May 17th 2015 - IoThingsMilan · Centric • Consumer IoT centric ... •Understanding the concept of Artificial Intelligence •Assessing methods of artificial intelligence for the

May 17th 2015

DATA

[+] Focusing on Sector needs - securely

[-] Not Enabling Wide Interactions

IoT Applications

INTERACTION

UNDERSTANDING

CONTROL

INSIGHT

& more

DATADATA

DATA

From the single application (business purpose)

towards a context with several applications and

several datasets

• An IoT Platform is an intelligent layer that connect the things to the network and abstract applications from the things in order to enable the development of services.

• We want all this because we want to achieve at least three main objectives:

• flexibility (being able to deploy things in different contexts),

• usability (being able to make the user experience easy),

• and productivity (enabling service creation in order to improve efficiency, but also enabling new service development)

New Entrants

Acquisitions

2016

2016

2019

2019

• Different platforms offer different services and differentdorms of specialisation.• Those are indications forgrouping platforms.• Application Dev Centric• Data Analytics Centric• Device Management Centric• Connectivity ManagementCentric• Consumer IoT centric• Vertical Centric• Developers Focus

DiscoveryDeep

AssessmentRFP

Exploring the IoTplatform landscape and short-list platforms that match requirements through tools such as mapping analysis and features assessment.

Technical and business assessment of the short-list through tools such as hands-on evaluation.

Writing specific and robust RFPs for final selection.

Discovery

Deep Assessment

User Account Creation

Device Connection to Platform

Data Exchange Device Platform

Rules Creation

Data Visualisation

Device Data Reading

Settings Changing

Reporting and Analytics

Indicators

Time

Usability

Documentation Quality

Support

Tasks

Community

Deep Assessment

In collaboration:

Data Collection and Storage

Data Orchestration

Basic Analytics, Reporting, Visualizations

Predictive Analytics

Advanced Analytics (Simulations, Prescriptive Analytics, ML, AI)

Related Issues: Edge Analytics,Connectivity Analytics

Deep Assessment

Birkbeck Institute for Data Analytics (BIDA)

In collaboration with:

The Artificial Intelligence of Things Project

Objectives: • Understanding the concept of Artificial Intelligence• Assessing methods of artificial intelligence for the Internet of Things• Assessing the applications of AI in IoT contexts• Assessing the market landscape of AI offer in the IoT space• Defining a framework for applying and using AI in IoT spaces

Artificial Intelligence as an overall concept. It is more thanjust automation, but it regards the capacity of a machine to Express an appropriate behaviour in response to changes andOpportunities in the environment in which the machine operates

Cognitive ComputingMachine LearningDeep LearningUnsupervised versus supervised

Assisted Intelligence

Augmented Intelligence

Autonomous Intelligence

Autonomous Intelligence• Dynamic nature of the task• Automated Decisions• Machine learns continuously• Example: autonomous vehicle

Assisted Intelligence• Static nature of the task• Task is automated• Humans do not learn• Machine learns

Augmented Intelligence• Dynamic nature of the task• Human-machine exchange• Example: Smart clinical decision support

AI AIAI

Source: Maranda, A, Kaczmarek, D. 2015, Selected Methods of Artificial Intelligence for Internet of Things Cooperation

AI

• IBM Watson• Hitachi Lumada• IoTFy• Cumulocity• PTC• Bosch• Oracle

• Smart Sensors – that bring together IoT capabilities and AI capabilities – are using for

predictive and prescriptive activities at the edge in industrial processes. • Adaptive and autonomous systems without human intervention

such as monorail systems.• Autonomous surveillance in construction sties via drones• Several activities in the healthcare sector:

• Using EEG of infants (12 months) to predict risk of developing specific diseases;

• Determining Parkinson’s score with the aim to move the test from the doctor’s office into the patient’s hands.

• Detecting knee range of motion for surgery recovery patients.

• Beecham Research is running research in the areas of IoT platforms with the objective of helping adopters to explore the landscape and understand how they can find answers in that landscape for their requirements.

• Artificial Intelligence will move the current states of the IoT space at the next level, the level of autonomous machines and predictiveness. But, we all need to learn what AI is about, how it can be used, for what purposes, and what that means from an organisational perspective.

• Partnerships are key also at the research level.

[email protected]

@Saverio_Romeo