Shahbaz Rezaei
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Shahbaz Rezaei has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
- Xin Liu
Department of Computer Science, University of California, Davis, CA 95616, USA
13 shared publications - Ali Mohammad Afshin Hemmatyar
Department of Computer Engineering, Sharif University of Technology, Tehran, Iran
3 shared publications - Ali Movaghar
Sharif University of Technology
3 shared publications - Mohammed Gharib
Institute for Research in Fundamental Sciences, Tehran, Iran
3 shared publications - Anpeng Huang
Peking University
2 shared publications - Guangyu Wang
State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing, China
2 shared publications
Latest Publications
- Open Access
On the Necessity of Multi-Domain Explanation: An Uncertainty Principle Approach for Deep Time Series Models
A prevailing approach to explain time series models is to generate attribution in time domain. A recent development in time series XAI is the concept of explanation spaces, where any model trained in the time domain can be interpreted with any existing XAI method in alternative domains, such as frequency. The prevailing approach is to present XAI attributions either in the time domain or in the domain where the attribution is most sparse. In this paper, we demonstrate that in certain cases, XAI methods can generate attributions that highlight fundamentally different features in the time and frequency domains that are not direct counterparts of one another. This suggests that both domains' attributions should be presented to achieve a more comprehensive interpretation. Thus it shows the necessity of multi-domain explanation. To quantify when such cases arise, we introduce the uncertainty principle (UP), originally developed in quantum mechanics and later studied in harmonic analysis and signal processing, to the XAI literature. This principle establishes a lower bound on how much a signal can be simultaneously localized in both the time and frequency domains. By leveraging this concept, we assess whether attributions in the time and frequency domains violate this bound, indicating that they emphasize distinct features. In other words, UP provides a sufficient condition that the time and frequency domain explanations do not match and, hence, should be both presented to the end user. We validate the effectiveness of this approach across various deep learning models, XAI methods, and a wide range of classification and forecasting datasets. The frequent occurrence of UP violations across various datasets and XAI methods highlights the limitations of existing approaches that focus solely on time-domain explanations. This underscores the need for multi-domain explanations as a new paradigm.
Open AccessOn the Necessity of Multi-Domain Explanation: An Uncertainty Principle Approach for Deep Time Series Models
A prevailing approach to explain time series models is to generate attribution in time domain. A recent development in time series XAI is the concept of explanation spaces, where any model trained in the time domain can be interpreted with any existing XAI method in alternative domains, such as frequency. The prevailing approach is to present XAI attributions either in the time domain or in the domain where the attribution is most sparse. In this paper, we demonstrate that in certain cases, XAI methods can generate attributions that highlight fundamentally different features in the time and frequency domains that are not direct counterparts of one another. This suggests that both domains' attributions should be presented to achieve a more comprehensive interpretation. Thus it shows the necessity of multi-domain explanation. To quantify when such cases arise, we introduce the uncertainty principle (UP), originally developed in quantum mechanics and later studied in harmonic analysis and signal processing, to the XAI literature. This principle establishes a lower bound on how much a signal can be simultaneously localized in both the time and frequency domains. By leveraging this concept, we assess whether attributions in the time and frequency domains violate this bound, indicating that they emphasize distinct features. In other words, UP provides a sufficient condition that the time and frequency domain explanations do not match and, hence, should be both presented to the end user. We validate the effectiveness of this approach across various deep learning models, XAI methods, and a wide range of classification and forecasting datasets. The frequent occurrence of UP violations across various datasets and XAI methods highlights the limitations of existing approaches that focus solely on time-domain explanations. This underscores the need for multi-domain explanations as a new paradigm.
- Open Access
Implet: A Post-hoc Subsequence Explainer for Time Series Models
Explainability in time series models is crucial for fostering trust, facilitating debugging, and ensuring interpretability in real-world applications. In this work, we introduce Implet, a novel post-hoc explainer that generates accurate and concise subsequence-level explanations for time series models. Our approach identifies critical temporal segments that significantly contribute to the model's predictions, providing enhanced interpretability beyond traditional feature-attribution methods. Based on it, we propose a cohort-based (group-level) explanation framework designed to further improve the conciseness and interpretability of our explanations. We evaluate Implet on several standard time-series classification benchmarks, demonstrating its effectiveness in improving interpretability. The code is available at https://github.com/LbzSteven/implet
Open AccessImplet: A Post-hoc Subsequence Explainer for Time Series Models
Explainability in time series models is crucial for fostering trust, facilitating debugging, and ensuring interpretability in real-world applications. In this work, we introduce Implet, a novel post-hoc explainer that generates accurate and concise subsequence-level explanations for time series models. Our approach identifies critical temporal segments that significantly contribute to the model's predictions, providing enhanced interpretability beyond traditional feature-attribution methods. Based on it, we propose a cohort-based (group-level) explanation framework designed to further improve the conciseness and interpretability of our explanations. We evaluate Implet on several standard time-series classification benchmarks, demonstrating its effectiveness in improving interpretability. The code is available at https://github.com/LbzSteven/implet
- Open Access
Dynamic Batch Norm Statistics Update for Natural Robustness
DNNs trained on natural clean samples have been shown to perform poorly on corrupted samples, such as noisy or blurry images. Various data augmentation methods have been recently proposed to improve DNN's robustness against common corruptions. Despite their success, they require computationally expensive training and cannot be applied to off-the-shelf trained models. Recently, it has been shown that updating BatchNorm (BN) statistics of an off-the-shelf model on a single corruption improves its accuracy on that corruption significantly. However, adopting the idea at inference time when the type of corruption is unknown and changing decreases the effectiveness of this method. In this paper, we harness the Fourier domain to detect the corruption type, a challenging task in the image domain. We propose a unified framework consisting of a corruption-detection model and BN statistics update that improves the corruption accuracy of any off-the-shelf trained model. We benchmark our framework on different models and datasets. Our results demonstrate about 8% and 4% accuracy improvement on CIFAR10-C and ImageNet-C, respectively. Furthermore, our framework can further improve the accuracy of state-of-the-art robust models, such as AugMix and DeepAug.
Open AccessDynamic Batch Norm Statistics Update for Natural Robustness
DNNs trained on natural clean samples have been shown to perform poorly on corrupted samples, such as noisy or blurry images. Various data augmentation methods have been recently proposed to improve DNN's robustness against common corruptions. Despite their success, they require computationally expensive training and cannot be applied to off-the-shelf trained models. Recently, it has been shown that updating BatchNorm (BN) statistics of an off-the-shelf model on a single corruption improves its accuracy on that corruption significantly. However, adopting the idea at inference time when the type of corruption is unknown and changing decreases the effectiveness of this method. In this paper, we harness the Fourier domain to detect the corruption type, a challenging task in the image domain. We propose a unified framework consisting of a corruption-detection model and BN statistics update that improves the corruption accuracy of any off-the-shelf trained model. We benchmark our framework on different models and datasets. Our results demonstrate about 8% and 4% accuracy improvement on CIFAR10-C and ImageNet-C, respectively. Furthermore, our framework can further improve the accuracy of state-of-the-art robust models, such as AugMix and DeepAug.
- Open Access
Uncovering the Gut–Liver Axis Biomarkers for Predicting Metabolic Burden in Mice
Western diet (WD) intake, aging, and inactivation of farnesoid X receptor (FXR) are risk factors for metabolic and chronic inflammation-related health issues ranging from metabolic dysfunction-associated steatotic liver disease (MASLD) to dementia. The progression of MASLD can be escalated when those risks are combined. Inactivation of FXR, the receptor for bile acid (BA), is cancer prone in both humans and mice. The current study used multi-omics including hepatic transcripts, liver, serum, and urine metabolites, hepatic BAs, as well as gut microbiota from mouse models to classify those risks using machine learning. A linear support vector machine with K-fold cross-validation was used for classification and feature selection. We have identified that increased urine sucrose alone achieved 91% accuracy in predicting WD intake. Hepatic lithocholic acid and serum pyruvate had 100% and 95% accuracy, respectively, to classify age. Urine metabolites (decreased creatinine and taurine as well as increased succinate) or increased gut bacteria (Dorea, Dehalobacterium, and Oscillospira) could predict FXR deactivation with greater than 90% accuracy. Human disease relevance is partly revealed using the metabolite–disease interaction network. Transcriptomics data were also compared with the human liver disease datasets. WD-reduced hepatic Cyp39a1 (cytochrome P450 family 39 subfamily a member 1) and increased Gramd1b (GRAM domain containing 1B) were also changed in human liver cancer and metabolic liver disease, respectively. Together, our data contribute to the identification of noninvasive biomarkers within the gut–liver axis to predict metabolic status.
Open AccessUncovering the Gut–Liver Axis Biomarkers for Predicting Metabolic Burden in Mice
Western diet (WD) intake, aging, and inactivation of farnesoid X receptor (FXR) are risk factors for metabolic and chronic inflammation-related health issues ranging from metabolic dysfunction-associated steatotic liver disease (MASLD) to dementia. The progression of MASLD can be escalated when those risks are combined. Inactivation of FXR, the receptor for bile acid (BA), is cancer prone in both humans and mice. The current study used multi-omics including hepatic transcripts, liver, serum, and urine metabolites, hepatic BAs, as well as gut microbiota from mouse models to classify those risks using machine learning. A linear support vector machine with K-fold cross-validation was used for classification and feature selection. We have identified that increased urine sucrose alone achieved 91% accuracy in predicting WD intake. Hepatic lithocholic acid and serum pyruvate had 100% and 95% accuracy, respectively, to classify age. Urine metabolites (decreased creatinine and taurine as well as increased succinate) or increased gut bacteria (Dorea, Dehalobacterium, and Oscillospira) could predict FXR deactivation with greater than 90% accuracy. Human disease relevance is partly revealed using the metabolite–disease interaction network. Transcriptomics data were also compared with the human liver disease datasets. WD-reduced hepatic Cyp39a1 (cytochrome P450 family 39 subfamily a member 1) and increased Gramd1b (GRAM domain containing 1B) were also changed in human liver cancer and metabolic liver disease, respectively. Together, our data contribute to the identification of noninvasive biomarkers within the gut–liver axis to predict metabolic status.
- Open Access
On the Discredibility of Membership Inference Attacks
With the wide-spread application of machine learning models, it has become critical to study the potential data leakage of models trained on sensitive data. Recently, various membership inference (MI) attacks are proposed to determine if a sample was part of the training set or not. The question is whether these attacks can be reliably used in practice. We show that MI models frequently misclassify neighboring nonmember samples of a member sample as members. In other words, they have a high false positive rate on the subpopulations of the exact member samples that they can identify. We then showcase a practical application of MI attacks where this issue has a real-world repercussion. Here, MI attacks are used by an external auditor (investigator) to show to a judge/jury that an auditee unlawfully used sensitive data. Due to the high false positive rate of MI attacks on member's subpopulations, auditee challenges the credibility of the auditor by revealing the performance of the MI attacks on these subpopulations. We argue that current membership inference attacks can identify memorized subpopulations, but they cannot reliably identify which exact sample in the subpopulation was used during the training.
Open AccessOn the Discredibility of Membership Inference Attacks
With the wide-spread application of machine learning models, it has become critical to study the potential data leakage of models trained on sensitive data. Recently, various membership inference (MI) attacks are proposed to determine if a sample was part of the training set or not. The question is whether these attacks can be reliably used in practice. We show that MI models frequently misclassify neighboring nonmember samples of a member sample as members. In other words, they have a high false positive rate on the subpopulations of the exact member samples that they can identify. We then showcase a practical application of MI attacks where this issue has a real-world repercussion. Here, MI attacks are used by an external auditor (investigator) to show to a judge/jury that an auditee unlawfully used sensitive data. Due to the high false positive rate of MI attacks on member's subpopulations, auditee challenges the credibility of the auditor by revealing the performance of the MI attacks on these subpopulations. We argue that current membership inference attacks can identify memorized subpopulations, but they cannot reliably identify which exact sample in the subpopulation was used during the training.
- Open Access
Accuracy-Privacy Trade-off in Deep Ensemble: A Membership Inference Perspective
Deep ensemble learning has been shown to improve accuracy by training multiple neural networks and averaging their outputs. Ensemble learning has also been suggested to defend against membership inference attacks that undermine privacy. In this paper, we empirically demonstrate a trade-off between these two goals, namely accuracy and privacy (in terms of membership inference attacks), in deep ensembles. Using a wide range of datasets and model architectures, we show that the effectiveness of membership inference attacks increases when ensembling improves accuracy. We analyze the impact of various factors in deep ensembles and demonstrate the root cause of the trade-off. Then, we evaluate common defenses against membership inference attacks based on regularization and differential privacy. We show that while these defenses can mitigate the effectiveness of membership inference attacks, they simultaneously degrade ensemble accuracy. We illustrate similar trade-off in more advanced and state-of-the-art ensembling techniques, such as snapshot ensembles and diversified ensemble networks. Finally, we propose a simple yet effective defense for deep ensembles to break the trade-off and, consequently, improve the accuracy and privacy, simultaneously.
Open AccessAccuracy-Privacy Trade-off in Deep Ensemble: A Membership Inference Perspective
Deep ensemble learning has been shown to improve accuracy by training multiple neural networks and averaging their outputs. Ensemble learning has also been suggested to defend against membership inference attacks that undermine privacy. In this paper, we empirically demonstrate a trade-off between these two goals, namely accuracy and privacy (in terms of membership inference attacks), in deep ensembles. Using a wide range of datasets and model architectures, we show that the effectiveness of membership inference attacks increases when ensembling improves accuracy. We analyze the impact of various factors in deep ensembles and demonstrate the root cause of the trade-off. Then, we evaluate common defenses against membership inference attacks based on regularization and differential privacy. We show that while these defenses can mitigate the effectiveness of membership inference attacks, they simultaneously degrade ensemble accuracy. We illustrate similar trade-off in more advanced and state-of-the-art ensembling techniques, such as snapshot ensembles and diversified ensemble networks. Finally, we propose a simple yet effective defense for deep ensembles to break the trade-off and, consequently, improve the accuracy and privacy, simultaneously.
- Open Access
User-Level Membership Inference Attack against Metric Embedding Learning
Membership inference (MI) determines if a sample was part of a victim model training set. Recent development of MI attacks focus on record-level membership inference which limits their application in many real-world scenarios. For example, in the person re-identification task, the attacker (or investigator) is interested in determining if a user's images have been used during training or not. However, the exact training images might not be accessible to the attacker. In this paper, we develop a user-level MI attack where the goal is to find if any sample from the target user has been used during training even when no exact training sample is available to the attacker. We focus on metric embedding learning due to its dominance in person re-identification, where user-level MI attack is more sensible. We conduct an extensive evaluation on several datasets and show that our approach achieves high accuracy on user-level MI task.
Open AccessUser-Level Membership Inference Attack against Metric Embedding Learning
Membership inference (MI) determines if a sample was part of a victim model training set. Recent development of MI attacks focus on record-level membership inference which limits their application in many real-world scenarios. For example, in the person re-identification task, the attacker (or investigator) is interested in determining if a user's images have been used during training or not. However, the exact training images might not be accessible to the attacker. In this paper, we develop a user-level MI attack where the goal is to find if any sample from the target user has been used during training even when no exact training sample is available to the attacker. We focus on metric embedding learning due to its dominance in person re-identification, where user-level MI attack is more sensible. We conduct an extensive evaluation on several datasets and show that our approach achieves high accuracy on user-level MI task.
- Open Access
On the Difficulty of Membership Inference Attacks
Recent studies propose membership inference (MI) attacks on deep models, where the goal is to infer if a sample has been used in the training process. Despite their apparent success, these studies only report accuracy, precision, and recall of the positive class (member class). Hence, the performance of these attacks have not been clearly reported on negative class (non-member class). In this paper, we show that the way the MI attack performance has been reported is often misleading because they suffer from high false positive rate or false alarm rate (FAR) that has not been reported. FAR shows how often the attack model mislabel non-training samples (non-member) as training (member) ones. The high FAR makes MI attacks fundamentally impractical, which is particularly more significant for tasks such as membership inference where the majority of samples in reality belong to the negative (non-training) class. Moreover, we show that the current MI attack models can only identify the membership of misclassified samples with mediocre accuracy at best, which only constitute a very small portion of training samples. We analyze several new features that have not been comprehensively explored for membership inference before, including distance to the decision boundary and gradient norms, and conclude that deep models' responses are mostly similar among train and non-train samples. We conduct several experiments on image classification tasks, including MNIST, CIFAR-10, CIFAR-100, and ImageNet, using various model architecture, including LeNet, AlexNet, ResNet, etc. We show that the current state-of-the-art MI attacks cannot achieve high accuracy and low FAR at the same time, even when the attacker is given several advantages. The source code is available at https://github.com/shrezaei/MI-Attack.
Open AccessOn the Difficulty of Membership Inference Attacks
Recent studies propose membership inference (MI) attacks on deep models, where the goal is to infer if a sample has been used in the training process. Despite their apparent success, these studies only report accuracy, precision, and recall of the positive class (member class). Hence, the performance of these attacks have not been clearly reported on negative class (non-member class). In this paper, we show that the way the MI attack performance has been reported is often misleading because they suffer from high false positive rate or false alarm rate (FAR) that has not been reported. FAR shows how often the attack model mislabel non-training samples (non-member) as training (member) ones. The high FAR makes MI attacks fundamentally impractical, which is particularly more significant for tasks such as membership inference where the majority of samples in reality belong to the negative (non-training) class. Moreover, we show that the current MI attack models can only identify the membership of misclassified samples with mediocre accuracy at best, which only constitute a very small portion of training samples. We analyze several new features that have not been comprehensively explored for membership inference before, including distance to the decision boundary and gradient norms, and conclude that deep models' responses are mostly similar among train and non-train samples. We conduct several experiments on image classification tasks, including MNIST, CIFAR-10, CIFAR-100, and ImageNet, using various model architecture, including LeNet, AlexNet, ResNet, etc. We show that the current state-of-the-art MI attacks cannot achieve high accuracy and low FAR at the same time, even when the attacker is given several advantages. The source code is available at https://github.com/shrezaei/MI-Attack.
- 29 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 8 (1) , 662-679
Modeling and Evaluation of Multi-Hop Wireless Networks Using SRNs
As multi-hop wireless networks are attracting more attention, the need to evaluate their performance becomes essential. In order to evaluate the performance metrics of multi-hop wireless networks, including sending and receiving rates of a node as well as the collision probability, a model based on Stochastic Reward Nets (SRNs) is proposed. The proposed SRN models a typical node in such networks, considered as a general template to be applied to any wireless node. The SRN model of a single node is designed to take transmission effects of all neighboring nodes into account, while ignoring the ones whose transmission has no effect on the node under-study. Applying the proposed SRN to each node in a multi-hop wireless network, we can model any arbitrary topology and evaluate the mean queue size, collision probability, sending and receiving rates of the nodes. Common issues related to Medium Access Control and physical layer of wireless networks, e.g. hidden and exposed node problems, collision, transmission, interference and carrier sense ranges, are considered in the proposed SRN. The results obtained by the proposed model are validated with the results obtained from the discrete-event simulation conducted in ns-2 to show the applicability and accuracy of the proposed model.
- 29 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Network Science and Engineering
- Vol. 8 (1) , 662-679
Modeling and Evaluation of Multi-Hop Wireless Networks Using SRNs
As multi-hop wireless networks are attracting more attention, the need to evaluate their performance becomes essential. In order to evaluate the performance metrics of multi-hop wireless networks, including sending and receiving rates of a node as well as the collision probability, a model based on Stochastic Reward Nets (SRNs) is proposed. The proposed SRN models a typical node in such networks, considered as a general template to be applied to any wireless node. The SRN model of a single node is designed to take transmission effects of all neighboring nodes into account, while ignoring the ones whose transmission has no effect on the node under-study. Applying the proposed SRN to each node in a multi-hop wireless network, we can model any arbitrary topology and evaluate the mean queue size, collision probability, sending and receiving rates of the nodes. Common issues related to Medium Access Control and physical layer of wireless networks, e.g. hidden and exposed node problems, collision, transmission, interference and carrier sense ranges, are considered in the proposed SRN. The results obtained by the proposed model are validated with the results obtained from the discrete-event simulation conducted in ns-2 to show the applicability and accuracy of the proposed model.
- Open Access
How to Achieve High Classification Accuracy with Just a Few Labels: A Semi-supervised Approach Using Sampled Packets
Network traffic classification, which has numerous applications from security to billing and network provisioning, has become a cornerstone of today's computer networks. Previous studies have developed traffic classification techniques using classical machine learning algorithms and deep learning methods when large quantities of labeled data are available. However, capturing large labeled datasets is a cumbersome and time-consuming process. In this paper, we propose a semi-supervised approach that obviates the need for large labeled datasets. We first pre-train a model on a large unlabeled dataset where the input is the time series features of a few sampled packets. Then the learned weights are transferred to a new model that is re-trained on a small labeled dataset. We show that our semi-supervised approach achieves almost the same accuracy as a fully-supervised method with a large labeled dataset, though we use only 20 samples per class. In tests based on a dataset generated from the more challenging QUIC protocol, our approach yields 98% accuracy. To show its efficacy, we also test our approach on two public datasets. Moreover, we study three different sampling techniques and demonstrate that sampling packets from an arbitrary portion of a flow is sufficient for classification.
Open AccessHow to Achieve High Classification Accuracy with Just a Few Labels: A Semi-supervised Approach Using Sampled Packets
Network traffic classification, which has numerous applications from security to billing and network provisioning, has become a cornerstone of today's computer networks. Previous studies have developed traffic classification techniques using classical machine learning algorithms and deep learning methods when large quantities of labeled data are available. However, capturing large labeled datasets is a cumbersome and time-consuming process. In this paper, we propose a semi-supervised approach that obviates the need for large labeled datasets. We first pre-train a model on a large unlabeled dataset where the input is the time series features of a few sampled packets. Then the learned weights are transferred to a new model that is re-trained on a small labeled dataset. We show that our semi-supervised approach achieves almost the same accuracy as a fully-supervised method with a large labeled dataset, though we use only 20 samples per class. In tests based on a dataset generated from the more challenging QUIC protocol, our approach yields 98% accuracy. To show its efficacy, we also test our approach on two public datasets. Moreover, we study three different sampling techniques and demonstrate that sampling packets from an arbitrary portion of a flow is sufficient for classification.