Predicting miRNA-Disease Association Based on Neural Inductive Matrix Completion with Graph Autoencoders and Self-Attention Mechanism

Many studies have clarified that microRNAs (miRNAs) are associated with many human diseases. Therefore, it is essential to predict potential miRNA-disease associations for disease pathogenesis and treatment. Numerous machine learning and deep learning approaches have been adopted to this problem. In this paper, we propose a Neural Inductive Matrix completion-based method with Graph Autoencoders (GAE) and Self-Attention mechanism for miRNA-disease associations prediction (NIMGSA). Some of the previous works based on matrix completion ignore the importance of label propagation procedure for inferring miRNA-disease associations, while others cannot integrate matrix completion and label propagation effectively. Varying from previous studies, NIMGSA unifies inductive matrix completion and label propagation via neural network architecture, through the collaborative training of two graph autoencoders. This neural inductive matrix completion-based method is also an implementation of self-attention mechanism for miRNA-disease associations prediction. This end-to-end framework can strengthen the robustness and preciseness of both matrix completion and label propagation. Cross validations indicate that NIMGSA outperforms current miRNA-disease prediction methods. Case studies demonstrate that NIMGSA is competent in detecting potential miRNA-disease associations.

Towards Understanding the Risks of Gradient Inversion in Federated Learning

Federated learning (FL) allows the collaborative training of AI models without needing to share raw data. This capability makes it especially interesting for healthcare applications where patient and data privacy is of utmost concern. However, recent works on the inversion of deep neural networks from model gradients raised concerns about the security of FL in preventing the leakage of training data. In this work, we show that these attacks presented in the literature are impractical in real FL use-cases and provide a new baseline attack that works for more realistic scenarios where the clients’ training involves updating the Batch Normalization (BN) statistics. Furthermore, we present new ways to measure and visualize potential data leakage in FL. Our work is a step towards establishing reproducible methods of measuring data leakage in FL and could help determine the optimal tradeoffs between privacy-preserving techniques, such as differential privacy, and model accuracy based on quantifiable metrics.

Towards Understanding the Risks of Gradient Inversion in Federated Learning

Federated learning (FL) allows the collaborative training of AI models without needing to share raw data. This capability makes it especially interesting for healthcare applications where patient and data privacy is of utmost concern. However, recent works on the inversion of deep neural networks from model gradients raised concerns about the security of FL in preventing the leakage of training data. In this work, we show that these attacks presented in the literature are impractical in real FL use-cases and provide a new baseline attack that works for more realistic scenarios where the clients’ training involves updating the Batch Normalization (BN) statistics. Furthermore, we present new ways to measure and visualize potential data leakage in FL. Our work is a step towards establishing reproducible methods of measuring data leakage in FL and could help determine the optimal tradeoffs between privacy-preserving techniques, such as differential privacy, and model accuracy based on quantifiable metrics.

Innovative Proposal Concerning the Human Factors Shaping Institutional Responses to Domestic Violence – the IMPRODOVA Project

The introductory chapter of this book presents the book's structure as a whole and gives a brief overview of its single chapters and their interrelatedness. The aim of IMPRODOVA - Improving Frontline Responses toHigh Impact Domestic Violence was to deliver recommendations, toolkits and collaborative training for European police organisations and medical and social work professionals to improve and integrate theinstitutional response to high-impact domestic violence. IMPRODOVA had two main components: analysis of current institutional responses to high-impact domestic violence and the development of effectivesolutions to improve those responses. Efforts were made to avoid a one-size-fits-all approach and contextualise our solutions, tools and guidelines to make them applicable to a wide range of societies.

Construction and Evaluation of a Virtual Reality Laparoscopic Surgery Collaborative Training Platform: Taking Laparoscopic Cholecystectomy as an Example

Backgroud: Virtual reality (VR) technology represents the future of medical education due to its unique advantages, especially with the Covid-19 pandemic lasting. We developed a laparoscopic VR surgery collaborative training platform hoping to shed light on future medical education in China.Methods: We constructed a VR surgery training platform and designed surgery curriculum on laparoscopic cholecystectomy (LC). 36 first-year postgraduate students in China standardized training program for resident doctor (C-STRD) from the Third Xiangya Hospital of Central South University were enrolled for validation trials. In the Phase I trial, 12 students performed LC in the exploration mode. After training in the surgery learning mode, they performed LC again. The LC scores before and after training were compared. In the Phase II trial, another 12 students were randomly assigned to either the collaborative group or the control group. The former trained with a senior surgeon collaboratively in the surgery learning mode and then performed LC alone in the exploration mode. The latter trained in the surgery learning mode by themselves and performed LC in the exploration mode. The LC scores between groups were compared. The user experience (intention to use, skills improvement, usability, degree of enjoyment) were analyzed through questionnaires from the above 24 students. Interest in surgery learning of Phase I students was compared with 12 students who didn’t experience the VR platform.Results: In Phase I trial, the mean LC scores of the students were elevated from 56.83 to 61.17 (p=0.042) after learning in surgery learning mode. In Phase II trial, collaborative group students had higher scores than their rivals (67.17 vs 61.33, p=0.014). Most students have a positive users’ experience regarding the intention to use and skills improvement. Collaborative group students had higher evaluation regarding usability. Students who experienced the VR platform were significantly more interested in future surgery learning (3.60 vs 2.58, p <0.05).Conclusion: Our study constructed a VR platform for collaborative surgery training, which showed an excellent training effect. Medical students rated the platform highly, and their interest in learning increased.

μDFL: A Secure Microchained Decentralized Federated Learning Fabric atop IoT Networks

<div>Federated Learning (FL) has been recognized as a privacy-preserving machine learning (ML) technology that enables collaborative training and learning of a global ML model based on the aggregation of distributed local model updates. However, security and privacy guarantees could be compromised due to malicious participants and the centralized aggregation manner. Possessing attractive features like decentralization, immutability and auditability, Blockchain is promising to enable a tamper-proof and trust-free framework to enhance performance and security in IoT based FL systems. However, directly integrating blockchains into the large scale IoT-based FL scenarios still faces many limitations, such as high computation and storage demands, low transactions throughput, poor scalability and challenges in privacy preservation. This paper proposes uDFL, a novel hierarchical IoT network fabric for decentralized federated learning (DFL) atop of a lightweight blockchain called microchain. Following the hierarchical infrastructure of FL, participants in uDFL are fragmented into multiple small scale microchains. Each microchain network relies on a hybrid Proof of Credit (PoC) block generation and Voting-based Chain Finality (VCF) consensus protocol to ensure efficiency and privacy-preservation at the network of edge. Meanwhile, microchains are federated vie a high-level inter-chain network, which adopts an efficient Byzantine Fault Tolerance (BFT) consensus protocol to achieve scalability and security.</div><div>A proof-of-concept prototype is implemented, and the experimental results verify the feasibility of the proposed uDFL solution in cross-devices FL settings with efficiency, security and privacy guarantees.</div>

μDFL: A Secure Microchained Decentralized Federated Learning Fabric atop IoT Networks

<div>Federated Learning (FL) has been recognized as a privacy-preserving machine learning (ML) technology that enables collaborative training and learning of a global ML model based on the aggregation of distributed local model updates. However, security and privacy guarantees could be compromised due to malicious participants and the centralized aggregation manner. Possessing attractive features like decentralization, immutability and auditability, Blockchain is promising to enable a tamper-proof and trust-free framework to enhance performance and security in IoT based FL systems. However, directly integrating blockchains into the large scale IoT-based FL scenarios still faces many limitations, such as high computation and storage demands, low transactions throughput, poor scalability and challenges in privacy preservation. This paper proposes uDFL, a novel hierarchical IoT network fabric for decentralized federated learning (DFL) atop of a lightweight blockchain called microchain. Following the hierarchical infrastructure of FL, participants in uDFL are fragmented into multiple small scale microchains. Each microchain network relies on a hybrid Proof of Credit (PoC) block generation and Voting-based Chain Finality (VCF) consensus protocol to ensure efficiency and privacy-preservation at the network of edge. Meanwhile, microchains are federated vie a high-level inter-chain network, which adopts an efficient Byzantine Fault Tolerance (BFT) consensus protocol to achieve scalability and security.</div><div>A proof-of-concept prototype is implemented, and the experimental results verify the feasibility of the proposed uDFL solution in cross-devices FL settings with efficiency, security and privacy guarantees.</div>