A Novel Self Deep Learning Semi-Supervised Approach to Classify Unlabeled Multivariate Time Series Data
, Professor, SENAI CIMATEC
Learn how to overcome the problem of unlabeled data with semi-supervised self-learning to develop a deep learning model, applied to a public dataset of an oil & gas offshore platform, containing different types of normal and pre-failure behaviors using multivariate time series data collected from sensors. We'll describe these key ingredients to enable this: (1) unlabeled multivariate time series data; (2) anomalous pattern recognition; and (3) semi-supervised self-learning. We used NVIDIA GPU Tesla V100-SXM3 to accelerate and develop the self deep learning semi-supervised model. With this, predictive maintenance solutions can be further developed, even in scenarios with few — or even no — labeled data.