innovation fund themed competition webinar - session 2

15
Challenge 2 Free up personnel through the application of innovative use of machine learning algorithms and artificial intelligence (AI) for military advantage

Upload: heather-fiona-egan

Post on 20-Mar-2017

191 views

Category:

Government & Nonprofit


0 download

TRANSCRIPT

Challenge 2Free up personnel through the application of innovative use of machine learning algorithms and artificial intelligence (AI) for military advantage

Next generation Air ForceInformatio

n collection

Human analytic capacity

People

TechnologyProcess

Cha

lleng

e

Decision advantage

Manage, analyse and exploit multiple information sources

…….at pace

Exponential data

Identify the right 1%

Constrained human capacity

RAF ISTAR* Force*Intelligence Surveillance Target Acquisition and Reconnaissance

E-3D Sentry Shadow R1

Rivet JointSentinel R1

Reaper - Protector

1 ISR Wing P-8 PoseidonTornado Tac Recce

Space

Exponential data – ISR Services

Multi-Intelligence Fusion & Cross-

Cue

Automation AI Analytics

Optimise Intelligence

Analyst Fusion /

Cross-Cue

ImageryMulti-SpectralHyper-Spectral

ElectronicCommunications

Foreign Intel SystemsMeasurement &

SignaturesCyber & EMAcousticsHuman

Open SourceHistorical /Archive

Direct

Collect

Process

Disseminate

PROCESS Information = Human / Machine Partnership

Decision Advantage

Human / machine analytics

Open source activity

Ground moving targets

Google imager

y

Cyber and electromagnetic activity

Airborne imagery

Synthetic radar

imagery

Recognised air

picture

Wider opportunitiesEngineering and logistics

• improve aviation safety• keep aircraft in the air for longer• environmental stress and trend analysis• work closer to mandated tolerance limits

Cyber defence• continuous activity on networks• identify the anomalies

Conclusion• decision advantage

• exponential data vs human capacity – close the gap

• the right 1% ..... at pace

• human and machine in partnership

OFFICIAL

Defence and Security Accelerator

Defence and Security AcceleratorDefence and Security Accelerator

Challenge 2: Technical perspectiveLeo Borrett, Capability Adviser, Data Science

OFFICIALUK OFFICIAL

LSVRC* classification challenge: error rates by year red line = human error rate

Face recognitionSpeech recognition Lip reading Machine translation

*Large Scale Visual Recognition Challenge

OFFICIAL

What do we want?

Over-fitting

Free and open-source software (where appropriate)

Solve one aspect of the problem well

OFFICIAL

Automated activity classificationMOD requires methods for automated detection and classification of activities and intents from multiple sensor types using state-of-the-art machine learning and artificial intelligence (AI)

Fathom neural computer stick

Adversarial machine learning example

• beyond simple feature extraction• ability to operate “at the edge”• semi-supervised and un-

supervised methods• approaches to enable robust

deployment (for example adversarial machine learning)

OFFICIAL

Cognitive computing

UK OFFICIAL

Automated speech recognition

Knowledge graphs

Natural language question answering

Automation of manual tasks

Flag adversary activity of interest

Infer new “knowledge”

Identification of false information

OFFICIAL

Combined human/machine derived models

UK OFFICIAL

MOD is interested in the combination of human-derived models, exploiting domain knowledge using a rules-based approach; with machine-derived models, which require large volumes of data and driven by machine learning technologies. How do we:• combine data and human derived models• build more robust statistical models of subjective measures

(for example assessment of threat)• ensure data-driven models are transparent and

understandable for analysts and operators?

OFFICIAL

Predictive analyticsApplication of machine learning in support of predictive modelling to guide military decision

making. MOD requires solutions which go beyond enhancing military understanding of current situations, but predicts future outcomes, including actions, anomalies, intent and

movements, to guide decision makers in support of operational planning.

UK OFFICIAL

Information overload

Situation understanding

Predictive analytics

Prescriptive analytics