Distributed LSTM-Learning from Differentially Private Label Proportions
This addresses privacy and efficiency challenges in decentralized spatio-temporal data analysis, but appears incremental as it builds on existing LSTM and differential privacy techniques.
The paper tackles the problem of learning from spatio-temporal data while preserving privacy and reducing communication bandwidth by proposing two distributed LSTM models with differential privacy. The evaluation on datasets like Pems-Bay and METR-LA shows a tradeoff between performance and data privacy, though no concrete numerical results are provided in the abstract.
Data privacy and decentralised data collection has become more and more popular in recent years. In order to solve issues with privacy, communication bandwidth and learning from spatio-temporal data, we will propose two efficient models which use Differential Privacy and decentralized LSTM-Learning: One, in which a Long Short Term Memory (LSTM) model is learned for extracting local temporal node constraints and feeding them into a Dense-Layer (LabelProportionToLocal). The other approach extends the first one by fetching histogram data from the neighbors and joining the information with the LSTM output (LabelProportionToDense). For evaluation two popular datasets are used: Pems-Bay and METR-LA. Additionally, we provide an own dataset, which is based on LuST. The evaluation will show the tradeoff between performance and data privacy.