The cognitive pillar, launched in September 2019, aims to provide the LIST research community with advanced technologies, services and solutions in data / business analytics, artificial intelligence and "big data".

It is based on three main parts: 2 located on premises and one in the cloud.

On premises, services are offered by a platform combining Open-source solutions and IBM Cloud Paks (Data, Applications, Multi-Cloud, and API). This offer is completed by public cloud solutions.

CogPil

CogPil is a powerful and innovative platform providing state-of-art technologies and services in data analytics and AI, based on a hybrid cloud architecture.

Applications and key expertise

KEY EXPERTISE:	APPLICATION DOMAINS
Analytics tools for Proof of concept (POC) development and/or projects about: Advanced Statistics Machine Learning Business Intelligence Advanced Visualisation Optimisation Simulation … on LIST’s premises or in Public Cloud	Used for: Process improvement Resource planning / scheduling Customer segmentation Forecasting Preventive maintenance Root cause analysis Anomaly detection Risk & fraud Analytics …

Equipment

In numbers:

• 34 physical servers, supporting Hadoop and Private Cloud solutions (including storage and backup)
• Network layer including Omni-Path connectivity
• Distributed storage layer including auto-tiering on Spectrum Scale/GPFS

Offered software capabilities:

• A cognitive and analytical layer based on IBM Watson (IBM Public Cloud and on-premise).
• A Hadoop enterprise cluster installed on physical servers with GPUs for distributed compute dedicated to Machine Learning, Deep Learning, Big Data analysis in batch mode, IoT stream analytics and real-time data processing.
• A tight integration between two worlds (Watson / Hadoop Open Source), offering users the benefits of both platforms.