Bilinear Graph Neural Network with Node Interactions

Graph Neural Network (GNN) is a powerful model to learn representations and make predictions on graph data. Existing efforts on GNN have largely defined the graph convolution as a weighted sum of the features of the connected nodes to form the representation of the target node. Nevertheless, the operation of weighted sum assumes the neighbor nodes are independent of each other, and ignores the possible interactions between them. When such interactions exist, such as the co-occurrence of two neighbor nodes is a strong signal of the target node's characteristics, existing GNN models may fail to capture the signal. In this work, we argue the importance of modeling the interactions between neighbor nodes in GNN. We propose a new graph convolution operator, which augments the weighted sum with pairwise interactions of the representations of neighbor nodes. We term this framework as Bilinear Graph Neural Network (BGNN), which improves GNN representation ability with bilinear interactions between neighbor nodes. In particular, we specify two BGNN models named BGCN and BGAT, based on the well-known GCN and GAT, respectively. Empirical results on three public benchmarks of semi-supervised node classification verify the effectiveness of BGNN --- BGCN (BGAT) outperforms GCN (GAT) by 1.6% (1.5%) in classification accuracy.

Deep Learning-based End-to-end Diagnosis System for Avascular Necrosis of Femoral Head

As the first diagnostic imaging modality of avascular necrosis of the femoral head (AVNFH), accurately staging AVNFH from a plain radiograph is critical and challenging for orthopedists. Thus, we propose a deep learning-based AVNFH diagnosis system (AVN-net). The proposed AVN-net reads plain radiographs of the pelvis, conducts diagnosis, and visualizes results automatically. Deep convolutional neural networks are trained to provide an end-to-end diagnosis solution, covering femoral head detection, exam-view/sides identification, AVNFH diagnosis, and key clinical note generation subtasks. AVN-net is able to obtain state-of-the-art testing AUC of 0.95 (95% CI: 0.92-0.98) in AVNFH detection and significantly greater F1 scores (p<0.01) than less-to-moderately experienced orthopedists in all diagnostic tests. Furthermore, two real-world pilot studies were conducted for diagnosis support and education assistance, respectively, to assess the utility of AVN-net. The experimental results are promising. With the AVN-net diagnosis as a reference, the diagnostic accuracy and consistency of all orthopedists considerably improved while requiring only 1/4 of the time. Students self-studying the AVNFH diagnosis using AVN-net can learn better and faster than the control group. To the best of our knowledge, this study is the first research on the prospective use of a deep learning-based diagnosis system for AVNFH by conducting two pilot studies representing real-world application scenarios. We have demonstrated that the proposed AVN-net achieves expert-level AVNFH diagnosis performance, provides efficient support in clinical decision-making, and effectively passes clinical experience to students.

LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation

Graph Convolution Network (GCN) has become new state-of-the-art for collaborative filtering. Nevertheless, the reasons of its effectiveness for recommendation are not well understood. Existing work that adapts GCN to recommendation lacks thorough ablation analyses on GCN, which is originally designed for graph classification tasks and equipped with many neural network operations. However, we empirically find that the two most common designs in GCNs --- feature transformation and nonlinear activation --- contribute little to the performance of collaborative filtering. Even worse, including them adds to the difficulty of training and degrades recommendation performance. In this work, we aim to simplify the design of GCN to make it more concise and appropriate for recommendation. We propose a new model named LightGCN, including only the most essential component in GCN --- neighborhood aggregation --- for collaborative filtering. Specifically, LightGCN learns user and item embeddings by linearly propagating them on the user-item interaction graph, and uses the weighted sum of the embeddings learned at all layers as the final embedding. Such simple, linear, and neat model is much easier to implement and train, exhibiting substantial improvements (about 16.5\% relative improvement on average) over Neural Graph Collaborative Filtering (NGCF) --- a state-of-the-art GCN-based recommender model --- under exactly the same experimental setting. Further analyses are provided towards the rationality of the simple LightGCN from both analytical and empirical perspectives.

Towards Understanding the Importance of Noise in Training Neural Networks

Numerous empirical evidence has corroborated that the noise plays a crucial rule in effective and efficient training of neural networks. The theory behind, however, is still largely unknown. This paper studies this fundamental problem through training a simple two-layer convolutional neural network model. Although training such a network requires solving a nonconvex optimization problem with a spurious local optimum and a global optimum, we prove that perturbed gradient descent and perturbed mini-batch stochastic gradient algorithms in conjunction with noise annealing is guaranteed to converge to a global optimum in polynomial time with arbitrary initialization. This implies that the noise enables the algorithm to efficiently escape from the spurious local optimum. Numerical experiments are provided to support our theory.

Inductive Bias of Gradient Descent based Adversarial Training on Separable Data

Adversarial training is a principled approach for training robust neural networks. Despite of tremendous successes in practice, its theoretical properties still remain largely unexplored. In this paper, we provide new theoretical insights of gradient descent based adversarial training by studying its computational properties, specifically on its inductive bias. We take the binary classification task on linearly separable data as an illustrative example, where the loss asymptotically attains its infimum as the parameter diverges to infinity along certain directions. Specifically, we show that when the adversarial perturbation during training has bounded $\ell_2$-norm, the classifier learned by gradient descent based adversarial training converges in direction to the maximum $\ell_2$-norm margin classifier at the rate of $\tilde{\mathcal{O}}(1/\sqrt{T})$, significantly faster than the rate $\mathcal{O}(1/\log T)$ of training with clean data. In addition, when the adversarial perturbation during training has bounded $\ell_q$-norm for some $q\ge 1$, the resulting classifier converges in direction to a maximum mixed-norm margin classifier, which has a natural interpretation of robustness, as being the maximum $\ell_2$-norm margin classifier under worst-case $\ell_q$-norm perturbation to the data. Our findings provide theoretical backups for adversarial training that it indeed promotes robustness against adversarial perturbation.

Similarity Grouping-Guided Neural Network Modeling for Maritime Time Series Prediction

Reliable and accurate prediction of time series plays a crucial role in maritime industry, such as economic investment, transportation planning, port planning and design, etc. The dynamic growth of maritime time series has the predominantly complex, nonlinear and non-stationary properties. To guarantee high-quality prediction performance, we propose to first adopt the empirical mode decomposition (EMD) and ensemble EMD (EEMD) methods to decompose the original time series into high- and low-frequency components. The low-frequency components can be easily predicted directly through traditional neural network (NN) methods. It is more difficult to predict high-frequency components due to their properties of weak mathematical regularity. To take advantage of the inherent self-similarities within high-frequency components, these components will be divided into several continuous small (overlapping) segments. The grouped segments with high similarities are then selected to form more proper training datasets for traditional NN methods. This regrouping strategy can assist in enhancing the prediction accuracy of high-frequency components. The final prediction result is obtained by integrating the predicted high- and low-frequency components. Our proposed three-step prediction frameworks benefit from the time series decomposition and similar segments grouping. Experiments on both port cargo throughput and vessel traffic flow have illustrated its superior performance in terms of prediction accuracy and robustness.

Imitating Targets from all sides: An Unsupervised Transfer Learning method for Person Re-identification

Person re-identification (Re-ID) models usually show a limited performance when they are trained on one dataset and tested on another dataset due to the inter-dataset bias (e.g. completely different identities and backgrounds) and the intra-dataset difference (e.g. camera invariance). In terms of this issue, given a labelled source training set and an unlabelled target training set, we propose an unsupervised transfer learning method characterized by 1) bridging inter-dataset bias and intra-dataset difference via a proposed ImitateModel simultaneously; 2) regarding the unsupervised person Re-ID problem as a semi-supervised learning problem formulated by a dual classification loss to learn a discriminative representation across domains; 3) exploiting the underlying commonality across different domains from the class-style space to improve the generalization ability of re-ID models. Extensive experiments are conducted on two widely employed benchmarks, including Market-1501 and DukeMTMC-reID, and experimental results demonstrate that the proposed method can achieve a competitive performance against other state-of-the-art unsupervised Re-ID approaches.

Transductive Zero-Shot Learning with Visual Structure Constraint

Zero-shot Learning (ZSL) aims to recognize objects of the unseen classes, whose instances may not have been seen during training. It associates seen and unseen classes with the common semantic space and provides the visual features for each data instance. Most existing methods first learn a compatible projection function between the semantic space and the visual space based on the data of source seen classes, then directly apply it to target unseen classes. However, in real scenarios, the data distribution between the source and target domain might not match well, thus causing the well-known domain shift problem. Based on the observation that visual features of test instances can be separated into different clusters, we propose a visual structure constraint on class centers for transductive ZSL, to improve the generality of the projection function (i.e. alleviate the above domain shift problem). Specifically, two different strategies (symmetric Chamfer-distance and bipartite matching) are adopted to align the projected unseen semantic centers and visual cluster centers of test instances. Experiments on three widely used datasets demonstrate that the proposed visual structure constraint can bring substantial performance gain consistently and achieve state-of-the-art results.

TechKG: A Large-Scale Chinese Technology-Oriented Knowledge Graph

Knowledge graph is a kind of valuable knowledge base which would benefit lots of AI-related applications. Up to now, lots of large-scale knowledge graphs have been built. However, most of them are non-Chinese and designed for general purpose. In this work, we introduce TechKG, a large scale Chinese knowledge graph that is technology-oriented. It is built automatically from massive technical papers that are published in Chinese academic journals of different research domains. Some carefully designed heuristic rules are used to extract high quality entities and relations. Totally, it comprises of over 260 million triplets that are built upon more than 52 million entities which come from 38 research domains. Our preliminary ex-periments indicate that TechKG has high adaptability and can be used as a dataset for many diverse AI-related applications. We released TechKG at: http://www.techkg.cn.