IoT, Biometria, Big Data & Analytics e o Impacto nos Negócios · NEC “Invariant Analysis” A quality and performance prediction analysis system that automatically identifies regularities

IoT, Biometria, Big Data & Analytics e o Impacto nos Negócios

Seminário - IoT - Internet das Coisas - “A Revolução das Coisas”

nos Negócios

Alexandre Soyama

Manager, Solutions and Engineering

NEC Latin America S.A.

A bit of History

2 © NEC Corporation 2016 NEC Group Internal Use Only

Dependency, Complementarity and SynergyDependency, Complementarity and Synergy

Big Data meets IoT

Real World Data

Network Data Big Data

IoT

- Binding physical world and cyber

world- Application-specific

- Cloud/edge/device computing


IT system Data

Data Mining

- Unstructured data- Indirect assessment of Real

World through SNS exchanges

- Correlation than Causation

- Stored data in IT system

Which IoT?


Industrial IoT - Types

▌Sensors in Products (by manufacturers)

� Aircraft engine, Wind Turbine, …

▌Sensors in Assets (by operators)

� Fleet Management, …

▌Manufacturing Process Source : GE


▌Manufacturing Process � “Industry 4.0”

LED display LCD displayGPS

3G/WiFiOn Board Computer

(OBC)Bus Operator /

Regulator

In Vehicle Systems (IVS)

Transit Management System (TMS)

Driver Display Unit (DDU)

Server

Passenger Information System (PIS)

Data Analysis

▌Big Data analysis based on Machine Learning

“Invariant Analysis”


Power DemandForecasting

Quality Forecasting

Deterioration Forecasting

Monitor for signs of failures in social infrastructures(Bridges, expressways)

Monitor for signs of failures in manufacturing plants

(assembly, chemicals, etc.)

Monitor for signs of failures at power plants

“Heterogeneous Mixture Learning”

Fresh FoodForecasting

NEC “Invariant Analysis”

▌A quality and performance prediction analysis system that automatically identifies regularities amongst large volumes of data and makes highly-accurate predictions about product quality and performance

Automatically build a normal operating model

De

te

ctio

n a

nd

Pre

dic

tio

n o

f

De

te

ctio

n a

nd

Pre

dic

tio

n o

f


Behavior

analysis

Model

building

Data analytics

Verification

and analytics

High speed

processing

matching

Machine Learning

Data

Sensor data

System Log

・・・

De

te

ctio

n a

nd

Pre

dic

tio

n o

f

An

om

alie

s

De

te

ctio

n a

nd

Pre

dic

tio

n o

f

An

om

alie

s

Detection ofabnormalities

NEC “Heterogeneous Mixture Learning” (HML)

▌Automatic classification of complicated heterogeneous big data into cases and automatic generation of a prediction model for demand forecast for each case

� No expert prior knowledge is required. And human understandable rules can be extracted from classified cases

Yｎ

=a0x0+a1x1+・・・+a50x50Yｎ

=a0x0+a1x1+・・・+a50x50Y =a x +a x +・・・+a x


100

150

200

250

300

2012/8/23 2012/8/30 2012/9/6 2012/9/13 2012/9/20 2012/9/27

実績値

予測値

学習期間（4ヶ月）予測Training Prediction

Predicted

Measured value

Yｎ

=a0x0+a1x1+・・・+a50x50Yｎ

=a0x0+a1x1+・・・+a50x50model ： Y

ｎ

=a0x0+a1x1+・・・+a50x50

Landslide Prediction System – Rio de Janeiro

Pilot Project in Rio de Janeiro City with COR (Rio Operation Center)Pilot Project in Rio de Janeiro City with COR (Rio Operation Center)

▌Big Data prediction system based on advanced physical studies of geology

▌Integrated with existing sensors,


▌2 (two) risk areas were chosen by the Civil Defense:

�Morro da Formiga (Tijuca)

�Morro da Cotia (Estrada Grajaú – Jacarepaguá)

�7 years of historical data of rain, past landslide events and low altitude soil study were

considered in the analysis

▌Integrated with existing sensors,

such a rain gage, radar, satellite and humans

Healthcare - The Concept of e-Pathologist

e-Pathologist supports to improve the efficiency andthe quality control of pathology diagnosis

Pathology

Slide IT Device

Image Analysis Module

10 © NEC Corporation 2015 Confidential

Gastric Colorectal Breast

Gastric Colorectal Breast

Slide Scanner

Clinical Information Pathology Report

Pathologist, TechnicianData Management Module

Slide Image Clinical Info Analysis Result

What is Biometrics? ... Binding the Real World

【【【【【【【【Physical CharacteristicsPhysical Characteristics】】】】】】】】【【【【【【【【Behavior CharacteristicsBehavior Characteristics】】】】】】】】

IrisIrisSignatureSignature

Blinking Blinking EyeEye

FingerprintFingerprintPalmPalmVeinVein…… FaceFace


SmellSmellVoiceVoice

IrisIrisSignatureSignature

Typing paceTyping pace

Walk paceWalk pace

Counting, age, gender, face recognition, behavior detection...


CounterGender

AgeBehavior

CRM

Password

Security

Audience Measurement Analytics

Shopping mall @São Paulo on 24/12/2014Christmas

NEC Biometrics Market Leadership


Biometrics for the Private Sector: Universal Studios Japan


Biometrics for Public Safety – John F. Kennedy Airport


Biometrics for Public Safety – Receita Federal, Brazil


Biometrics for Public Safety – Tigre, Argentina

Improve Surveillance Efficiency by detecting human and traffic related abnormalbehaviors in specific context /scenarios: Railways station, Banking areas, Accessroads, Streets, Parks & Open places


Biometrics for Payment – Tokyo, Japan

NEC uses Face Recognition for Payment Authorization500 (kick-off) �� 100,000 employees


http://www.nikkei.com/article/DGXLASDZ21HIM_R20C16A6000000/

w

Thief Customer Staff Fresh Food Internal & External

Physical Loss Fresh Food Data Loss

Thief Customer Staff Fresh Food Internal & External

12% 7%

Sources:

Supermarket Loss Estimates for Fresh Fruit, Vegetables, Meat, Poultry, and Seafood... / EIB-44 - Economic Research Service/USDA

Global Retail Theft Barometer

Fraud costs US retailers $54bn a year, according to new Kount Volumatic 2013 survey

How retailers are coping with the challenge of food waste, Ian Walters, head of retail at supply chain standards organisation GS1

1,48%Loss

(% net sales)

Avoidable <50% <77% <50%

1,3%

System that automatically identifies regular

patterns hidden within past sales data (big

data), and forecasts product demand figures

with a high level of precision, places orders

automatically and assists with the decision-

making process.

123

HYBRID LEARNING

LEARNING FORECASTING RE-LEARNING

AUTOMATED ORDERING

DELIVERS MORE VALUE

Demand Forecasting &

Automated Ordering Solution

Fresh Food

Preventing stock-out

Lower waste ratio

Lower ordering man-hours

3 ADVANTAGES

NEC’s AFIM vs Traditional Forecast Engine

*based on results of customer project

Retail Revenue Generators

Sources:

Tightwad Promotion: sales programs that sell - John Kuraoka

Adidas' Digital Signage Boosts Sales By 40% - February 2013 Integrated Solutions For Retailers By Bob Johns, associate editor

Journal of Emerging Trends in Computing and Information Sciences - CIS Journal Vol. 5, No. 3 March 2014

FREQUENT CUSTOMER

PROGRAM (VIP)

CUSTOMER BROWSING, BEHAVIOR ANALYSIS AND

DIGITAL SIGNAGE

<23%Sales

Increase<40%

INTERESTING CASES

BiometricsBiometrics

Biometrics for Public Safety – Okayama, Japan


The Future of Biometrics: IoT


Biometrics for the Private Sector: SoftBank Pepper


The Future of Biometrics: Blockchain


‘NEC the WISE’ � Menu de Soluções de Inteligência Artificial

▌Facilitar a cooperação entre pessoas e a IA na resolução de questões da nossa sociedade, acelerando o desenvolvimento de aplicações de IA para a IoT, o Big Data e a segurança, seja para resolver problemas sociais ou para transformar empresas


NEC’s TECHNOLOGIES � White Label for Service Providers

Collection of large-scale data

Analysis and prediction

Multiple Solutions

• Invariant analysis

• Heterogeneous mixture learning

• Facial image analysis

• Surveillance cameras

• Diverse sensors

Diverse sensors and

human interface

technologies

High-performance/high-

reliability core IT

technologies

IoTBIG

DATA

Unique

Unique

No.1


• Facial image analysis

• Behavior analysis• Smart devices

SDN : Software-Defined Networking

・Network virtualization

・Network Function Virtualization

Next-generation network

technologiesSDN

* Rated as No. 1 among organizations participating in an evaluation task organized by the U.S. National Institute of Standards and Technology (NIST)

Unique

No.1*

From the seafloor

to outer space

World’s first SDN controller & switch

Essential to future information system

Aeroporto

Aeroporto Internacional@Rio de Janeiro

Metro

São Paulo Linha 6

#1 NEC filiada fora do Japão- Desde 1968

#1 Integrador de Sistemas paraInstalações Críticas- 2014 Copa do Mundo

- 2016 Olimpíadas

- Aeroporto e Porto no Rio

- São Paulo Metro Linha 6

#1 Soluções de Biometria

#1 Soluções para Hotelaria

#1 Segurança Cibernética

Porto

Porto Açu@Rio de Janeiro

NEC no Brasil – 2016 Perfil Corporativo


Cinema & Displays

300+ Cinemas500+ LFD instalações

Comunicação Unificada

600+ clientes diretosMaior projeto de telefonia IP

com 3C no mundo

Supercomputador

Previsão do Tempo15+ de operação

Soluções para Operadoras & DC- IP network

- Rádio

700+ Clientes Diretos

NOC / SOC

Governo Federal (TCU)Estado de São Paulo (SEFAZ)

Arenas

Copa do MundoOlimpíadas

Biometria

5 Estados – Civil e Criminal14 Aeroportos InternacionaisPrevenção de Perdas Varejo

Operadoras e DC

#1 Rádio#1 Juniper SI

SDN / NFV

1o Projeto de vCPE com a Telefonica

Documents

IoT, Biometria, Big Data & Analytics e o Impacto nos Negócios · NEC “Invariant Analysis” A quality and performance prediction analysis system that automatically identifies regularities