Högskolan i Skövde

Federated learning (FL) has emerged as a powerful solution for collaborative model training in domains with strict data privacy requirements, such as medical imaging. However, FL remains vulnerable to data poisoning attacks, which can significantly compromise the integrity of the global model. This study investigates the impact of two representative poisoning strategies—label flipping and backdoor injection—within an FL setup using a convolutional neural network trained on chest X-ray images for pneumonia detection. Our experimental results reveal that both attacks can severely degrade the global model’s performance, either by reducing classification accuracy or embedding hidden misclassification behaviors triggered during inference. To address these vulnerabilities, we propose an adaptive aggregation strategy that assigns weights to client updates based on their performance on a clean validation set. This approach enhances robustness against poisoning without requiring modifications on the client side. Experimental results demonstrate that the proposed defense effectively mitigates the impact of both attack types, maintaining high accuracy on clean data while minimizing the influence of poisoned updates. These findings highlight the urgent need for integrated security measures in FL systems, particularly in high-stakes applications such as clinical diagnostics. 

Accurate prediction of cutting forces is critical in milling operations, with implications for cost reduction and improved manufacturing efficiency. While traditional mechanistic models provide high accuracy, their reliance on extensive milling data for force coefficient fitting poses challenges. The eXtreme Gradient Boosting algorithm offers a potential solution with reduced data requirements, yet the optimal utilization of eXtreme Gradient Boosting remains unexplored. This study investigates its effectiveness in predicting cutting forces during down-milling of Al2024. A novel framework is proposed optimizing its precision, efficiency, and user-friendliness. The model training incorporates the mechanistic force model in both time and frequency domains as new features. Through rigorous experimentation, various aspects of the eXtreme Gradient Boosting configuration are explored, including identifying the optimal number of periods for the training dataset, determining the best normalization and scaling technique, and assessing the hyperparameters’ impact on model performance in terms of accuracy and computational time. The results show the remarkable effectiveness of the eXtreme Gradient Boosting model with an average normalized root mean square error of 14.7%, surpassing the 21.9% obtained by the mechanistic force model. Additionally, the machine learning model could capture the runout effect. These findings enable optimized milling operations regarding cost, accuracy and computation time.

Leukemia is a heterogeneous group of hematologic malignancies, with acute lymphoblastic leukemia (ALL) being one of the most harmful forms. Accurate and early diagnosis is crucial for effective treatment, potentially saving lives. Recent advances in machine learning (ML) and deep learning (DL) have significantly enhanced diagnostic capabilities. However, these advancements often compromise the confidentiality of sensitive medical data. In this paper, we propose a federated learning (FL) framework for the binary classification of ALL versus normal cases. This framework leverages decentralized data from multiple clients, where each client trains its model locally on its own data, transmitting only model updates to a central server. The central server then aggregates these updates using the FedAvg algorithm, creating a global model while ensuring that patient data remains at its source, thereby preserving confidentiality. Using an EfficientNetV2S-based model architecture and a dataset of 10,661 images containing normal cells and lymphoblasts, our experiments demonstrate that the proposed FL approach achieves an accuracy of 95.6% and a kappa coefficient of 0.89. This performance is competitive with centralized methods while maintaining data privacy. These results highlight the potential of FL to revolutionize the clinical detection of acute lymphoblastic leukemia, offering a scalable and privacy-preserving solution for medical applications.

Brain tumors pose a significant global health challenge, driving ongoing research advancements in early detection methods. Artificial intelligence (AI) and deep learning (DL) techniques have shown great potential in this field, enabling the creation of highly accurate models for brain tumor identification from medical images. However, centralized approaches to these methods often raise critical concerns regarding patient data privacy and security. This paper presents a novel federated learning (FL) framework for brain tumor identification that effectively addresses these privacy concerns. FL enables collaborative model training across multiple institutions without the need for raw data sharing. Each participating institution trains the model locally on their Magnetic Resonance Imaging (MRI) datasets and only transmits model updates to a central server for secure aggregation. This iterative process results in a robust global model trained on a distributed dataset while preserving patient data confidentiality. The proposed FL model is evaluated using a dataset of 3,000 MRI images. Experimental results demonstrate the effectiveness of our approach, achieving a high accuracy rate of 96.88% for brain tumor identification. These findings suggest that FL provides a viable solution for privacy-preserving brain tumor identification, maintaining comparable performance to centralized models while ensuring the security of patient data.

Federated learning (FL) enables edge nodes to collaboratively contribute to constructing a global model without sharing their data. This is accomplished by devices computing local, private model updates that are then aggregated by a server. However, computational resource constraints and network communication can become a severe bottleneck for larger model sizes typical for deep learning (DL) applications. Edge nodes tend to have limited hardware resources (RAM, CPU), and the network bandwidth and reliability at the edge is a concern for scaling federated fleet applications. In this paper, we propose and evaluate a FL strategy inspired by transfer learning in order to reduce resource utilization on devices, as well as the load on the server and network in each global training round. For each local model update, we randomly select layers to train, freezing the remaining part of the model. In doing so, we can reduce both server load and communication costs per round by excluding all untrained layer weights from being transferred to the server. The goal of this study is to empirically explore the potential trade-off between resource utilization on devices and global model convergence under the proposed strategy. We implement the approach using the FL framework FEDn. A number of experiments were carried out over different datasets (CIFAR-10, CASA, and IMDB), performing different tasks using different DL model architectures. Our results show that training the model partially can accelerate the training process, efficiently utilizes resources on-device, and reduce the data transmission by around 75% and 53% when we train 25%, and 50% of the model layers, respectively, without harming the resulting global model accuracy. Furthermore, our results demonstrate a negative correlation between the number of participating clients in the training process and the number of layers that need to be trained on each client’s side. As the number of clients increases, there is a decrease in the required number of layers. This observation highlights the potential of the approach, particularly in cross-device use cases.

Chatbots are mainly data-driven and usually based on utterances that might be sensitive. However, training deep learning models on shared data can violate user privacy. Such issues have commonly existed in chatbots since their inception. In the literature, there have been many approaches to deal with privacy, such as differential privacy and secure multi-party computation, but most of them need to have access to users' data. In this context, Federated Learning (FL) aims to protect data privacy through distributed learning methods that keep the data in its location. This paper presents Fedbot, a proof-of-concept (POC) privacy-preserving chatbot that leverages large-scale customer support data. The POC combines Deep Bidirectional Transformer models and federated learning algorithms to protect customer data privacy during collaborative model training. The results of the proof-of-concept showcase the potential for privacy-preserving chatbots to transform the customer support industry by delivering personalized and efficient customer service that meets data privacy regulations and legal requirements. Furthermore, the system is specifically designed to improve its performance and accuracy over time by leveraging its ability to learn from previous interactions.