SlideShare ist ein Scribd-Unternehmen logo
1 von 34
Downloaden Sie, um offline zu lesen
A CLOJURE FUSION OF SYMBOLIC
AND DATA DRIVEN AI
Huahai Yang, Ph.D.
Juji, Inc.
Nov. 29, 2018
Navigate : Space / Arrow Keys | - Menu | - Fullscreen | - Overview | - Blackout | - Speaker | - HelpM F O B S ?

1 / 34
Psychologist
Computer scientist
Coding in Clojure since 2012
Huahai Yang
 huahaiy 
Cofounder & CTO
Juji Inc.
© 2018 Juji, Inc.

2 / 34
JUJI BUILDS CHATBOT PLATFORM
A challenging problem
© 2018 Juji, Inc.

3 / 34
JUJI BUILDS CHATBOT PLATFORM
It is not hard to pass Turing Test, done in 70s'
PARRY: 33 psychiatrists cannot tell it from
paranoid patients
© 2018 Juji, Inc.

4 / 34
PARRY SAMPLE
© 2018 Juji, Inc.

5 / 34
JUJI BUILDS CHATBOT PLATFORM
It is harder to build useful
chatbots
interview people to collect
feedback
receive visitors to sites
screen job candidates
check up on trainees
© 2018 Juji, Inc.

6 / 34
JUJI BUILDS CHATBOT PLATFORM
When used for survey
2X completion rate
76% better quality responses
 the whole time i was doing this survey it felt like i was
talking to a friend and sharing the same common
ground. i loved that i wish it didnt have to end
 very dynamic and very fluid conversation you have
great quality thanks
© 2018 Juji, Inc.

7 / 34
JUJI APPROACH:
SYMBOLIC + DATA DRIVEN
Symbolic system as the
bones
Data-driven component
as the flesh
Done in a Clojure DSL,
REP
© 2018 Juji, Inc.

8 / 34
AI SUMMER IS BACK
Watson Jeopardy beats human
AlphaGo beats human
Many AI assistants on phone and in home
Many commercial products in enterprises
© 2018 Juji, Inc.

9 / 34
RISE OF DEEP LEARNING (DL)
Recently hugely successful
For many: AI = DL
© 2018 Juji, Inc.

10 / 34
DL SOLVES PERCEPTION PROBLEM
DL maps raw data (pixels, text characters) into:
known labels (classification)
desirable numbers (regression)
fixed length vectors (embedding)
 Perception is the organization, identification, and
interpretation of sensory information in order to
represent and understand the presented information, or
the environment.
© 2018 Juji, Inc.

11 / 34
PERCEPTION FEELS LIKE INTELLIGENCE
Reporter: "How many moves do you see ahead
while playing chess?"
Capablanca: "Only one, but it's always the right
one."
© 2018 Juji, Inc.

12 / 34
PERCEPTION IS NOT YET INTELLIGENCE
© 2018 Juji, Inc.

13 / 34
INTELLIGENCE CANNOT BE SOLVED WITH
DATA ALONE
Bottom-up
data driven
sub-symbolic
Top-down
goal/hypothesis driven
symbolic (human-readable)
© 2018 Juji, Inc.

14 / 34
TIME TO BRING BACK SYMBOLIC AI
(Semi-)solving perception lays the foundation for
symbolic AI
The same forces leading to the rise of DL apply to
symbolic AI
More powerful hardware, help graph search
More abundant realistic data, help knowledge
base construction
Better so ware tools and practices
© 2018 Juji, Inc.

15 / 34
TRADE-OFFS
Data driven
Easy to defeat/abuse by adversaries (e.g. Tay)
Hard to debug and bend it to the creator's will
By design, unlikely to be fixable
Symbolic
Easy to build rigid/brittle systems
Hard to develop, hard for human to think like
machines
In principle fixable, in practice, not so easy
© 2018 Juji, Inc.

16 / 34
DL IS FOOLED
© 2018 Juji, Inc.

17 / 34
GIBBON
© 2018 Juji, Inc.

18 / 34
TWO ROADS TO INTEGRATION
Implement symbolic phenomenon with sub-
symbolic system
Mimic brain
Not yet practical
Symbolic + sub-symbolic
Engineer's method
Practical today
© 2018 Juji, Inc.

19 / 34
SYMBOLIC + DATA DRIVEN
Symbolic system as the bones
for its potential for growth and adaptability,
despite the rigidity
DL/ML components as the flesh
for its flexibility and ease of development, despite
the obscurity
© 2018 Juji, Inc.

20 / 34
JUJI ARCHITECTURE
© 2018 Juji, Inc.

21 / 34
EDN DATA ALL THE WAY
1 User select chat template
2 User configure chat in GUI
3 Generate script from GUI
4 Chat: script compiles and runs
© 2018 Juji, Inc.

22 / 34
DEFTOPIC: THE BUILDING BLOCK
Topic: a set of rules
01 (deftopic hello-world ; topic name
02 [] ; parameters
03
04 [] ; trigger
05 ["Hello world!"]) ; action
© 2018 Juji, Inc.

23 / 34
PRODUCTION RULE
Rule: trigger (if), action (then) and associated
followup topics
Followup topics are primed when a rule fired
01 [:1 hello hi hey howdy] ; alternative pattern trigger
02 ["Nice to meet you!"] ; action: a string output
03 (talk-about-wheather) ; followup topic invocation
© 2018 Juji, Inc.

24 / 34
TOPIC COMPOSITIONS
A topic may include rules of other topics
01 (deftopic greetings
02 []
03 {:include-before [(morning-greetings)
04 (evening-greetings)]}
05
06 [:1 hello hi hey howdy]
07 ["Hello"])
© 2018 Juji, Inc.

25 / 34
PATTERNS
Token based regular expressions
01 [I love :1-. pizza] ; sequence with a wild car
02 [love ? [:1- pizza bacon]] ; nested, inner alternativ
03 [:0. "I love pizza"] ; start and string pattern
© 2018 Juji, Inc.

26 / 34
ML BASED TAG AND CLASS PATTERNS
Tags for annotating text
keywords for placeholders of content classes
01 [he #pos/verb dog tree] ; parts of speech tag
02 [she love :phrase/NP] ; noun phrase class
03 [I work at :entity/org] ; organization entity class
04 [did in :entity/duration] ; duration entity class
© 2018 Juji, Inc.

27 / 34
DL/ML FOR CLASSIFICATION FUNCTIONS
Neural networks are universal function
approximator, should be used as such
Patterns are and together in a rule
Rules are or together, so a topic matches a DNF
01 [programming
02 (input-in-this-category? "self-intro-relevance" 0.7)
03 ["You must be a smart person"]
© 2018 Juji, Inc.

28 / 34
DL FOR SIMILARITY BASED MATCHES
Calculate similarity using Tensorflow sentence
embedding
01 [(> (max-similarity-score
02 ["What does your product cost?"
03 "How much does your product cost?"
04 "What's the price of your product?"
05 "How expensive is your product?"])
06 0.9)]
© 2018 Juji, Inc.

29 / 34
ROLES
ML/DL models cover broad cases
Symbolic covers specific cases
misses by DL/ML
detailed refinement
01 [(input-in-this-category? "self-intro-relevance"
02 0.7)]
03 ([programming]
04 "You must be a smart person."
05
06 [art]
07 "I enjoy art too."
08
09 "Thank you for the introduction.")
© 2018 Juji, Inc.

30 / 34
META-CIRCULARITY
Turn a topic into a function, then use the function in
another topic
01 [(create-topic-func
02 custom/why-u-here :extract-why-u-here)
03 "I see, you are here to "
04 (exec-topic-func :extract-why-u-here)]
© 2018 Juji, Inc.

31 / 34
AUTOMATIC DIALOG MANAGEMENT
REP is a declarative language
System pushes topics around
Agenda queue
Ad-lib queue
Exception queue
Main stack
© 2018 Juji, Inc.

32 / 34
CONCLUSION
Symbolic + data driven = practical AI today
Clojure is a great choice for doing so
Lisp was and still is the language of symbolic AI
Data orientation of Clojure makes it easy to
integrate both
© 2018 Juji, Inc.

33 / 34
Huahai Yang
Juji, Inc.
https://juji.io
© 2018 Juji, Inc.

34 / 34

Weitere ähnliche Inhalte

Ähnlich wie A Clojure Fusion of Symbolic and Data Driven AI

Software Engineering Project Report. Help for Admission Test Students (HATS).
Software Engineering Project Report. Help for Admission Test Students (HATS).Software Engineering Project Report. Help for Admission Test Students (HATS).
Software Engineering Project Report. Help for Admission Test Students (HATS).Md. Shafaat Jamil Rokon
 
17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf
17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf
17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdfBrunoAtti1
 
IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...
IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...
IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...Niklas Heidloff
 
20191030 bitas hayashi mamoru_english
20191030 bitas hayashi mamoru_english20191030 bitas hayashi mamoru_english
20191030 bitas hayashi mamoru_englishit-innovation
 
How Business Acceleration Cloud is Revolutionizing Software Sales
How Business Acceleration Cloud is Revolutionizing Software SalesHow Business Acceleration Cloud is Revolutionizing Software Sales
How Business Acceleration Cloud is Revolutionizing Software SalesCloudShare
 
Hashitalks 2021 Infrastructure Drift & Driftctl
Hashitalks 2021 Infrastructure Drift & Driftctl Hashitalks 2021 Infrastructure Drift & Driftctl
Hashitalks 2021 Infrastructure Drift & Driftctl Stephane Jourdan
 
An illustrated guide to microservices (boston python meetup - Aug 2016)
An illustrated guide to microservices (boston python meetup - Aug 2016)An illustrated guide to microservices (boston python meetup - Aug 2016)
An illustrated guide to microservices (boston python meetup - Aug 2016)Ambassador Labs
 
The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...
The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...
The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...Neo4j
 
Making PHP Smarter - Dutch PHP 2023.pptx
Making PHP Smarter - Dutch PHP 2023.pptxMaking PHP Smarter - Dutch PHP 2023.pptx
Making PHP Smarter - Dutch PHP 2023.pptxAdam Englander
 
Break up the Monolith: Testing Microservices
Break up the Monolith: Testing MicroservicesBreak up the Monolith: Testing Microservices
Break up the Monolith: Testing MicroservicesMarcus Merrell
 
Soc biz to improve stakeholder comms (devdays13)
Soc biz to improve stakeholder comms (devdays13)Soc biz to improve stakeholder comms (devdays13)
Soc biz to improve stakeholder comms (devdays13)Thorsten Gau
 
Fun and Games Linux Game Servers
Fun and Games Linux Game ServersFun and Games Linux Game Servers
Fun and Games Linux Game Serversevolutionaryit
 
Free and Open Source Software for Geoinformatics in Geosciences Operative an...
Free and Open Source Software for  Geoinformatics in Geosciences Operative an...Free and Open Source Software for  Geoinformatics in Geosciences Operative an...
Free and Open Source Software for Geoinformatics in Geosciences Operative an...Carlos Gabriel Asato
 
IT Business Analyst (NTP, PG, 08.10.2013)
IT Business Analyst (NTP, PG, 08.10.2013)IT Business Analyst (NTP, PG, 08.10.2013)
IT Business Analyst (NTP, PG, 08.10.2013)JIT Solutions
 
A short study on telecom information models & offerings
A short study on telecom information models & offeringsA short study on telecom information models & offerings
A short study on telecom information models & offeringsSayak Majumder
 
Jboss jbpm and drools 1 introduction to drools architecture
Jboss jbpm and drools   1 introduction to drools architectureJboss jbpm and drools   1 introduction to drools architecture
Jboss jbpm and drools 1 introduction to drools architectureZoran Hristov
 
Ricoh generates new ideas with lateral communication
Ricoh generates new ideas with lateral communicationRicoh generates new ideas with lateral communication
Ricoh generates new ideas with lateral communicationIBM Software India
 
Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...
Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...
Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...Pedro Moreira da Silva
 

Ähnlich wie A Clojure Fusion of Symbolic and Data Driven AI (20)

Lightweight DITA
Lightweight DITALightweight DITA
Lightweight DITA
 
Software Engineering Project Report. Help for Admission Test Students (HATS).
Software Engineering Project Report. Help for Admission Test Students (HATS).Software Engineering Project Report. Help for Admission Test Students (HATS).
Software Engineering Project Report. Help for Admission Test Students (HATS).
 
17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf
17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf
17h25_closing_keynote_stefano_stinchi_-_innovation_story.pdf
 
IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...
IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...
IBM Connect 2013 - BP212: Apps, Apps and more Apps: Meet the Very Best Open S...
 
20191030 bitas hayashi mamoru_english
20191030 bitas hayashi mamoru_english20191030 bitas hayashi mamoru_english
20191030 bitas hayashi mamoru_english
 
How Business Acceleration Cloud is Revolutionizing Software Sales
How Business Acceleration Cloud is Revolutionizing Software SalesHow Business Acceleration Cloud is Revolutionizing Software Sales
How Business Acceleration Cloud is Revolutionizing Software Sales
 
Hashitalks 2021 Infrastructure Drift & Driftctl
Hashitalks 2021 Infrastructure Drift & Driftctl Hashitalks 2021 Infrastructure Drift & Driftctl
Hashitalks 2021 Infrastructure Drift & Driftctl
 
An illustrated guide to microservices (boston python meetup - Aug 2016)
An illustrated guide to microservices (boston python meetup - Aug 2016)An illustrated guide to microservices (boston python meetup - Aug 2016)
An illustrated guide to microservices (boston python meetup - Aug 2016)
 
Greenslade togaf2
Greenslade togaf2Greenslade togaf2
Greenslade togaf2
 
The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...
The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...
The perfect couple: Uniting Large Language Models and Knowledge Graphs for En...
 
Making PHP Smarter - Dutch PHP 2023.pptx
Making PHP Smarter - Dutch PHP 2023.pptxMaking PHP Smarter - Dutch PHP 2023.pptx
Making PHP Smarter - Dutch PHP 2023.pptx
 
Break up the Monolith: Testing Microservices
Break up the Monolith: Testing MicroservicesBreak up the Monolith: Testing Microservices
Break up the Monolith: Testing Microservices
 
Soc biz to improve stakeholder comms (devdays13)
Soc biz to improve stakeholder comms (devdays13)Soc biz to improve stakeholder comms (devdays13)
Soc biz to improve stakeholder comms (devdays13)
 
Fun and Games Linux Game Servers
Fun and Games Linux Game ServersFun and Games Linux Game Servers
Fun and Games Linux Game Servers
 
Free and Open Source Software for Geoinformatics in Geosciences Operative an...
Free and Open Source Software for  Geoinformatics in Geosciences Operative an...Free and Open Source Software for  Geoinformatics in Geosciences Operative an...
Free and Open Source Software for Geoinformatics in Geosciences Operative an...
 
IT Business Analyst (NTP, PG, 08.10.2013)
IT Business Analyst (NTP, PG, 08.10.2013)IT Business Analyst (NTP, PG, 08.10.2013)
IT Business Analyst (NTP, PG, 08.10.2013)
 
A short study on telecom information models & offerings
A short study on telecom information models & offeringsA short study on telecom information models & offerings
A short study on telecom information models & offerings
 
Jboss jbpm and drools 1 introduction to drools architecture
Jboss jbpm and drools   1 introduction to drools architectureJboss jbpm and drools   1 introduction to drools architecture
Jboss jbpm and drools 1 introduction to drools architecture
 
Ricoh generates new ideas with lateral communication
Ricoh generates new ideas with lateral communicationRicoh generates new ideas with lateral communication
Ricoh generates new ideas with lateral communication
 
Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...
Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...
Simplifying complexity at GitLab (2023-07-31 @ OutSystems Product Design Unwr...
 

Kürzlich hochgeladen

Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 3652toLead Limited
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerThousandEyes
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonetsnaman860154
 
Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Paola De la Torre
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking MenDelhi Call girls
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitecturePixlogix Infotech
 
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Alan Dix
 
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of ServiceCNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Servicegiselly40
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsMaria Levchenko
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonAnna Loughnan Colquhoun
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | DelhiFULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhisoniya singh
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Drew Madelung
 
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...gurkirankumar98700
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountPuma Security, LLC
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Igalia
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking MenDelhi Call girls
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfEnterprise Knowledge
 

Kürzlich hochgeladen (20)

Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 
Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC Architecture
 
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Swan(sea) Song – personal research during my six years at Swansea ... and bey...
 
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of ServiceCNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Service
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | DelhiFULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
FULL ENJOY 🔝 8264348440 🔝 Call Girls in Diplomatic Enclave | Delhi
 
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
 
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
Kalyanpur ) Call Girls in Lucknow Finest Escorts Service 🍸 8923113531 🎰 Avail...
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path Mount
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
 
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdfThe Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
The Role of Taxonomy and Ontology in Semantic Layers - Heather Hedden.pdf
 

A Clojure Fusion of Symbolic and Data Driven AI

  • 1. A CLOJURE FUSION OF SYMBOLIC AND DATA DRIVEN AI Huahai Yang, Ph.D. Juji, Inc. Nov. 29, 2018 Navigate : Space / Arrow Keys | - Menu | - Fullscreen | - Overview | - Blackout | - Speaker | - HelpM F O B S ?  1 / 34
  • 2. Psychologist Computer scientist Coding in Clojure since 2012 Huahai Yang  huahaiy  Cofounder & CTO Juji Inc. © 2018 Juji, Inc.  2 / 34
  • 3. JUJI BUILDS CHATBOT PLATFORM A challenging problem © 2018 Juji, Inc.  3 / 34
  • 4. JUJI BUILDS CHATBOT PLATFORM It is not hard to pass Turing Test, done in 70s' PARRY: 33 psychiatrists cannot tell it from paranoid patients © 2018 Juji, Inc.  4 / 34
  • 5. PARRY SAMPLE © 2018 Juji, Inc.  5 / 34
  • 6. JUJI BUILDS CHATBOT PLATFORM It is harder to build useful chatbots interview people to collect feedback receive visitors to sites screen job candidates check up on trainees © 2018 Juji, Inc.  6 / 34
  • 7. JUJI BUILDS CHATBOT PLATFORM When used for survey 2X completion rate 76% better quality responses  the whole time i was doing this survey it felt like i was talking to a friend and sharing the same common ground. i loved that i wish it didnt have to end  very dynamic and very fluid conversation you have great quality thanks © 2018 Juji, Inc.  7 / 34
  • 8. JUJI APPROACH: SYMBOLIC + DATA DRIVEN Symbolic system as the bones Data-driven component as the flesh Done in a Clojure DSL, REP © 2018 Juji, Inc.  8 / 34
  • 9. AI SUMMER IS BACK Watson Jeopardy beats human AlphaGo beats human Many AI assistants on phone and in home Many commercial products in enterprises © 2018 Juji, Inc.  9 / 34
  • 10. RISE OF DEEP LEARNING (DL) Recently hugely successful For many: AI = DL © 2018 Juji, Inc.  10 / 34
  • 11. DL SOLVES PERCEPTION PROBLEM DL maps raw data (pixels, text characters) into: known labels (classification) desirable numbers (regression) fixed length vectors (embedding)  Perception is the organization, identification, and interpretation of sensory information in order to represent and understand the presented information, or the environment. © 2018 Juji, Inc.  11 / 34
  • 12. PERCEPTION FEELS LIKE INTELLIGENCE Reporter: "How many moves do you see ahead while playing chess?" Capablanca: "Only one, but it's always the right one." © 2018 Juji, Inc.  12 / 34
  • 13. PERCEPTION IS NOT YET INTELLIGENCE © 2018 Juji, Inc.  13 / 34
  • 14. INTELLIGENCE CANNOT BE SOLVED WITH DATA ALONE Bottom-up data driven sub-symbolic Top-down goal/hypothesis driven symbolic (human-readable) © 2018 Juji, Inc.  14 / 34
  • 15. TIME TO BRING BACK SYMBOLIC AI (Semi-)solving perception lays the foundation for symbolic AI The same forces leading to the rise of DL apply to symbolic AI More powerful hardware, help graph search More abundant realistic data, help knowledge base construction Better so ware tools and practices © 2018 Juji, Inc.  15 / 34
  • 16. TRADE-OFFS Data driven Easy to defeat/abuse by adversaries (e.g. Tay) Hard to debug and bend it to the creator's will By design, unlikely to be fixable Symbolic Easy to build rigid/brittle systems Hard to develop, hard for human to think like machines In principle fixable, in practice, not so easy © 2018 Juji, Inc.  16 / 34
  • 17. DL IS FOOLED © 2018 Juji, Inc.  17 / 34
  • 18. GIBBON © 2018 Juji, Inc.  18 / 34
  • 19. TWO ROADS TO INTEGRATION Implement symbolic phenomenon with sub- symbolic system Mimic brain Not yet practical Symbolic + sub-symbolic Engineer's method Practical today © 2018 Juji, Inc.  19 / 34
  • 20. SYMBOLIC + DATA DRIVEN Symbolic system as the bones for its potential for growth and adaptability, despite the rigidity DL/ML components as the flesh for its flexibility and ease of development, despite the obscurity © 2018 Juji, Inc.  20 / 34
  • 21. JUJI ARCHITECTURE © 2018 Juji, Inc.  21 / 34
  • 22. EDN DATA ALL THE WAY 1 User select chat template 2 User configure chat in GUI 3 Generate script from GUI 4 Chat: script compiles and runs © 2018 Juji, Inc.  22 / 34
  • 23. DEFTOPIC: THE BUILDING BLOCK Topic: a set of rules 01 (deftopic hello-world ; topic name 02 [] ; parameters 03 04 [] ; trigger 05 ["Hello world!"]) ; action © 2018 Juji, Inc.  23 / 34
  • 24. PRODUCTION RULE Rule: trigger (if), action (then) and associated followup topics Followup topics are primed when a rule fired 01 [:1 hello hi hey howdy] ; alternative pattern trigger 02 ["Nice to meet you!"] ; action: a string output 03 (talk-about-wheather) ; followup topic invocation © 2018 Juji, Inc.  24 / 34
  • 25. TOPIC COMPOSITIONS A topic may include rules of other topics 01 (deftopic greetings 02 [] 03 {:include-before [(morning-greetings) 04 (evening-greetings)]} 05 06 [:1 hello hi hey howdy] 07 ["Hello"]) © 2018 Juji, Inc.  25 / 34
  • 26. PATTERNS Token based regular expressions 01 [I love :1-. pizza] ; sequence with a wild car 02 [love ? [:1- pizza bacon]] ; nested, inner alternativ 03 [:0. "I love pizza"] ; start and string pattern © 2018 Juji, Inc.  26 / 34
  • 27. ML BASED TAG AND CLASS PATTERNS Tags for annotating text keywords for placeholders of content classes 01 [he #pos/verb dog tree] ; parts of speech tag 02 [she love :phrase/NP] ; noun phrase class 03 [I work at :entity/org] ; organization entity class 04 [did in :entity/duration] ; duration entity class © 2018 Juji, Inc.  27 / 34
  • 28. DL/ML FOR CLASSIFICATION FUNCTIONS Neural networks are universal function approximator, should be used as such Patterns are and together in a rule Rules are or together, so a topic matches a DNF 01 [programming 02 (input-in-this-category? "self-intro-relevance" 0.7) 03 ["You must be a smart person"] © 2018 Juji, Inc.  28 / 34
  • 29. DL FOR SIMILARITY BASED MATCHES Calculate similarity using Tensorflow sentence embedding 01 [(> (max-similarity-score 02 ["What does your product cost?" 03 "How much does your product cost?" 04 "What's the price of your product?" 05 "How expensive is your product?"]) 06 0.9)] © 2018 Juji, Inc.  29 / 34
  • 30. ROLES ML/DL models cover broad cases Symbolic covers specific cases misses by DL/ML detailed refinement 01 [(input-in-this-category? "self-intro-relevance" 02 0.7)] 03 ([programming] 04 "You must be a smart person." 05 06 [art] 07 "I enjoy art too." 08 09 "Thank you for the introduction.") © 2018 Juji, Inc.  30 / 34
  • 31. META-CIRCULARITY Turn a topic into a function, then use the function in another topic 01 [(create-topic-func 02 custom/why-u-here :extract-why-u-here) 03 "I see, you are here to " 04 (exec-topic-func :extract-why-u-here)] © 2018 Juji, Inc.  31 / 34
  • 32. AUTOMATIC DIALOG MANAGEMENT REP is a declarative language System pushes topics around Agenda queue Ad-lib queue Exception queue Main stack © 2018 Juji, Inc.  32 / 34
  • 33. CONCLUSION Symbolic + data driven = practical AI today Clojure is a great choice for doing so Lisp was and still is the language of symbolic AI Data orientation of Clojure makes it easy to integrate both © 2018 Juji, Inc.  33 / 34
  • 34. Huahai Yang Juji, Inc. https://juji.io © 2018 Juji, Inc.  34 / 34