Search

Federated learning is a distributed machine learning framework that has been widely applied in scenarios that require data privacy. To obtain a neural network model that performs well, when the model falls into a bug, existing solutions retrain it on a larger training dataset or the carefully selected samples from model diagnosis. To overcome this challenge, this paper presents Fed-DNN-Debugger, which can automatically and efficiently fix DNN models in federated learning. Fed-DNN-Debugger fixes the federated model by fixing each client model. Fed-DNN-Debugger consists of two modules for debugging a client model: nonintrusive metadata capture (NIMC) and automated neural network model debugging (ANNMD). NIMC collects the metadata with deep learning software syntax automatically. It does not insert any code for metadata collection into modeling scripts. ANNMD scores samples according to metadata and searches for high-quality samples. Models are retrained on the selected samples to repair their weights. Our experiments with popular federated models show that Fed-DNN-Debugger can improve the test accuracy by 8% by automatically fixing models. [ABSTRACT FROM AUTHOR]

In this paper, we study the problem of privacy-preserving data synthesis (PPDS) for tabular data in a distributed multi-party environment. In a decentralized setting, for PPDS, federated generative models with differential privacy are used by the existing methods. Unfortunately, the existing models apply only to images or text data and not to tabular data. Unlike images, tabular data usually consist of mixed data types (discrete and continuous attributes) and real-world datasets with highly imbalanced data distributions. Existing methods hardly model such scenarios due to the multimodal distributions in the decentralized continuous columns and highly imbalanced categorical attributes of the clients. To solve these problems, we propose a federated generative model for decentralized tabular data synthesis (HT-Fed-GAN). There are three important parts of HT-Fed-GAN: the federated variational Bayesian Gaussian mixture model (Fed-VB-GMM), which is designed to solve the problem of multimodal distributions; federated conditional one-hot encoding with conditional sampling for global categorical attribute representation and rebalancing; and a privacy consumption-based federated conditional GAN for privacy-preserving decentralized data modeling. The experimental results on five real-world datasets show that HT-Fed-GAN obtains the best trade-off between the data utility and privacy level. For the data utility, the tables generated by HT-Fed-GAN are the most statistically similar to the original tables and the evaluation scores show that HT-Fed-GAN outperforms the state-of-the-art model in terms of machine learning tasks. [ABSTRACT FROM AUTHOR]