Qiuyu Zhu

Qiuyu Zhu, Liang Zhang, Qianxiong Xu et al.

Despite the success of Heterogeneous Graph Neural Networks (HGNNs) in modeling real-world Heterogeneous Information Networks (HINs), challenges such as expressiveness limitations and over-smoothing have prompted researchers to explore Graph Transformers (GTs) for enhanced HIN representation learning. However, research on GT in HINs remains limited, with two key shortcomings in existing work: (1) A node's neighbors at different distances in HINs convey diverse semantics. Unfortunately, existing methods ignore such differences and uniformly treat neighbors within a given distance in a coarse manner, which results in semantic confusion. (2) Nodes in HINs have various types, each with unique semantics. Nevertheless, existing methods mix nodes of different types during neighbor aggregation, hindering the capture of proper correlations between nodes of diverse types. To bridge these gaps, we design an innovative structure named (k,t)-ring neighborhood, where nodes are initially organized by their distance, forming different non-overlapping k-ring neighborhoods for each distance. Within each k-ring structure, nodes are further categorized into different groups according to their types, thus emphasizing the heterogeneity of both distances and types in HINs naturally. Based on this structure, we propose a novel Hierarchical Heterogeneous Graph Transformer (HHGT) model, which seamlessly integrates a Type-level Transformer for aggregating nodes of different types within each k-ring neighborhood, followed by a Ring-level Transformer for aggregating different k-ring neighborhoods in a hierarchical manner. Extensive experiments are conducted on downstream tasks to verify HHGT's superiority over 14 baselines, with a notable improvement of up to 24.75% in NMI and 29.25% in ARI for node clustering task on the ACM dataset compared to the best baseline.

Qiuyu Zhu, Guohui Zheng, Yingying Yan

Deep neural networks suffer from the overconfidence issue in the open world, meaning that classifiers could yield confident, incorrect predictions for out-of-distribution (OOD) samples. Thus, it is an urgent and challenging task to detect these samples drawn far away from training distribution based on the security considerations of artificial intelligence. Many current methods based on neural networks mainly rely on complex processing strategies, such as temperature scaling and input preprocessing, to obtain satisfactory results. In this paper, we propose an effective algorithm for detecting out-of-distribution examples utilizing PEDCC-Loss. We mathematically analyze the nature of the confidence score output by the PEDCC (Predefined Evenly-Distribution Class Centroids) classifier, and then construct a more effective scoring function to distinguish in-distribution (ID) and out-of-distribution. In this method, there is no need to preprocess the input samples and the computational burden of the algorithm is reduced. Experiments demonstrate that our method can achieve better OOD detection performance.

Qiuyu Zhu, Liheng Hu, Sijin Wang

In the face of complex natural images, existing deep clustering algorithms fall significantly short in terms of clustering accuracy when compared to supervised classification methods, making them less practical. This paper introduces an image clustering algorithm based on self-supervised pretrained models and latent feature distribution optimization, substantially enhancing clustering performance. It is found that: (1) For complex natural images, we effectively enhance the discriminative power of latent features by leveraging self-supervised pretrained models and their fine-tuning, resulting in improved clustering performance. (2) In the latent feature space, by searching for k-nearest neighbor images for each training sample and shortening the distance between the training sample and its nearest neighbor, the discriminative power of latent features can be further enhanced, and clustering performance can be improved. (3) In the latent feature space, reducing the distance between sample features and the nearest predefined cluster centroids can optimize the distribution of latent features, therefore further improving clustering performance. Through experiments on multiple datasets, our approach outperforms the latest clustering algorithms and achieves state-of-the-art clustering results. When the number of categories in the datasets is small, such as CIFAR-10 and STL-10, and there are significant differences between categories, our clustering algorithm has similar accuracy to supervised methods without using pretrained models, slightly lower than supervised methods using pre-trained models. The code linked algorithm is https://github.com/LihengHu/semi.

Qiuyu Zhu, Hao Wang, Xuewen Zu et al.

For the convolutional neural network (CNN) used for pattern classification, the training loss function is usually applied to the final output of the network, except for some regularization constraints on the network parameters. However, with the increasing of the number of network layers, the influence of the loss function on the network front layers gradually decreases, and the network parameters tend to fall into local optimization. At the same time, it is found that the trained network has significant information redundancy at all stages of features, which reduces the effectiveness of feature mapping at all stages and is not conducive to the change of the subsequent parameters of the network in the direction of optimality. Therefore, it is possible to obtain a more optimized solution of the network and further improve the classification accuracy of the network by designing a loss function for restraining the front stage features and eliminating the information redundancy of the front stage features .For CNN, this article proposes a multi-stage feature decorrelation loss (MFD Loss), which refines effective features and eliminates information redundancy by constraining the correlation of features at all stages. Considering that there are many layers in CNN, through experimental comparison and analysis, MFD Loss acts on multiple front layers of CNN, constrains the output features of each layer and each channel, and performs supervision training jointly with classification loss function during network training. Compared with the single Softmax Loss supervised learning, the experiments on several commonly used datasets on several typical CNNs prove that the classification performance of Softmax Loss+MFD Loss is significantly better. Meanwhile, the comparison experiments before and after the combination of MFD Loss and some other typical loss functions verify its good universality.

Qiuyu Zhu, Yiwei He

Out-of-distribution (OOD) detection, crucial for reliable pattern classification, discerns whether a sample originates outside the training distribution. This paper concentrates on the high-dimensional features output by the final convolutional layer, which contain rich image features. Our key idea is to project these high-dimensional features into two specific feature subspaces, leveraging the dimensionality reduction capacity of the network's linear layers, trained with Predefined Evenly-Distribution Class Centroids (PEDCC)-Loss. This involves calculating the cosines of three projection angles and the norm values of features, thereby identifying distinctive information for in-distribution (ID) and OOD data, which assists in OOD detection. Building upon this, we have modified the batch normalization (BN) and ReLU layer preceding the fully connected layer, diminishing their impact on the output feature distributions and thereby widening the distribution gap between ID and OOD data features. Our method requires only the training of the classification network model, eschewing any need for input pre-processing or specific OOD data pre-tuning. Extensive experiments on several benchmark datasets demonstrates that our approach delivers state-of-the-art performance. Our code is available at https://github.com/Hewell0/ProjOOD.

Qiuyu Zhu, Xuewen Zu

In the field of pattern classification, the training of deep learning classifiers is mostly end-to-end learning, and the loss function is the constraint on the final output (posterior probability) of the network, so the existence of Softmax is essential. In the case of end-to-end learning, there is usually no effective loss function that completely relies on the features of the middle layer to restrict learning, resulting in the distribution of sample latent features is not optimal, so there is still room for improvement in classification accuracy. Based on the concept of Predefined Evenly-Distributed Class Centroids (PEDCC), this article proposes a Softmax-free loss function based on predefined optimal-distribution of latent features-POD Loss. The loss function only restricts the latent features of the samples, including the norm-adaptive Cosine distance between the latent feature vector of the sample and the center of the predefined evenly-distributed class, and the correlation between the latent features of the samples. Finally, Cosine distance is used for classification. Compared with the commonly used Softmax Loss, some typical Softmax related loss functions and PEDCC-Loss, experiments on several commonly used datasets on several typical deep learning classification networks show that the classification performance of POD Loss is always significant better and easier to converge. Code is available in https://github.com/TianYuZu/POD-Loss.

Qiuyu Zhu, Zhengyong Wang

In this paper, we propose a novel, effective and simpler end-to-end image clustering auto-encoder algorithm: ICAE. The algorithm uses PEDCC (Predefined Evenly-Distributed Class Centroids) as the clustering centers, which ensures the inter-class distance of latent features is maximal, and adds data distribution constraint, data augmentation constraint, auto-encoder reconstruction constraint and Sobel smooth constraint to improve the clustering performance. Specifically, we perform one-to-one data augmentation to learn the more effective features. The data and the augmented data are simultaneously input into the autoencoder to obtain latent features and the augmented latent features whose similarity are constrained by an augmentation loss. Then, making use of the maximum mean discrepancy distance (MMD), we combine the latent features and augmented latent features to make their distribution close to the PEDCC distribution (uniform distribution between classes, Dirac distribution within the class) to further learn clustering-oriented features. At the same time, the MSE of the original input image and reconstructed image is used as reconstruction constraint, and the Sobel smooth loss to build generalization constraint to improve the generalization ability. Finally, extensive experiments on three common datasets MNIST, Fashion-MNIST, COIL20 are conducted. The experimental results show that the algorithm has achieved the best clustering results so far. In addition, we can use the predefined PEDCC class centers, and the decoder to clearly generate the samples of each class. The code can be downloaded at https://github.com/zyWang-Power/Clustering!

Qiuyu Zhu, Pengju Zhang, Xin Ye

With the development of convolutional neural networks (CNNs) in recent years, the network structure has become more and more complex and varied, and has achieved very good results in pattern recognition, image classification, object detection and tracking. For CNNs used for image classification, in addition to the network structure, more and more research is now focusing on the improvement of the loss function, so as to enlarge the inter-class feature differences, and reduce the intra-class feature variations as soon as possible. Besides the traditional Softmax, typical loss functions include L-Softmax, AM-Softmax, ArcFace, and Center loss, etc. Based on the concept of predefined evenly-distributed class centroids (PEDCC) in CSAE network, this paper proposes a PEDCC-based loss function called PEDCC-Loss, which can make the inter-class distance maximal and intra-class distance small enough in hidden feature space. Multiple experiments on image classification and face recognition have proved that our method achieve the best recognition accuracy, and network training is stable and easy to converge. Code is available in https://github.com/ZLeopard/PEDCC-Loss

Qiuyu Zhu, Ruixin Zhang

In order to enhance the real-time performance of convolutional neural networks(CNNs), more and more researchers are focusing on improving the efficiency of CNN. Based on the analysis of some CNN architectures, such as ResNet, DenseNet, ShuffleNet and so on, we combined their advantages and proposed a very efficient model called Highly Efficient Networks(HENet). The new architecture uses an unusual way to combine group convolution and channel shuffle which was mentioned in ShuffleNet. Inspired by ResNet and DenseNet, we also proposed a new way to use element-wise addition and concatenation connection with each block. In order to make greater use of feature maps, pooling operations are removed from HENet. The experiments show that our model's efficiency is more than 1 times higher than ShuffleNet on many open source datasets, such as CIFAR-10/100 and SVHN.

Qiuyu Zhu, Liang Zhang, Qianxiong Xu et al.

Representation learning on heterogeneous text-rich networks (HTRNs), which consist of multiple types of nodes and edges with each node associated with textual information, is essential for various real-world applications. Given the success of pretrained language models (PLMs) in processing text data, recent efforts have focused on integrating PLMs into HTRN representation learning. These methods typically handle textual and structural information separately, using both PLMs and heterogeneous graph neural networks (HGNNs). However, this separation fails to capture the critical interactions between these two types of information within HTRNs. Additionally, it necessitates an extra alignment step, which is challenging due to the fundamental differences between distinct embedding spaces generated by PLMs and HGNNs. To deal with it, we propose HierPromptLM, a novel pure PLM-based framework that seamlessly models both text data and graph structures without the need for separate processing. Firstly, we develop a Hierarchical Prompt module that employs prompt learning to integrate text data and heterogeneous graph structures at both the node and edge levels, within a unified textual space. Building upon this foundation, we further introduce two innovative HTRN-tailored pretraining tasks to fine-tune PLMs for representation learning by emphasizing the inherent heterogeneity and interactions between textual and structural information within HTRNs. Extensive experiments on two real-world HTRN datasets demonstrate HierPromptLM outperforms state-of-the-art methods, achieving significant improvements of up to 6.08% for node classification and 10.84% for link prediction.

Philipp Andelfinger, Jieyi Bi, Qiuyu Zhu et al.

In manufacturing, capacity planning is the process of allocating production resources in accordance with variable demand. The current industry practice in semiconductor manufacturing typically applies heuristic rules to prioritize actions, such as future change lists that account for incoming machine and recipe dedications. However, while offering interpretability, heuristics cannot easily account for the complex interactions along the process flow that can gradually lead to the formation of bottlenecks. Here, we present a neural network-based model for capacity planning on the level of individual machines, trained using deep reinforcement learning. By representing the policy using a heterogeneous graph neural network, the model directly captures the diverse relationships among machines and processing steps, allowing for proactive decision-making. We describe several measures taken to achieve sufficient scalability to tackle the vast space of possible machine-level actions. Our evaluation results cover Intel's small-scale Minifab model and preliminary experiments using the popular SMT2020 testbed. In the largest tested scenario, our trained policy increases throughput and decreases cycle time by about 1.8% each.

Zhengyong Wang, Liquan Shen, Mei Yu et al.

Most deep models for underwater image enhancement resort to training on synthetic datasets based on underwater image formation models. Although promising performances have been achieved, they are still limited by two problems: (1) existing underwater image synthesis models have an intrinsic limitation, in which the homogeneous ambient light is usually randomly generated and many important dependencies are ignored, and thus the synthesized training data cannot adequately express characteristics of real underwater environments; (2) most of deep models disregard lots of favorable underwater priors and heavily rely on training data, which extensively limits their application ranges. To address these limitations, a new underwater synthetic dataset is first established, in which a revised ambient light synthesis equation is embedded. The revised equation explicitly defines the complex mathematical relationship among intensity values of the ambient light in RGB channels and many dependencies such as surface-object depth, water types, etc, which helps to better simulate real underwater scene appearances. Secondly, a unified framework is proposed, named ANA-SYN, which can effectively enhance underwater images under collaborations of priors (underwater domain knowledge) and data information (underwater distortion distribution). The proposed framework includes an analysis network and a synthesis network, one for priors exploration and another for priors integration. To exploit more accurate priors, the significance of each prior for the input image is explored in the analysis network and an adaptive weighting module is designed to dynamically recalibrate them. Meanwhile, a novel prior guidance module is introduced in the synthesis network, which effectively aggregates the prior and data features and thus provides better hybrid information to perform the more reasonable image enhancement.

Haiping Hu, Yingying Yan, Qiuyu Zhu et al.

Predefined evenly-distributed class centroids (PEDCC) can be widely used in models and algorithms of pattern classification, such as CNN classifiers, classification autoencoders, clustering, and semi-supervised learning, etc. Its basic idea is to predefine the class centers, which are evenly-distributed on the unit hypersphere in feature space, to maximize the inter-class distance. The previous method of generating PEDCC uses an iterative algorithm based on a charge model, that is, the initial values of various centers (charge positions) are randomly set from the normal distribution, and the charge positions are updated iteratively with the help of the repulsive force between charges of the same polarity. The class centers generated by the algorithm will produce some errors with the theoretically evenly-distributed points, and the generation time will be longer. This paper takes advantage of regular polyhedron in high-dimensional space and the evenly distribution of points on the n dimensional hypersphere to generate PEDCC mathematically. Then, we discussed the basic and extensive characteristics of the frames formed by PEDCC. Finally, experiments show that new algorithm is not only faster than the iterative method, but also more accurate in position. The mathematical analysis and experimental results of this paper can provide a theoretical tool for using PEDCC to solve the key problems in the field of pattern recognition, such as interpretable supervised/unsupervised learning, incremental learning, uncertainty analysis and so on.

Joel Q. L. Chang, Qiuyu Zhu, Vincent Y. F. Tan

The multi-armed bandit (MAB) problem is a ubiquitous decision-making problem that exemplifies the exploration-exploitation tradeoff. Standard formulations exclude risk in decision making. Risk notably complicates the basic reward-maximising objective, in part because there is no universally agreed definition of it. In this paper, we consider a popular risk measure in quantitative finance known as the Conditional Value at Risk (CVaR). We explore the performance of a Thompson Sampling-based algorithm CVaR-TS under this risk measure. We provide comprehensive comparisons between our regret bounds with state-of-the-art L/UCB-based algorithms in comparable settings and demonstrate their clear improvement in performance. We also include numerical simulations to empirically verify that CVaR-TS outperforms other L/UCB-based algorithms.

Qiuyu Zhu, Vincent Y. F. Tan

The multi-armed bandit (MAB) problem is a classical learning task that exemplifies the exploration-exploitation tradeoff. However, standard formulations do not take into account {\em risk}. In online decision making systems, risk is a primary concern. In this regard, the mean-variance risk measure is one of the most common objective functions. Existing algorithms for mean-variance optimization in the context of MAB problems have unrealistic assumptions on the reward distributions. We develop Thompson Sampling-style algorithms for mean-variance MAB and provide comprehensive regret analyses for Gaussian and Bernoulli bandits with fewer assumptions. Our algorithms achieve the best known regret bounds for mean-variance MABs and also attain the information-theoretic bounds in some parameter regimes. Empirical simulations show that our algorithms significantly outperform existing LCB-based algorithms for all risk tolerances.

Qiuyu Zhu, Tiantian Li

Compared to supervised learning, semi-supervised learning reduces the dependence of deep learning on a large number of labeled samples. In this work, we use a small number of labeled samples and perform data augmentation on unlabeled samples to achieve image classification. Our method constrains all samples to the predefined evenly-distributed class centroids (PEDCC) by the corresponding loss function. Specifically, the PEDCC-Loss for labeled samples, and the maximum mean discrepancy loss for unlabeled samples are used to make the feature distribution closer to the distribution of PEDCC. Our method ensures that the inter-class distance is large and the intra-class distance is small enough to make the classification boundaries between different classes clearer. Meanwhile, for unlabeled samples, we also use KL divergence to constrain the consistency of the network predictions between unlabeled and augmented samples. Our semi-supervised learning method achieves the state-of-the-art results, with 4000 labeled samples on CIFAR10 and 1000 labeled samples on SVHN, and the accuracy is 95.10% and 97.58% respectively.

Qiuyu Zhu, Zikuang He, Xin Ye

The main purpose of incremental learning is to learn new knowledge while not forgetting the knowledge which have been learned before. At present, the main challenge in this area is the catastrophe forgetting, namely the network will lose their performance in the old tasks after training for new tasks. In this paper, we introduce an ensemble method of incremental classifier to alleviate this problem, which is based on the cosine distance between the output feature and the pre-defined center, and can let each task to be preserved in different networks. During training, we make use of PEDCC-Loss to train the CNN network. In the stage of testing, the prediction is determined by the cosine distance between the network latent features and pre-defined center. The experimental results on EMINST and CIFAR100 show that our method outperforms the recent LwF method, which use the knowledge distillation, and iCaRL method, which keep some old samples while training for new task. The method can achieve the goal of not forgetting old knowledge while training new classes, and solve the problem of catastrophic forgetting better.

Qiuyu Zhu, Ruixin Zhang

Classic variational autoencoders are used to learn complex data distributions, that are built on standard function approximators. Especially, VAE has shown promise on a lot of complex task. In this paper, a new autoencoder model - classification supervised autoencoder (CSAE) based on predefined evenly-distributed class centroids (PEDCC) is proposed. Our method uses PEDCC of latent variables to train the network to ensure the maximization of inter-class distance and the minimization of inner-class distance. Instead of learning mean/variance of latent variables distribution and taking reparameterization of VAE, latent variables of CSAE are directly used to classify and as input of decoder. In addition, a new loss function is proposed to combine the loss function of classification. Based on the basic structure of the universal autoencoder, we realized the comprehensive optimal results of encoding, decoding, classification, and good model generalization performance at the same time. Theoretical advantages are reflected in experimental results.