The “Talent Lab” project
The Calabria Region’s “Talent Lab” initiative aims to support the growth of innovative startups. In this context, Revelis has developed a platform for the development of Artificial Intelligence solutions based on machine learning, deep learning and reasoning techniques.
POR CALABRIA FESR-FSE 2014-2020
ASSE I – PROMOZIONE DELLA RICERCA E DELL’INNOVAZIONE
Obiettivo Specifico 1.4 “Aumento dell’incidenza di specializzazioni innovative in perimetri applicativi ad alta intensità di conoscenza”
Azione 1.4.1 “Sostegno alla creazione e al consolidamento di startup innovative ad alta intensità di applicazione di conoscenza e alle iniziative di spin-off della ricerca”.
Talent Lab: Revelis Artificial Intelligence Smart Environment
Through the “Talent Lab” project, Revelis has created the Revelis Artificial Intelligence Smart Environment which offers:
– machine learning/deep learning and automatic reasoning functionalities, thus enabling the development of Artificial Intelligence solutions on Big Data
– explanation mechanisms of the decision-making models learned, in order to increase the effectiveness of man-machine interaction in the context of industrial and operational processes that involve human actors and intelligent automatic systems
The Revelis Artificial Intelligence Smart Environment can be delivered in “on-premise” mode (at the customer’s infrastructure) or in “Software-as-a-Service” (SaaS) mode
Data Layer
Allows the storage and querying of:
– structured data
– texts
– streams
Business Layer
– tools for multidimensional analysis
– framework for distributed pre-processing
– libraries for machine / deep learning (induction)
– engine for automatic reasoning (deduction)
Presentation Layer
Provides visual analysis of:
– data
– models (Insights)
Benefits
The Revelis Artificial Intelligence Smart Environment is a “general purpose” platform, and therefore can be used in AI applications in different applicative sectors, both in cloud processing and at-the-edge
Minimization of the time needed to move from the “design” phase to the “implementation” phase of AI solutions
Combination of inductive and deductive techniques to ensure maximum flexibility in satisfying user needs
Availability of tools that, by explaining the behavior of automatic decision-making, allow humans to interact more effectively with machines
Massive and real-time processing, ensuring the maximum efficiency of the algorithms and computing resources used
Predictive Maintenance
PlugAIn
The verticalization of Revelis Artificial Intelligence Smart Environment for predictive maintenance will meet the needs of industrial companies that, in the production / maintenance phase of the plants, need to move from a “planned” or “reactive” maintenance model to a model capable of verifying the state of health of machinery in order to predict when a failure will occur, and consequently to determine an optimization of maintenance activities, with enormous improvements and savings for the customer.
Diagnostic time series management
Use of ontologies for plants modeling
Fault prediction and anomaly detection machine/deep learning models
Fault explanation
Innovation Plan
The Talent Lab Project is formed by 5 workpackages
Study of the state of the art of methods for Artificial Intelligence on Big Data (inductive and deductive approaches)
1. Deep Learning techniques
2. Knowledge Representation and Reasoning (KR&R) techniques
Design and implementation of Revelis Artificial Intelligence Smart Environment
1. State of the art of Machine/Deep Learning tools
2. Design of a Machine/Deep Learning software component
3. Design of a Big Data ASP component
4. Implementation Revelis Artificial Intelligence Smart Environment
Mining algorithms and techniques for predictive maintenance
1. Preprocessing and feature engineering techniques for sensor-based big data analysis
2. Anomaly detection and outlier explanation techniques for streaming data
3. Deep Learning techniques for predictive maintenance
Reasoning models and methods for predictive maintenance
1. Deductive Root Cause Analysis methods
2. Domain model for a civil or industrial use case
PlugAIn implementation
1. Implementation of machine/deep learning for predictive maintenance
2. Implementation of reasoning models for predictive maintenance
3. Integration of software components and deploy of PlugAIn
