic gradient descent, into two parts during analysis, i.e., a pre-condition component and a learning component, in which the output of the pre-condition one is the input of the learning one. We use a filtering method to characterize the frequency distribution of a high-dimensional function. Based on expe…
Universitat de Barcelona（巴塞罗那大学）；Bucknell University（巴克内尔大学）
Neural networks are more expressive when they have multiple layers. In turn, conventional training methods are only successful if the depth does not lead to numerical issues such as exploding or vanishing gradients, which occur less frequently when the layers are sufficiently wide. However, increasing width to attain greater depth entails the use of heavier computational resources and leads to overparameterized models. These subsequent issues have been partially addressed by model compression methods such as quantization and pruning, some of which relying on normalization-based regularization …
Laboratoire de Probabilités, Statistique et Modélisation（概率、统计与建模实验室）；Sorbonne Université（索邦大学）
Training a very deep neural network is a challenging task, as the deeper a neural network is, the more non-linear it is. We compare the performances of various preconditioned Langevin algorithms with their non-Langevin counterparts for the training of neural networks of increasing depth. For shallow neural networks, Langevin algorithms do not lead to any improvement, however the deeper the network is and the greater are the gains provided by Langevin algorithms. Adding noise to the gradient descent allows to escape from local traps, which are more frequent for very deep neural networks. Follow…
Towards the Training of Deeper Predictive Coding Neural Networks
Institute of Logic and Computation, Vienna University of Technology（逻辑与计算研究所，维也纳技术大学）；University of Oxford（牛津大学）；VERSES AI Research Lab（VERSES AI研究实验室）
Istanbul Technical University（伊斯坦布尔技术大学）；University of Maryland Baltimore County（马里兰大学巴尔的摩县分校）；Istanbul Medeniyet University（伊斯坦布尔梅德尼耶特大学）
Graph autoencoders are efficient at embedding graph-based data sets. Most graph autoencoder architectures have shallow depths which limits their ability to capture meaningful relations between nodes separated by multi-hops. In this paper, we propose Residual Variational Graph Autoencoder, ResVGAE, a deep variational graph autoencoder model with multiple residual modules. We show that our multiple residual modules, a convolutional layer with residual connection, improve the average precision of the graph autoencoders. Experimental results suggest that our proposed model with residual modules ou…
Deeper Insights into Deep Graph Convolutional Networks: Stability and Generalization
Department of Mathematics, College of Mathematics and Informatics, South China Agricultural University, Guangzhou, China（数学系，数学与信息学学院，华南农业大学，广州，中国）；Zhejiang Key Laboratory of Intelligent Education Technology and Application, Zhejiang Normal University, Jinhua, China（智能教育技术与应用浙江省重点实验室，浙江师范大学，金华，中国）；Department of Mathematics, City University of Hong Kong, Hong Kong, China（数学系，香港城市大学，香港，中国）
Osnabrück University（奥斯纳布吕克大学）；Osnabrück University of Applied Sciences（奥斯纳布吕克应用科学大学）
When optimizing convolutional neural networks (CNN) for a specific image-based task, specialists commonly overshoot the number of convolutional layers in their designs. By implication, these CNNs are unnecessarily resource intensive to train and deploy, with diminishing beneficial effects on the predictive performance. The features a convolutional layer can process are strictly limited by its receptive field. By layer-wise analyzing the size of the receptive fields, we can reliably predict sequences of layers that will not contribute qualitatively to the test accuracy in the given CNN architec…
Indian Institute of Technology Hyderabad（印度理工学院海得拉巴分校）
Loss landscape analysis is extremely useful for a deeper understanding of the generalization ability of deep neural network models. In this work, we propose a layerwise loss landscape analysis where the loss surface at every layer is studied independently and also on how each correlates to the overall loss surface. We study the layerwise loss landscape by studying the eigenspectra of the Hessian at each layer. In particular, our results show that the layerwise Hessian geometry is largely similar to the entire Hessian. We also report an interesting phenomenon where the Hessian eigenspectrum of …
CSIRO Health and Biosecurity（联邦科学与工业研究组织健康与生物安全部门）；The Australian eHealth Research Centre（澳大利亚电子健康研究中心）；CSIRO Data61（联邦科学与工业研究组织数据61所）
Brain morphometry from magnetic resonance imaging (MRI) is a consolidated biomarker for many neurodegenerative diseases. Recent advances in this domain indicate that deep convolutional neural networks can infer morphometric measurements within a few seconds. Nevertheless, the accuracy of the devised model for insightful bio-markers (mean curvature and thickness) remains unsatisfactory. In this paper, we propose a more accurate and efficient neural network model for brain morphometry named HerstonNet. More specifically, we develop a 3D ResNet-based neural network to learn rich features directly…
Fast Think-on-Graph: Wider, Deeper and Faster Reasoning of Large Language Model on Knowledge Graph
School of Cyberspace Security, Beijing University Of Posts And Telecommunications（北京邮电大学网络空间安全学院）；Peng Cheng Laboratory（鹏城实验室）；School of Computer Science and Technology, Harbin Institute of Technology (Shenzhen)（哈尔滨工业大学（深圳）计算机科学与技术学院）
HOTCAKE: Higher Order Tucker Articulated Kernels for Deeper CNN Compression
Institute for Clarity in Documentation（清晰度文献研究所）；Inria Paris-Rocquencourt（法国国家信息与自动化研究所巴黎-罗康库尔中心）；Rajiv Gandhi University（拉吉夫·甘地大学）；Tsinghua University（清华大学）；Palmer Research Laboratories（帕尔默研究实验室）
Universidade da Beira Interior（贝拉内陆大学）；NOVA Lincs（NOVA 林肯斯实验室）
Neural Architecture Search (NAS) benchmarks significantly improved the capability of developing and comparing NAS methods while at the same time drastically reduced the computational overhead by providing meta-information about thousands of trained neural networks. However, tabular benchmarks have several drawbacks that can hinder fair comparisons and provide unreliable results. These usually focus on providing a small pool of operations in heavily constrained search spaces -- usually cell-based neural networks with pre-defined outer-skeletons. In this work, we conducted an empirical analysis …
Austria Institute of Technology（奥地利技术研究所）；The University of Danang（岘港大学）；University of Essex（埃塞克斯大学）；Hongik University（弘益大学）
In this paper, we present a robust and low complexity system for Acoustic Scene Classification (ASC), the task of identifying the scene of an audio recording. We first construct an ASC baseline system in which a novel inception-residual-based network architecture is proposed to deal with the mismatched recording device issue. To further improve the performance but still satisfy the low complexity model, we apply two techniques: ensemble of multiple spectrograms and channel reduction on the ASC baseline system. By conducting extensive experiments on the benchmark DCASE 2020 Task 1A Development …
KingSoft AI Lab（金山AI实验室）；University of Groningen（格罗宁根大学）
Learning deeper models is usually a simple and effective approach to improve model performance, but deeper models have larger model parameters and are more difficult to train. To get a deeper model, simply stacking more layers of the model seems to work well, but previous works have claimed that it cannot benefit the model. We propose to train a deeper model with recurrent mechanism, which loops the encoder and decoder blocks of Transformer in the depth direction. To address the increasing of model parameters, we choose to share parameters in different recursive moments. We conduct our experim…
Confidence-gated training for efficient early-exit neural networks
International University of Rabat（拉巴特国际大学）；TICLab（TIC实验室）；Thales Alenia Space（泰勒斯阿莱尼亚空间）；Mohammed 6 Polytechnic University（穆莱·6技术大学）；University of Leeds（利兹大学）
RIKEN Center for Emergent Matter Science (CEMS)（理化学研究所涌现物质科学中心（CEMS））；The University of Tokyo（东京大学）
It has been recognized that a heavily overparameterized artificial neural network exhibits surprisingly good generalization performance in various machine-learning tasks. Recent theoretical studies have made attempts to unveil the mystery of the overparameterization. In most of those previous works, the overparameterization is achieved by increasing the width of the network, while the effect of increasing the depth has remained less well understood. In this work, we investigate the effect of increasing the depth within an overparameterized regime. To gain an insight into the advantage of depth…
