With the rapid adoption of Federated Learning (FL) as the training and tuning protocol for applications utilizing Large Language Models (LLMs), recent research highlights the need for significant modifications to FL to accommodate the large-scale of LLMs. While substantial adjustments to the protocol have been introduced as a response, comprehensive privacy analysis for the adapted FL protocol is currently lacking. To address this gap, our work delves into an extensive examination of the privacy analysis of FL when used for training LLMs, both from theoretical and practical perspectives. In particular, we design two active membership inference attacks with guaranteed theoretical success rates to assess the privacy leakages of various adapted FL configurations. Our theoretical findings are translated into practical attacks, revealing substantial privacy vulnerabilities in popular LLMs, including BERT, RoBERTa, DistilBERT, and OpenAI's GPTs, across multiple real-world language datasets. Additionally, we conduct thorough experiments to evaluate the privacy leakage of these models when data is protected by state-of-the-art differential privacy (DP) mechanisms.
Federated Learning (FL) has garnered significant attention for its potential to protect user privacy while enhancing model training efficiency. However, recent research has demonstrated that FL protocols can be easily compromised by active reconstruction attacks executed by dishonest servers. These attacks involve the malicious modification of global model parameters, allowing the server to obtain a verbatim copy of users' private data by inverting their gradient updates. Tackling this class of attack remains a crucial challenge due to the strong threat model. In this paper, we propose OASIS, a defense mechanism based on image augmentation that effectively counteracts active reconstruction attacks while preserving model performance. We first uncover the core principle of gradient inversion that enables these attacks and theoretically identify the main conditions by which the defense can be robust regardless of the attack strategies. We then construct OASIS with image augmentation showing that it can undermine the attack principle. Comprehensive evaluations demonstrate the efficacy of OASIS highlighting its feasibility as a solution.
Federated learning (FL) was originally regarded as a framework for collaborative learning among clients with data privacy protection through a coordinating server. In this paper, we propose a new active membership inference (AMI) attack carried out by a dishonest server in FL. In AMI attacks, the server crafts and embeds malicious parameters into global models to effectively infer whether a target data sample is included in a client's private training data or not. By exploiting the correlation among data features through a non-linear decision boundary, AMI attacks with a certified guarantee of success can achieve severely high success rates under rigorous local differential privacy (LDP) protection; thereby exposing clients' training data to significant privacy risk. Theoretical and experimental results on several benchmark datasets show that adding sufficient privacy-preserving noise to prevent our attack would significantly damage FL's model utility.
Recent development in the field of explainable artificial intelligence (XAI) has helped improve trust in Machine-Learning-as-a-Service (MLaaS) systems, in which an explanation is provided together with the model prediction in response to each query. However, XAI also opens a door for adversaries to gain insights into the black-box models in MLaaS, thereby making the models more vulnerable to several attacks. For example, feature-based explanations (e.g., SHAP) could expose the top important features that a black-box model focuses on. Such disclosure has been exploited to craft effective backdoor triggers against malware classifiers. To address this trade-off, we introduce a new concept of achieving local differential privacy (LDP) in the explanations, and from that we establish a defense, called XRand, against such attacks. We show that our mechanism restricts the information that the adversary can learn about the top important features, while maintaining the faithfulness of the explanations.
Despite the great potential of Federated Learning (FL) in large-scale distributed learning, the current system is still subject to several privacy issues due to the fact that local models trained by clients are exposed to the central server. Consequently, secure aggregation protocols for FL have been developed to conceal the local models from the server. However, we show that, by manipulating the client selection process, the server can circumvent the secure aggregation to learn the local models of a victim client, indicating that secure aggregation alone is inadequate for privacy protection. To tackle this issue, we leverage blockchain technology to propose a verifiable client selection protocol. Owing to the immutability and transparency of blockchain, our proposed protocol enforces a random selection of clients, making the server unable to control the selection process at its discretion. We present security proofs showing that our protocol is secure against this attack. Additionally, we conduct several experiments on an Ethereum-like blockchain to demonstrate the feasibility and practicality of our solution.
Federated learning is known to be vulnerable to security and privacy issues. Existing research has focused either on preventing poisoning attacks from users or on protecting user privacy of model updates. However, integrating these two lines of research remains a crucial challenge since they often conflict with one another with respect to the threat model. In this work, we develop a framework to combine secure aggregation with defense mechanisms against poisoning attacks from users, while maintaining their respective privacy guarantees. We leverage zero-knowledge proof protocol to let users run the defense mechanisms locally and attest the result to the central server without revealing any information about their model updates. Furthermore, we propose a new secure aggregation protocol for federated learning using homomorphic encryption that is robust against malicious users. Our framework enables the central server to identify poisoned model updates without violating the privacy guarantees of secure aggregation. Finally, we analyze the computation and communication complexity of our proposed solution and benchmark its performance.
In this paper, we use pre-trained ResNet models as backbone architectures for classification of adventitious lung sounds and respiratory diseases. The knowledge of the pre-trained model is transferred by using vanilla fine-tuning, co-tuning, stochastic normalization and the combination of the co-tuning and stochastic normalization techniques. Furthermore, data augmentation in both time domain and time-frequency domain is used to account for the class imbalance of the ICBHI and our multi-channel lung sound dataset. Additionally, we apply spectrum correction to consider the variations of the recording device properties on the ICBHI dataset. Empirically, our proposed systems mostly outperform all state-of-the-art lung sound classification systems for the adventitious lung sounds and respiratory diseases of both datasets.
Large annotated lung sound databases are publicly available and might be used to train algorithms for diagnosis systems. However, it might be a challenge to develop a well-performing algorithm for small non-public data, which have only a few subjects and show differences in recording devices and setup. In this paper, we use transfer learning to tackle the mismatch of the recording setup. This allows us to transfer knowledge from one dataset to another dataset for crackle detection in lung sounds. In particular, a single input convolutional neural network (CNN) model is pre-trained on a source domain using ICBHI 2017, the largest publicly available database of lung sounds. We use log-mel spectrogram features of respiratory cycles of lung sounds. The pre-trained network is used to build a multi-input CNN model, which shares the same network architecture for respiratory cycles and their corresponding respiratory phases. The multi-input model is then fine-tuned on the target domain of our self-collected lung sound database for classifying crackles and normal lung sounds. Our experimental results show significant performance improvements of 9.84% (absolute) in F-score on the target domain using the multi-input CNN model based on transfer learning for crackle detection in adventitious lung sound classification task.
This paper presents a robust deep learning framework developed to detect respiratory diseases from recordings of respiratory sounds. The complete detection process firstly involves front end feature extraction where recordings are transformed into spectrograms that convey both spectral and temporal information. Then a back-end deep learning model classifies the features into classes of respiratory disease or anomaly. Experiments, conducted over the ICBHI benchmark dataset of respiratory sounds, evaluate the ability of the framework to classify sounds. Two main contributions are made in this paper. Firstly, we provide an extensive analysis of how factors such as respiratory cycle length, time resolution, and network architecture, affect final prediction accuracy. Secondly, a novel deep learning based framework is proposed for detection of respiratory diseases and shown to perform extremely well compared to state of the art methods.