Deep Learning Toolkit
The Deep Learning Toolkit (DLT) is a tool which leverages the Machine Learning and Deep Learning techniques for time series data, like Long Short Term Memory (LSTM) and Encoder-Decoder, to provide solutions of predictive maintenance and/or anomaly detection in the Industry scenarios. The DLT provides as well, using the same techniques, time series forecasting that can be used to predict the prices of goods.
The Deep Learning Toolkit is made of two different modules, the price prediction and the predictive maintenance one. Both have been developed during the COMPOSITION project and have then been adapted to the EFPF ecosystem. The modules are independent and can be independently deployed, while being part of the same suite and sharing part of the codebase.
Both modules do not include a graphical user interface or a web portal by default. This is because of the variety of the processes they can be used in, so that an ad-hoc interface can be developed according to the needs of each user.
The Deep Learning Toolkit for Price Forecasting has a UI provided by CERTH through its Visual and Data Analytics Tool.
The Deep Learning Toolkit for Predictive Maintenance has a graphical user interface provided by LINKS, which is made available directly to the final user.