Bias issues of neural networks garner significant attention along with its promising advancement. Among various bias issues, mitigating two predominant biases is crucial in advancing fair and trustworthy AI: (1) ensuring neural networks yields even performance across demographic groups, and (2) ensuring algorithmic decision-making does not rely on protected attributes. However, upon the investigation of \pc papers in the relevant literature, we find that there exists a persistent, extensive but under-explored confusion regarding these two types of biases. Furthermore, the confusion has already significantly hampered the clarity of the community and subsequent development of debiasing methodologies. Thus, in this work, we aim to restore clarity by providing two mathematical definitions for these two predominant biases and leveraging these definitions to unify a comprehensive list of papers. Next, we highlight the common phenomena and the possible reasons for the existing confusion. To alleviate the confusion, we provide extensive experiments on synthetic, census, and image datasets, to validate the distinct nature of these biases, distinguish their different real-world manifestations, and evaluate the effectiveness of a comprehensive list of bias assessment metrics in assessing the mitigation of these biases. Further, we compare these two types of biases from multiple dimensions including the underlying causes, debiasing methods, evaluation protocol, prevalent datasets, and future directions. Last, we provide several suggestions aiming to guide researchers engaged in bias-related work to avoid confusion and further enhance clarity in the community.

Recent works have shown that generative models leave traces of their underlying generative process on the generated samples, broadly referred to as fingerprints of a generative model, and have studied their utility in detecting synthetic images from real ones. However, the extend to which these fingerprints can distinguish between various types of synthetic image and help identify the underlying generative process remain under-explored. In particular, the very definition of a fingerprint remains unclear, to our knowledge. To that end, in this work, we formalize the definition of artifact and fingerprint in generative models, propose an algorithm for computing them in practice, and finally study its effectiveness in distinguishing a large array of different generative models. We find that using our proposed definition can significantly improve the performance on the task of identifying the underlying generative process from samples (model attribution) compared to existing methods. Additionally, we study the structure of the fingerprints, and observe that it is very predictive of the effect of different design choices on the generative process.

Ensuring a neural network is not relying on protected attributes (e.g., race, sex, age) for predictions is crucial in advancing fair and trustworthy AI. While several promising methods for removing attribute bias in neural networks have been proposed, their limitations remain under-explored. In this work, we mathematically and empirically reveal an important limitation of attribute bias removal methods in presence of strong bias. Specifically, we derive a general non-vacuous information-theoretical upper bound on the performance of any attribute bias removal method in terms of the bias strength. We provide extensive experiments on synthetic, image, and census datasets to verify the theoretical bound and its consequences in practice. Our findings show that existing attribute bias removal methods are effective only when the inherent bias in the dataset is relatively weak, thus cautioning against the use of these methods in smaller datasets where strong attribute bias can occur, and advocating the need for methods that can overcome this limitation.

Ensuring a neural network is not relying on protected attributes (e.g., race, sex, age) for prediction is crucial in advancing fair and trustworthy AI. While several promising methods for removing attribute bias in neural networks have been proposed, their limitations remain under-explored. To that end, in this work, we mathematically and empirically reveal the limitation of existing attribute bias removal methods in presence of strong bias and propose a new method that can mitigate this limitation. Specifically, we first derive a general non-vacuous information-theoretical upper bound on the performance of any attribute bias removal method in terms of the bias strength, revealing that they are effective only when the inherent bias in the dataset is relatively weak. Next, we derive a necessary condition for the existence of any method that can remove attribute bias regardless of the bias strength. Inspired by this condition, we then propose a new method using an adversarial objective that directly filters out protected attributes in the input space while maximally preserving all other attributes, without requiring any specific target label. The proposed method achieves state-of-the-art performance in both strong and moderate bias settings. We provide extensive experiments on synthetic, image, and census datasets, to verify the derived theoretical bound and its consequences in practice, and evaluate the effectiveness of the proposed method in removing strong attribute bias.

Discovering causal relations among semantic factors is an emergent topic in representation learning. Most causal representation learning (CRL) methods are fully supervised, which is impractical due to costly labeling. To resolve this restriction, weakly supervised CRL methods were introduced. To evaluate CRL performance, four existing datasets, Pendulum, Flow, CelebA(BEARD) and CelebA(SMILE), are utilized. However, existing CRL datasets are limited to simple graphs with few generative factors. Thus we propose two new datasets with a larger number of diverse generative factors and more sophisticated causal graphs. In addition, current real datasets, CelebA(BEARD) and CelebA(SMILE), the originally proposed causal graphs are not aligned with the dataset distributions. Thus, we propose modifications to them.

Precision and Recall are two prominent metrics of generative performance, which were proposed to separately measure the fidelity and diversity of generative models. Given their central role in comparing and improving generative models, understanding their limitations are crucially important. To that end, in this work, we identify a critical flaw in the common approximation of these metrics using k-nearest-neighbors, namely, that the very interpretations of fidelity and diversity that are assigned to Precision and Recall can fail in high dimensions, resulting in very misleading conclusions. Specifically, we empirically and theoretically show that as the number of dimensions grows, two model distributions with supports at equal point-wise distance from the support of the real distribution, can have vastly different Precision and Recall regardless of their respective distributions, hence an emergent asymmetry in high dimensions. Based on our theoretical insights, we then provide simple yet effective modifications to these metrics to construct symmetric metrics regardless of the number of dimensions. Finally, we provide experiments on real-world datasets to illustrate that the identified flaw is not merely a pathological case, and that our proposed metrics are effective in alleviating its impact.

Due to data's unavailability or large size, and the high computational and human labor costs of training machine learning models, it is a common practice to rely on open source pre-trained models whenever possible. However, this practice is worry some from the security perspective. Pre-trained models can be infected with Trojan attacks, in which the attacker embeds a trigger in the model such that the model's behavior can be controlled by the attacker when the trigger is present in the input. In this paper, we present our preliminary work on a novel method for Trojan model detection. Our method creates a signature for a model based on activation optimization. A classifier is then trained to detect a Trojan model given its signature. Our method achieves state of the art performance on two public datasets.

Causal representation learning has been proposed to encode relationships between factors presented in the high dimensional data. However, existing methods suffer from merely using a large amount of labeled data and ignore the fact that samples generated by the same causal mechanism follow the same causal relationships. In this paper, we seek to explore such information by leveraging do-operation to reduce supervision strength. We propose a framework that implements do-operation by swapping latent cause and effect factors encoded from a pair of inputs. Moreover, we also identify the inadequacy of existing causal representation metrics empirically and theoretically and introduce new metrics for better evaluation. Experiments conducted on both synthetic and real datasets demonstrate the superiorities of our method compared with state-of-the-art methods.

In this paper, we introduce a novel non-linear activation function that spontaneously induces class-compactness and regularization in the embedding space of neural networks. The function is dubbed DOME for Difference Of Mirrored Exponential terms. The basic form of the function can replace the sigmoid or the hyperbolic tangent functions as an output activation function for binary classification problems. The function can also be extended to the case of multi-class classification, and used as an alternative to the standard softmax function. It can also be further generalized to take more flexible shapes suitable for intermediate layers of a network. In this version of the paper, we only introduce the concept. In a subsequent version, experimental evaluation will be added.

The ability to capture complex linguistic structures and long-term dependencies among words in the passage is essential for many natural language understanding tasks. In relation extraction, dependency trees that contain rich syntactic clues have been widely used to help capture long-term dependencies in text. Graph neural networks (GNNs), one of the means to encode dependency graphs, has been shown effective in several prior works. However, relatively little attention has been paid to the receptive fields of GNNs, which can be crucial in tasks with extremely long text that go beyond single sentences and require discourse analysis. In this work, we leverage the idea of graph pooling and propose the Mirror Graph Convolution Network (MrGCN), a GNN model with pooling-unpooling structures tailored to relation extraction. The pooling branch reduces the graph size and enables the GCN to obtain larger receptive fields within less layers; the unpooling branch restores the pooled graph to its original resolution such that token-level relation extraction can be performed. Experiments on two datasets demonstrate the effectiveness of our method, showing significant improvements over previous results.