ºÝºÝߣ

ºÝºÝߣShare a Scribd company logo

Uthought executive overview

•
•
Glenn West
Glenn West

Containers, Microservices, Big Data, And deep neural networks, on a massive scale predicting stock prices. This is what you can do with Docker, Kubernetes, and Openshift Enterprise. I takes a couple of handful of container types, and a little bit of hardware, but can be done with a small development team.

1 of 14
Uthought executive overview
DISRUPTIVE CHANGE IN TECHNOLOGY • WHAT IF: • YOU COULD ANALYZE EQUITIES FROM DIFFERENT POINT OF VIEWS • TRY MULTIPLE ALGORITHMS IN PARALLEL FOR EACH POINT OF VIEW • SELECT THE BEST OF THESE SOLUTIONS THAT GIVE GOOD YIELD • EXECUTE TRADES ON THEM • AND DO THIS ON A MASSIVE SCALE, WITH MINIMAL HARDWARE?
OVERVIEW • MULTIPLE STOCK TICKER PROVIDERS • ALLOWS SYSTEM TO OPERATE MULTIPLE MARKETS • STOCK QUALIFIER SYSTEM • SET METRICS IN ENGLISH LANGUAGE TO ESTABLISH EQUITIES OF INTEREST • POV CREATORS • REUSABLE POINT OF VIEW SYSTEM • DEEP NEURAL NETWORKS • 15 DIFFERENT TYPES
OVERVIEW • ANALYTICS AND BIG DATA GENERATOR • RANKS AND REVIEWS THE DATA • TRADE PRE-FILTER • USES ENGLISH LANGUAGE RULES TO SELECT PROSPECTIVE TRADES • TRADE APPROVAL SYSTEM • PORTFOLIO REVIEW
uThought – Tech Overview • LEVERAGES MICRO SERVICE ARCHITECTURE • DOCKER/CONTAINERS • MESSAGE Q’S WITH PUBLISH SUBSCRIBE • NODEJS – THE FASTEST GROWING ENVIRONMENT FOR DEVELOPMENT • DEEP NEURAL NETWORKS FOR LOW LEVEL MACHINE INTELLIGENCE • META-AS-A-SERVICE – MACRO-SCALE ADAPTATION AND GROWTH OF SYSTEM
WHY? • ABILITY TO FIND PATTERNS AND CONVERT THEM INTO BUYING DECISIONS • ABILITY TO USE A ORGANIC APPROACH TO A TRADITIONAL DIGITAL SYSTEM • ABILITY TO RUN MANY ALGORITHMS WHY KEEPING OVERHEAD LOW • USES BIG DATA CONCEPTS AND SELF LEARNING TO SELF ADAPT • SMALL REUSABLE COMPONENTS THAT ARE CONNECTED IN COMPLEX WAYS • ABILITY TO YOU ANY SIZE HARDWARE ENVIRONMENT TO BEST EFFECT
MICRO SERVICES • MICRO SERVICE – A SMALL UNIT OF WORK – TYPICALLY USES ONE OR MORE API’S IN AND ONE OR MORE API OUT • WELL DEFINED API BUT FLEXIBLE DEFINITION TO FIT THE NEED • ABLE TO BE IN ANY LANGUAGE • USE REST FOR PUBLIC API AND ESB/AMQ FOR INTERNAL API • REUSABLE
BEST WAY TO IMPLEMENT A MICRO SERVICE • MICRO SERVICE ARE BEST IMPLEMENTED IN A CONTAINER • REUSABLE AND SALEABLE • ALLOWS COMPLEX THINGS TO BE DONE BY SIMPLY INSTANCING MORE • BREAKS DEPENDENCIES BETWEEN SERVICES
MICROSERVICES – REST/AMQ • REST IS A STANDARD WAY OF PUBLISHING PUBLIC API ON THE WEB • INSIDE THE SYSTEM, THE NEED FOR 1-TO-MANY COMMUNICATIONS MAKES AMQ A BETTER CHOICE • MORE ORGANIC GROWTH AND SCALABILITY OF SYSTEM USING AMQ • RELIABILITY FOR FINANCIAL APPLICATIONS IS USEFUL
DEEP NEURAL NETWORKS Deep learning is a class of machine learning algorithms that[ use a cascade of many layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. The algorithms may be supervised or unsupervised and applications include pattern analysis (unsupervised) and classification (supervised).are based on the (unsupervised) learning of multiple levels of features or representations of the data.
POINT OF VIEWS • POINT OF VIEWS ARE THE PRECLASSIFICATION OF DATA, TO LOOK AT DATA IN PREDETERMINED FASHION. EXAMPLES OF THIS COULD BE AS SIMPLE AS: • FIRST TUESDAY OF A MONTH • FIRST OF QUARTER • HOLIDAYS AWARENESS • AND MANY OTHERS
GENETIC AND PARALLEL ALGORITHM • USING MULTIPLE IMPLEMENTATIONS OF DEEP LEARNING NETWORKS ON THESAME POV AND EQUITY, TO SEE WHICH ALGORITHMS FIND CORRELATION, AN D GIVE GOOD RESULTS • A SINGLE POV/EQUITY PAIR COULD HAVE 10’S OR 100’S OF PARALLEL IMPLEMENTATION
HOW BIG IS THE INFRASTRUCTURE IS THIS? • WHILE THE SYSTEM EVEN IN POC COULD USE 10,000 CONTAINERS, UP TO 1000 OF THESE CAN FIT ON ONE MACHINE • ARCHITECTURE PROVEN BY GOOGLE IN MASSIVE SCALE • SYSTEM ABLE TO USE STANDARD INFRASTRUCTURE, PRIVATE, PUBLIC OR OPEN HYBRID
HOW TO LEARN MORE: • FOLLOW ME ON LINKEDIN. GWEST@REDHAT.COM

More Related Content

Uthought executive overview

  • 2. DISRUPTIVE CHANGE IN TECHNOLOGY • WHAT IF: • YOU COULD ANALYZE EQUITIES FROM DIFFERENT POINT OF VIEWS • TRY MULTIPLE ALGORITHMS IN PARALLEL FOR EACH POINT OF VIEW • SELECT THE BEST OF THESE SOLUTIONS THAT GIVE GOOD YIELD • EXECUTE TRADES ON THEM • AND DO THIS ON A MASSIVE SCALE, WITH MINIMAL HARDWARE?
  • 3. OVERVIEW • MULTIPLE STOCK TICKER PROVIDERS • ALLOWS SYSTEM TO OPERATE MULTIPLE MARKETS • STOCK QUALIFIER SYSTEM • SET METRICS IN ENGLISH LANGUAGE TO ESTABLISH EQUITIES OF INTEREST • POV CREATORS • REUSABLE POINT OF VIEW SYSTEM • DEEP NEURAL NETWORKS • 15 DIFFERENT TYPES
  • 4. OVERVIEW • ANALYTICS AND BIG DATA GENERATOR • RANKS AND REVIEWS THE DATA • TRADE PRE-FILTER • USES ENGLISH LANGUAGE RULES TO SELECT PROSPECTIVE TRADES • TRADE APPROVAL SYSTEM • PORTFOLIO REVIEW
  • 5. uThought – Tech Overview • LEVERAGES MICRO SERVICE ARCHITECTURE • DOCKER/CONTAINERS • MESSAGE Q’S WITH PUBLISH SUBSCRIBE • NODEJS – THE FASTEST GROWING ENVIRONMENT FOR DEVELOPMENT • DEEP NEURAL NETWORKS FOR LOW LEVEL MACHINE INTELLIGENCE • META-AS-A-SERVICE – MACRO-SCALE ADAPTATION AND GROWTH OF SYSTEM
  • 6. WHY? • ABILITY TO FIND PATTERNS AND CONVERT THEM INTO BUYING DECISIONS • ABILITY TO USE A ORGANIC APPROACH TO A TRADITIONAL DIGITAL SYSTEM • ABILITY TO RUN MANY ALGORITHMS WHY KEEPING OVERHEAD LOW • USES BIG DATA CONCEPTS AND SELF LEARNING TO SELF ADAPT • SMALL REUSABLE COMPONENTS THAT ARE CONNECTED IN COMPLEX WAYS • ABILITY TO YOU ANY SIZE HARDWARE ENVIRONMENT TO BEST EFFECT
  • 7. MICRO SERVICES • MICRO SERVICE – A SMALL UNIT OF WORK – TYPICALLY USES ONE OR MORE API’S IN AND ONE OR MORE API OUT • WELL DEFINED API BUT FLEXIBLE DEFINITION TO FIT THE NEED • ABLE TO BE IN ANY LANGUAGE • USE REST FOR PUBLIC API AND ESB/AMQ FOR INTERNAL API • REUSABLE
  • 8. BEST WAY TO IMPLEMENT A MICRO SERVICE • MICRO SERVICE ARE BEST IMPLEMENTED IN A CONTAINER • REUSABLE AND SALEABLE • ALLOWS COMPLEX THINGS TO BE DONE BY SIMPLY INSTANCING MORE • BREAKS DEPENDENCIES BETWEEN SERVICES
  • 9. MICROSERVICES – REST/AMQ • REST IS A STANDARD WAY OF PUBLISHING PUBLIC API ON THE WEB • INSIDE THE SYSTEM, THE NEED FOR 1-TO-MANY COMMUNICATIONS MAKES AMQ A BETTER CHOICE • MORE ORGANIC GROWTH AND SCALABILITY OF SYSTEM USING AMQ • RELIABILITY FOR FINANCIAL APPLICATIONS IS USEFUL
  • 10. DEEP NEURAL NETWORKS Deep learning is a class of machine learning algorithms that[ use a cascade of many layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. The algorithms may be supervised or unsupervised and applications include pattern analysis (unsupervised) and classification (supervised).are based on the (unsupervised) learning of multiple levels of features or representations of the data.
  • 11. POINT OF VIEWS • POINT OF VIEWS ARE THE PRECLASSIFICATION OF DATA, TO LOOK AT DATA IN PREDETERMINED FASHION. EXAMPLES OF THIS COULD BE AS SIMPLE AS: • FIRST TUESDAY OF A MONTH • FIRST OF QUARTER • HOLIDAYS AWARENESS • AND MANY OTHERS
  • 12. GENETIC AND PARALLEL ALGORITHM • USING MULTIPLE IMPLEMENTATIONS OF DEEP LEARNING NETWORKS ON THESAME POV AND EQUITY, TO SEE WHICH ALGORITHMS FIND CORRELATION, AN D GIVE GOOD RESULTS • A SINGLE POV/EQUITY PAIR COULD HAVE 10’S OR 100’S OF PARALLEL IMPLEMENTATION
  • 13. HOW BIG IS THE INFRASTRUCTURE IS THIS? • WHILE THE SYSTEM EVEN IN POC COULD USE 10,000 CONTAINERS, UP TO 1000 OF THESE CAN FIT ON ONE MACHINE • ARCHITECTURE PROVEN BY GOOGLE IN MASSIVE SCALE • SYSTEM ABLE TO USE STANDARD INFRASTRUCTURE, PRIVATE, PUBLIC OR OPEN HYBRID
  • 14. HOW TO LEARN MORE: • FOLLOW ME ON LINKEDIN. GWEST@REDHAT.COM