Minimum Excess Risk (MER) in Bayesian learning is defined as the difference between the minimum expected loss achievable when learning from data and the minimum expected loss that could be achieved if the underlying parameter $W$ was observed. In this paper, we build upon and extend the recent results of (Xu & Raginsky, 2020) to analyze the MER in Bayesian learning and derive information-theoretic bounds on it. We formulate the problem as a (constrained) rate-distortion optimization and show how the solution can be bounded above and below by two other rate-distortion functions that are easier to study. The lower bound represents the minimum possible excess risk achievable by \emph{any} process using $R$ bits of information from the parameter $W$. For the upper bound, the optimization is further constrained to use $R$ bits from the training set, a setting which relates MER to information-theoretic bounds on the generalization gap in frequentist learning. We derive information-theoretic bounds on the difference between these upper and lower bounds and show that they can provide order-wise tight rates for MER. This analysis gives more insight into the information-theoretic nature of Bayesian learning as well as providing novel bounds.
Disease prediction is a well-known classification problem in medical applications. Graph neural networks provide a powerful tool for analyzing the patients' features relative to each other. Recently, Graph Convolutional Networks (GCNs) have particularly been studied in the field of disease prediction. Due to the nature of such medical datasets, the class imbalance is a familiar issue in the field of disease prediction. When the class imbalance is present in the data, the existing graph-based classifiers tend to be biased towards the major class(es). Meanwhile, the correct diagnosis of the rare true-positive cases among all the patients is vital. In conventional methods, such imbalance is tackled by assigning appropriate weights to classes in the loss function; however, this solution is still dependent on the relative values of weights, sensitive to outliers, and in some cases biased towards the minor class(es). In this paper, we propose Re-weighted Adversarial Graph Convolutional Network (RA-GCN) to enhance the performance of the graph-based classifier and prevent it from emphasizing the samples of any particular class. This is accomplished by automatically learning to weigh the samples of the classes. For this purpose, a graph-based network is associated with each class, which is responsible for weighing the class samples and informing the classifier about the importance of each sample. Therefore, the classifier adjusts itself and determines the boundary between classes with more attention to the important samples. The parameters of the classifier and weighing networks are trained by an adversarial approach. At the end of the adversarial training process, the boundary of the classifier is more accurate and unbiased. We show the superiority of RA-GCN on synthetic and three publicly available medical datasets compared to the recent method.
In recent years, extending variational autoencoder's framework to learn disentangled representations has received much attention. We address this problem by proposing a framework capable of disentangling class-related and class-independent factors of variation in data. Our framework employs an attention mechanism in its latent space in order to improve the process of extracting class-related factors from data. We also deal with the multimodality of data distribution by utilizing mixture models as learnable prior distributions, as well as incorporating the Bhattacharyya coefficient in the objective function to prevent highly overlapping mixtures. Our model's encoder is further trained in a semi-supervised manner, with a small fraction of labeled data, to improve representations' interpretability. Experiments show that our framework disentangles class-related and class-independent factors of variation and learns interpretable features. Moreover, we demonstrate our model's performance with quantitative and qualitative results on various datasets.
Deep generative models have achieved great success in areas such as image, speech, and natural language processing in the past few years. Thanks to the advances in graph-based deep learning, and in particular graph representation learning, deep graph generation methods have recently emerged with new applications ranging from discovering novel molecular structures to modeling social networks. This paper conducts a comprehensive survey on deep learning-based graph generation approaches and classifies them into five broad categories, namely, autoregressive, autoencoder-based, RL-based, adversarial, and flow-based graph generators, providing the readers a detailed description of the methods in each class. We also present publicly available source codes, commonly used datasets, and the most widely utilized evaluation metrics. Finally, we highlight the existing challenges and discuss future research directions.
Although GAN-based methods have received many achievements in the last few years, they have not been such successful in generating discrete data. The most important challenge of these methods is the difficulty of passing the gradient from the discriminator to the generator when the generator outputs are discrete. Despite several attempts done to alleviate this problem, none of the existing GAN-based methods has improved the performance of text generation (using measures that evaluate both the quality and the diversity of generated samples) compared to a generative RNN that is simply trained by the maximum likelihood approach. In this paper, we propose a new framework for generating discrete data by an adversarial approach in which we do not need to pass the gradient to the generator. In the proposed method, the update of either the generator or the discriminator can be accomplished straightforwardly. Moreover, we leverage the discreteness of data to explicitly model the data distribution and ensure the normalization of the generated distribution and consequently the convergence properties of the proposed method. Experimental results generally show the superiority of the proposed DGSAN method compared to the other GAN-based approaches for generating discrete sequential data.
Text generation is an important Natural Language Processing task with various applications. Although several metrics have already been introduced to evaluate the text generation methods, each of them has its own shortcomings. The most widely used metrics such as BLEU only consider the quality of generated sentences and neglect their diversity. For example, repeatedly generation of only one high quality sentence would result in a high BLEU score. On the other hand, the more recent metric introduced to evaluate the diversity of generated texts known as Self-BLEU ignores the quality of generated texts. In this paper, we propose metrics to evaluate both the quality and diversity simultaneously by approximating the distance of the learned generative model and the real data distribution. For this purpose, we first introduce a metric that approximates this distance using n-gram based measures. Then, a feature-based measure which is based on a recent highly deep model trained on a large text corpus called BERT is introduced. Finally, for oracle training mode in which the generator's density can also be calculated, we propose to use the distance measures between the corresponding explicit distributions. Eventually, the most popular and recent text generation models are evaluated using both the existing and the proposed metrics and the preferences of the proposed metrics are determined.
Recently, graph embedding emerges as an effective approach for graph analysis tasks such as node classification and link prediction. The goal of network embedding is to find low dimensional representation of graph nodes that preserves the graph structure. Since there might be signals on nodes as features, recent methods like Graph Convolutional Networks (GCNs) try to consider node signals besides the node relations. On the other hand, multi-layered graph analysis has been received much attention. However, the recent methods for node embedding have not been explored in these networks. In this paper, we study the problem of node embedding in multi-layered graphs and propose a deep method that embeds nodes using both relations (connections within and between layers of the graph) and nodes signals. We evaluate our method on node classification tasks. Experimental results demonstrate the superiority of the proposed method to other multi-layered and single-layered competitors and also proves the effect of using cross-layer edges.
Adversarial approach has been widely used for data generation in the last few years. However, this approach has not been extensively utilized for classifier training. In this paper, we propose an adversarial framework for classifier training that can also handle imbalanced data. Indeed, a network is trained via an adversarial approach to give weights to samples of the majority class such that the obtained classification problem becomes more challenging for the discriminator and thus boosts its classification capability. In addition to the general imbalanced classification problems, the proposed method can also be used for problems such as graph representation learning in which it is desired to discriminate similar nodes from dissimilar nodes. Experimental results on imbalanced data classification and on the tasks like graph link prediction show the superiority of the proposed method compared to the state-of-the-art methods.
Multi-label classification has received considerable interest in recent years. Multi-label classifiers have to address many problems including: handling large-scale datasets with many instances and a large set of labels, compensating missing label assignments in the training set, considering correlations between labels, as well as exploiting unlabeled data to improve prediction performance. To tackle datasets with a large set of labels, embedding-based methods have been proposed which seek to represent the label assignments in a low-dimensional space. Many state-of-the-art embedding-based methods use a linear dimensionality reduction to represent the label assignments in a low-dimensional space. However, by doing so, these methods actually neglect the tail labels - labels that are infrequently assigned to instances. We propose an embedding-based method that non-linearly embeds the label vectors using an stochastic approach, thereby predicting the tail labels more accurately. Moreover, the proposed method have excellent mechanisms for handling missing labels, dealing with large-scale datasets, as well as exploiting unlabeled data. With the best of our knowledge, our proposed method is the first multi-label classifier that simultaneously addresses all of the mentioned challenges. Experiments on real-world datasets show that our method outperforms stateof-the-art multi-label classifiers by a large margin, in terms of prediction performance, as well as training time.
In some of object recognition problems, labeled data may not be available for all categories. Zero-shot learning utilizes auxiliary information (also called signatures) describing each category in order to find a classifier that can recognize samples from categories with no labeled instance. In this paper, we propose a novel semi-supervised zero-shot learning method that works on an embedding space corresponding to abstract deep visual features. We seek a linear transformation on signatures to map them onto the visual features, such that the mapped signatures of the seen classes are close to labeled samples of the corresponding classes and unlabeled data are also close to the mapped signatures of one of the unseen classes. We use the idea that the rich deep visual features provide a representation space in which samples of each class are usually condensed in a cluster. The effectiveness of the proposed method is demonstrated through extensive experiments on four public benchmarks improving the state-of-the-art prediction accuracy on three of them.