SlideShare a Scribd company logo
1 of 50
Download to read offline
Error-proof Machine Learning with explainability and human-centric design
WE ARE AN END-TO-END ARTIFICIAL
INTELLIGENCE SOFTWARE FACTORY
Human-in-the-loop
AI systems inevitably lose accuracy
over time. Brainjar builds systems that
are able to continually improve by
working together with their human
colleagues.
Technology agnostic
Because of our broad technology
expertise, we always choose the best
solution for problem. This can be AI
API, pre-trained model or custom
model and is always cloud agnostic.
End-to-end
Brainjar works end-to-end with the
customer to build and integrate the AI
solution. We start with a data analysis,
followed by labeling, model training,
front-end building and deployment.
WE ARE PART OF
THE RACCOONS GROUP
Software
prototyping
AI end-to-end
Chatbot framework
Blockchain services
Full stack Javascript
Quantum computing
Edge AI
Digital production house
AI Business
Strategy
AI is...
NOT new (but has changed)
Early AI (1950’s – 2000’s)
• Deterministic, human-made algorithms
• Centered around planning or abstract reasoning
• Scaled horribly
• Required maintenance of structured “knowledge
systems”
Modern AI (2000’s – Now)
• “Learning” algorithms -> Machine Learning
• Centered around human senses and prediction
• Can scale with data
• Can process unstructured data
Source: Stanford news
Probabilistic
(Modern) AI makes mistakes!
• ~95-99% accuracy is typical (depending on usecase)
• Requires strategy for 5% (human review)
• Algorithms can provide “confidence level”, allowing
humans to only review edge cases
Source: wikipedia
LESSON 1
Think about mistakes
NOT Self-Learning!
Machine Learning learns from VERIFIED data
• Many tasks require human verification for the
system to learn & improve
• Initial training can be done using historic data
• Human feedback still required during production
Source: Medium
NOT Self-Learning!
support question central support domain experts support question ML algorithm domain experts support question ML algorithm domain experts
Example: Mail Forwarding
(1) (2) (3)
LESSON 2
Learn from your users
Human-Centered Machine Learning
Cooperation benefits all parties
• AI eliminates tedium
• Humans focus on edge cases
• Two key principle builds trust with management
• High quality data is produced
• AI improves/adapts continuously
Source: wikipedia
LESSON 3
Learn your users
Our work
CASE 1
ORDER ADMINSTRATION
AT TVH
TVH Group is a Belgian company active in the market of forklift trucks, aerial work platforms, agricultural tractors and industrial vehicles.
The company has an order department that handles orders from customers via email.
TVH has a large order administration
department.
O R D E R A D M I N I S T R AT I O N
Customers can place orders by sending an
email with their request to TVH.
E M A I L O R D E R S
The rental administrators handle these
requests by reading the mails and
registering the order in the ERP system.
R E P E T I T I V E W O R K
A R T I F I C I A L I N T E L L I G E N C E
Determine if the email is a rental
order or not & extracts all entities
in the email
Present an intuitive verification
step for users with automatic
retraining
Structured order is placed in the
ERP system
C O N N E C T T O
M A I L B O X
SOLUTION
Connects with existing mailbox &
gets triggered when there is an
incoming mail
V E R I F I C A T I O N P L A C E O R D E R
Customers of TVH send orders via email to the order administration. A typical example is shown below.
CONNECT TO MAILBOX
V E R I F I C A T I O N P L A C E O R D E RC O N N E C T T O M A I L B O X
A R T I F I C I A L
I N T E L L I G E N C E
SOLUTION
Determine if the email is a rental
order or not & extracts all entities
in the email
Present an intuitive verification
step for users with automatic
retraining
Structured order is placed in the
ERP system
Connects with existing mailbox &
gets triggered when there is an
incoming mail
Automatically determines if the email is a rental order & extracts all entities in the email.
ARTIFICIAL INTELLIGENCE
P L A C E O R D E R
V E R I F I C AT I O N
SOLUTION
Determine if the email is a rental
order or not & extracts all entities
in the email
Present an intuitive verification
step for users with automatic
retraining
Structured order is placed in the
ERP system
Connects with existing mailbox &
gets triggered when there is an
incoming mail
C O N N E C T T O M A I L B O X A R T I F I C I A L I N T E L L I G E N C E
P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g .
VERIFICATION
P L A C E O R D E R
S t r u c t u r e d o r d e r i s p l a c e d
i n t h e E R P s y s t e m
P L A C E O R D E R
SOLUTION
Determine if the email is a rental
order or not & extracts all entities
in the email
Present an intuitive verification
step for users with automatic
retraining
Structured order is placed in the
ERP system
Connects with existing mailbox &
gets triggered when there is an
incoming mail
C O N N E C T T O M A I L B O X A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N
BUSINESS CASE
The current order administration in each branch of TVH has 7 full time equivalents handling over 400 incoming orders via email per day.
5 MIN
T I M E P E R E M A I L
1 MIN
BEFORE:
AFTER:
80%
T I M E R E D U C T I O N
THE KLASSIF.AI SOLUTION
K l a s s i f . a i h a s i d e n t i f i e d f o u r k e y c o m p o n e n t s t h a t a l l o w o r g a n i s a t i o n s t o l e v e r a g e m a c h i n e l e a r n i n g t o
i m p r o v e t h e i r d o c u m e n t p r o c e s s i n g w o r k f l o w s . C o m b i n i n g t h e s e c o m p o n e n t s a l l o w s K l a s s i f . a i t o p r o c e s s
d o c u m e n t s a t c o m p u t e r s p e e d s w i t h h u m a n r e l i a b i l i t y .
For your experts, it’s business as usual.
Meanwhile, Klassif.ai is gathering training
data to learn the ins and outs of your
documents. The result is a training set that
represents the real world.
L A B E L I N G
Just like any mature software project
needs versioning and logging, so does a
machine learning model! Klassif.ai provides
a model history, along with statistics that
show you the model’s performance on
real, unseen data.
G O V E R N A N C E
Klassif.ai integrates seamlessly into your
current document processing workflow,
enriching your unstructured documents and
connecting them to your structured systems.
I N T E G R AT I O N
Every document type requires its own
approach. That’s why Klassif.ai classifies each
document before extracting the required
information. After all, an invoice needs to be
processed differently to an employee
contract.
C L A S S I F I C AT I O N &
E X T R A C T I O N
KLASSIF.AI
Platform features
OCR USER MANAGEMENT
AUTOMATIC
RETRAINING
SECURITY NOTIFICATIONS
ADMIN PANEL
MULTILINGUAL
SUPPORT
SCALABILITY
MODULAR AI MODELS ON-PREMISE OR IN THE CLOUD
PROJECT
APPROACH
K L A S S I F . A I
We map and visualise your business
processes
1 . I N TA K E
Your domain experts generate data while
doing their work as usual
4 . L A B E L
We customise our platform based on your
way of working
3 . C U S T O M I S E
We integrate our services with your
systems
2 . I N T E G R AT E
We train a machine learning algorithm
that meets your needs
5 . T R A I N M O D E L
We ready the platform for your experts
6 . D E P L O Y
P O C L E A D T I M E :
1 - 3 M O N T H S
P R O D U C T I O N R E A D Y:
2 - 5 M O N T H S
CASE 2
HEALTHCARE DATA EXTRACTION
AT RIZIV
RIZIV is the Belgian federal institute for health and disability insurance. The institute is, among others, responsible for handling files on
medical accidents.
Hospitals scan all case documents and
email them to RIZIV. An administrator at
RIZIV makes sure the case is complete.
C A S E D O C U M E N T S
A doctor from RIZIV goes over all case
documents and decides on refund of
medical costs.
R E V I E W B Y D O C T O R
An intern prints all case documents and
sorts these on date before scanning them
again.
M A N U A L W O R K
O C R A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N
B Y D O C T O R
Convert scanned document to
computer readable text
Extract all entities such as, doctor
name, procedure date and
procedure type from document
Present an intuitive verification
step for doctors with automatic
retraining
U P L O A D
D O C U M E N T S
SOLUTION
Doctors can upload all case
documents into application
U P L O A D D O C U M E N T S
O C R
SOLUTION
Convert scanned document to
computer readable text
Extract all entities such as, doctor
name, procedure date and
procedure type from document
Present an intuitive verification
step for doctors with automatic
retraining
Doctors can upload all case
documents into application
A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N
B Y D O C T O R
A R T I F I C I A L
I N T E L L I G E N C EU P L O A D D O C U M E N T S O C R
SOLUTION
Convert scanned document to
computer readable text
Extract all entities such as, doctor
name, procedure date and
procedure type from document
Present an intuitive verification
step for doctors with automatic
retraining
Doctors can upload all case
documents into application
V E R I F I C A T I O N
B Y D O C T O R
P L A C E O R D E R
S t r u c t u r e d o r d e r i s p l a c e d
i n t h e E R P s y s t e m
V E R I F I C AT I O N
B Y D O C T O RU P L O A D D O C U M E N T S O C R
SOLUTION
Convert scanned document to
computer readable text
Extract all entities such as, doctor
name, procedure date and
procedure type from document
Present an intuitive verification
step for doctors with automatic
retraining
Doctors can upload all case
documents into application
A R T I F I C I A L I N T E L L I G E N C E
Present an intuitive verification step for doctors with automatic retraining
VERIFICATION BY DOCTOR
CASE 3
DIGITALISING UNEMPLOYMENT BENEFITS
AT HVW
HVW, the Belgian relief fund for unemployment benefits, is a public institution for social security which pays unemployed citizens. Each
month HVW pays more than 120.000 benefit recipients their replacement income.
HVW has a large administration
department who processes all
unemployment benefits
B E N E F I T A D M I N I S T R AT I O N
The majority of unemployment forms are
still submitted on paper
PA P E R F O R M S
Due to human mistakes, often error are
introduced during the input step requiring
extra steps to fix mistakes
E R R O R P R O N E
H A N D W R I T I N G
R E C O G N I T I O N
A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N
C o n v e r t h a n d w r i t t e n
c o n t a c t i n f o r m a t i o n t o
c o m p u t e r r e a d a b l e t e x t
E x t r a c t c a l e n d a r
i n f o r m a t i o n f r o m p h o t o
P r e s e n t a n i n t u i t i v e
v e r i f i c a t i o n s t e p f o r u s e r s
w i t h a u t o m a t i c r e t r a i n i n g
U P L O A D
P H O T O
U s e r s c a n u p l o a d a p h o t o o f
t h e i r u n e m p l o y m e n t c a r d
SOLUTION
U s e r s c a n u p l o a d a p h o t o o f t h e i r u n e m p l o y m e n t c a r d
UPLOAD PHOTO
U P L O A D P H O T O
U s e r s c a n u p l o a d a p h o t o
o f t h e i r u n e m p l o y m e n t
c a r d
H A N D W R I T I N G
R E C O G N I T I O N
C o n v e r t h a n d w r i t t e n c o n t a c t
i n f o r m a t i o n t o c o m p u t e r
r e a d a b l e t e x t
A R T I F I C I A L I N T E L L I G E N C E
E x t r a c t c a l e n d a r
i n f o r m a t i o n f r o m p h o t o
V E R I F I C A T I O N
P r e s e n t a n i n t u i t i v e
v e r i f i c a t i o n s t e p f o r u s e r s
w i t h a u t o m a t i c r e t r a i n i n g
SOLUTION
A R T I F I C I A L
I N T E L L I G E N C E
E x t r a c t c a l e n d a r i n f o r m a t i o n
f r o m p h o t o
U P L O A D P H O T O
U s e r s c a n u p l o a d a p h o t o
o f t h e i r u n e m p l o y m e n t
c a r d
H A N D W R I T I N G
R E C O G N I T I O N
C o n v e r t h a n d w r i t t e n
c o n t a c t i n f o r m a t i o n t o
c o m p u t e r r e a d a b l e t e x t
V E R I F I C A T I O N
P r e s e n t a n i n t u i t i v e
v e r i f i c a t i o n s t e p f o r u s e r s
w i t h a u t o m a t i c r e t r a i n i n g
SOLUTION
P L A C E O R D E R
S t r u c t u r e d o r d e r i s p l a c e d
i n t h e E R P s y s t e m
V E R I F I C AT I O N
P r e s e n t a n i n t u i t i v e
v e r i f i c a t i o n s t e p f o r u s e r s
w i t h a u t o m a t i c r e t r a i n i n g
U P L O A D P H O T O
U s e r s c a n u p l o a d a p h o t o
o f t h e i r u n e m p l o y m e n t
c a r d
H A N D W R I T I N G
R E C O G N I T I O N
C o n v e r t h a n d w r i t t e n
c o n t a c t i n f o r m a t i o n t o
c o m p u t e r r e a d a b l e t e x t
A R T I F I C I A L I N T E L L I G E N C E
E x t r a c t c a l e n d a r
i n f o r m a t i o n f r o m p h o t o
SOLUTION
P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g
VERIFICATION
CASE 4
AI PRESENTATION ANALIST
In an international organisation a lot of
effort is expended in keeping the
personnel up-to-date using training
videos.
T R A I N I N G
To help train their trainers, we built an AI
system that provides feedback on their
presentation style.
A U T O M AT E F E E D B A C K
Often trainers have extensive knowledge
on their subject but are not natural born
presenters.
T R A I N T H E T R A I N E R
A R C H I T E C T U R E A R T I F I C I A L I N T E L L I G E N C E U S E R I N T E R F A C E
Our clever architecture only
enables the expensive cloud
components on demand. Audio
and video are separated for
further processing.
Several AI components are
combined to analyse the audio
and video of the presentation.
Present an intuitive user interface
for presenters to view the
feedback.
SOLUTION
U P L O A D V I D E O
Users can upload their videos to
our application for processing
U s e r s c a n v i e w t h e f e e d b a c k a b o u t t h e i r p r e s e n t a t i o n
USER INTERFACE
U s e r s c a n v i e w t h e f e e d b a c k a b o u t t h e i r p r e s e n t a t i o n
USER INTERFACE
U s e r s c a n v i e w t h e f e e d b a c k a b o u t t h e i r p r e s e n t a t i o n
USER INTERFACE
And what about explainability?
Explainability is mainly relevant for decision making
AI (which we haven’t been asked to build yet)
• Still early days
• Be careful about feedback to external users!
➢ Micro loan approval experiment
Source: wikipedia
The choice among algorithm categories can partially be made based on the user persona's
ability to intervene at different parts of a machine learning pipeline.
If the user is allowed to modify the training data, then pre-processing can be used.
If the user is allowed to change the learning algorithm, then in-processing can be used.
If the user can only treat the learned model as a black box without any ability to modify the
training data or learning algorithm, then only post-processing can be used.
AIF360 recommends the earliest mediation category in the pipeline that the user has
permission to apply because it gives the most flexibility and opportunity to correct bias as
much as possible. If possible, all algorithms from all permissible categories should be tested
because the ultimate performance depends on dataset characteristics: there is no one best
algorithm independent of dataset
NEED FOR A HUMAN IN THE LINK?
LET’S GET STARTED!
CONTACT US STAY IN TOUCH
info@brainjar.ai
+32 (0) 473 36 15 55
Gaston Geenslaan 11, B4
3001 Leuven
https://brainjar.ai
https://twitter.com/brainjarai
https://nl.linkedin.com/company/brainjarbe

More Related Content

Similar to Implementing error-proof, business-critical Machine Learning, presentation by Deevid De Meyer, co-founder of Brainjar, presented at the Trustworthy and Ethical AI Conference on Feb 13th in Brussels

SAI - Serverless Integration Architectures - 09/2019
SAI - Serverless Integration Architectures - 09/2019SAI - Serverless Integration Architectures - 09/2019
SAI - Serverless Integration Architectures - 09/2019Samuel Vandecasteele
 
How to Transform Into a Data-Driven Organization
How to Transform Into a Data-Driven OrganizationHow to Transform Into a Data-Driven Organization
How to Transform Into a Data-Driven OrganizationWarrenCruz3
 
Advanced Codeless Testing for Web Apps
Advanced Codeless Testing for Web AppsAdvanced Codeless Testing for Web Apps
Advanced Codeless Testing for Web AppsPerfecto by Perforce
 
ERP presentation ejaz ahmed bhatti
ERP  presentation  ejaz ahmed bhattiERP  presentation  ejaz ahmed bhatti
ERP presentation ejaz ahmed bhattiEjaz Bhatti
 
The future of customer service handout
The future of customer service handoutThe future of customer service handout
The future of customer service handoutMartin Hill-Wilson
 
Microsoft DevOps Journey
Microsoft DevOps JourneyMicrosoft DevOps Journey
Microsoft DevOps JourneyMayank Srivastava
 
Microsoft Project Server/Online Addons
Microsoft Project Server/Online AddonsMicrosoft Project Server/Online Addons
Microsoft Project Server/Online AddonsPJ Mistry
 
Unify Corporate Profile
Unify Corporate ProfileUnify Corporate Profile
Unify Corporate ProfileAnuragBhimrajka1
 
Realising the true value of DevOps
Realising the true value of DevOpsRealising the true value of DevOps
Realising the true value of DevOpstlevey
 
Alert Framework - Alert your organization to errors, changes, and stalled tra...
Alert Framework - Alert your organization to errors, changes, and stalled tra...Alert Framework - Alert your organization to errors, changes, and stalled tra...
Alert Framework - Alert your organization to errors, changes, and stalled tra...Smart ERP Solutions, Inc.
 
Meteor - not just for rockstars
Meteor - not just for rockstarsMeteor - not just for rockstars
Meteor - not just for rockstarsStephan Hochhaus
 
Innovation-at-Hyper-scale-Outlook-on-Emerging-Technologies
Innovation-at-Hyper-scale-Outlook-on-Emerging-TechnologiesInnovation-at-Hyper-scale-Outlook-on-Emerging-Technologies
Innovation-at-Hyper-scale-Outlook-on-Emerging-TechnologiesAmazon Web Services
 
Power Matrix Presentation1
Power Matrix Presentation1Power Matrix Presentation1
Power Matrix Presentation1Anuj Agarwal
 
Power Matrix Presentation1
Power Matrix Presentation1Power Matrix Presentation1
Power Matrix Presentation1Anuj Agarwal
 
Softengi - Inspired Software Engineering
Softengi - Inspired Software EngineeringSoftengi - Inspired Software Engineering
Softengi - Inspired Software EngineeringSoftengi
 
Neoaug 2013 critical success factors for data quality management-chain-sys-co...
Neoaug 2013 critical success factors for data quality management-chain-sys-co...Neoaug 2013 critical success factors for data quality management-chain-sys-co...
Neoaug 2013 critical success factors for data quality management-chain-sys-co...Chain Sys Corporation
 
DevOpsRoadTrip San Francisco Final Speaking Deck
DevOpsRoadTrip San Francisco Final Speaking Deck DevOpsRoadTrip San Francisco Final Speaking Deck
DevOpsRoadTrip San Francisco Final Speaking Deck VictorOps
 
Asp Abstracts, Sample Copy 15+ Abstracts
Asp Abstracts, Sample Copy 15+ AbstractsAsp Abstracts, Sample Copy 15+ Abstracts
Asp Abstracts, Sample Copy 15+ Abstractsncct
 

Similar to Implementing error-proof, business-critical Machine Learning, presentation by Deevid De Meyer, co-founder of Brainjar, presented at the Trustworthy and Ethical AI Conference on Feb 13th in Brussels (20)

SAI - Serverless Integration Architectures - 09/2019
SAI - Serverless Integration Architectures - 09/2019SAI - Serverless Integration Architectures - 09/2019
SAI - Serverless Integration Architectures - 09/2019
 
Skillaware Intro slide deck - U.S. launch
Skillaware Intro slide deck - U.S. launchSkillaware Intro slide deck - U.S. launch
Skillaware Intro slide deck - U.S. launch
 
How to Transform Into a Data-Driven Organization
How to Transform Into a Data-Driven OrganizationHow to Transform Into a Data-Driven Organization
How to Transform Into a Data-Driven Organization
 
Advanced Codeless Testing for Web Apps
Advanced Codeless Testing for Web AppsAdvanced Codeless Testing for Web Apps
Advanced Codeless Testing for Web Apps
 
ERP presentation ejaz ahmed bhatti
ERP  presentation  ejaz ahmed bhattiERP  presentation  ejaz ahmed bhatti
ERP presentation ejaz ahmed bhatti
 
The future of customer service handout
The future of customer service handoutThe future of customer service handout
The future of customer service handout
 
Microsoft DevOps Journey
Microsoft DevOps JourneyMicrosoft DevOps Journey
Microsoft DevOps Journey
 
Microsoft Project Server/Online Addons
Microsoft Project Server/Online AddonsMicrosoft Project Server/Online Addons
Microsoft Project Server/Online Addons
 
Unify Corporate Profile
Unify Corporate ProfileUnify Corporate Profile
Unify Corporate Profile
 
Realising the true value of DevOps
Realising the true value of DevOpsRealising the true value of DevOps
Realising the true value of DevOps
 
Alert Framework - Alert your organization to errors, changes, and stalled tra...
Alert Framework - Alert your organization to errors, changes, and stalled tra...Alert Framework - Alert your organization to errors, changes, and stalled tra...
Alert Framework - Alert your organization to errors, changes, and stalled tra...
 
Meteor - not just for rockstars
Meteor - not just for rockstarsMeteor - not just for rockstars
Meteor - not just for rockstars
 
Zero to Automated in Under a Year
Zero to Automated in Under a YearZero to Automated in Under a Year
Zero to Automated in Under a Year
 
Innovation-at-Hyper-scale-Outlook-on-Emerging-Technologies
Innovation-at-Hyper-scale-Outlook-on-Emerging-TechnologiesInnovation-at-Hyper-scale-Outlook-on-Emerging-Technologies
Innovation-at-Hyper-scale-Outlook-on-Emerging-Technologies
 
Power Matrix Presentation1
Power Matrix Presentation1Power Matrix Presentation1
Power Matrix Presentation1
 
Power Matrix Presentation1
Power Matrix Presentation1Power Matrix Presentation1
Power Matrix Presentation1
 
Softengi - Inspired Software Engineering
Softengi - Inspired Software EngineeringSoftengi - Inspired Software Engineering
Softengi - Inspired Software Engineering
 
Neoaug 2013 critical success factors for data quality management-chain-sys-co...
Neoaug 2013 critical success factors for data quality management-chain-sys-co...Neoaug 2013 critical success factors for data quality management-chain-sys-co...
Neoaug 2013 critical success factors for data quality management-chain-sys-co...
 
DevOpsRoadTrip San Francisco Final Speaking Deck
DevOpsRoadTrip San Francisco Final Speaking Deck DevOpsRoadTrip San Francisco Final Speaking Deck
DevOpsRoadTrip San Francisco Final Speaking Deck
 
Asp Abstracts, Sample Copy 15+ Abstracts
Asp Abstracts, Sample Copy 15+ AbstractsAsp Abstracts, Sample Copy 15+ Abstracts
Asp Abstracts, Sample Copy 15+ Abstracts
 

More from Patrick Van Renterghem

Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...
Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...
Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...Patrick Van Renterghem
 
Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...
Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...
Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...Patrick Van Renterghem
 
AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...
AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...
AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...Patrick Van Renterghem
 
Responsible AI: An Example AI Development Process with Focus on Risks and Con...
Responsible AI: An Example AI Development Process with Focus on Risks and Con...Responsible AI: An Example AI Development Process with Focus on Risks and Con...
Responsible AI: An Example AI Development Process with Focus on Risks and Con...Patrick Van Renterghem
 
Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...
Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...
Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...Patrick Van Renterghem
 
How obedient digital twins and intelligent beings contribute to ethics and ex...
How obedient digital twins and intelligent beings contribute to ethics and ex...How obedient digital twins and intelligent beings contribute to ethics and ex...
How obedient digital twins and intelligent beings contribute to ethics and ex...Patrick Van Renterghem
 
He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...
He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...
He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...Patrick Van Renterghem
 
Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...
Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...
Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...Patrick Van Renterghem
 
Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...
Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...
Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...Patrick Van Renterghem
 
Digital Workplace Case Study: How the Municipality of Duffel successfully swi...
Digital Workplace Case Study: How the Municipality of Duffel successfully swi...Digital Workplace Case Study: How the Municipality of Duffel successfully swi...
Digital Workplace Case Study: How the Municipality of Duffel successfully swi...Patrick Van Renterghem
 
Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...
Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...
Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...Patrick Van Renterghem
 
The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...
The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...
The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...Patrick Van Renterghem
 
Engie's Digital Workplace and "Connecting the company" business case, present...
Engie's Digital Workplace and "Connecting the company" business case, present...Engie's Digital Workplace and "Connecting the company" business case, present...
Engie's Digital Workplace and "Connecting the company" business case, present...Patrick Van Renterghem
 
Face your communication challenges when implementing a digital workplace, bas...
Face your communication challenges when implementing a digital workplace, bas...Face your communication challenges when implementing a digital workplace, bas...
Face your communication challenges when implementing a digital workplace, bas...Patrick Van Renterghem
 
The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...
The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...
The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...Patrick Van Renterghem
 
Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...
Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...
Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...Patrick Van Renterghem
 
Tim scottkoenverheyenpresentation
Tim scottkoenverheyenpresentationTim scottkoenverheyenpresentation
Tim scottkoenverheyenpresentationPatrick Van Renterghem
 
Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...
Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...
Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...Patrick Van Renterghem
 
Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...
Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...
Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...Patrick Van Renterghem
 
Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...
Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...
Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...Patrick Van Renterghem
 

More from Patrick Van Renterghem (20)

Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...
Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...
Ethical AI at VDAB, presented by Vincent Buekenhout (Ethical AI Lead, VDAB) a...
 
Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...
Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...
Building Trust and Explainability into Chatbots: the Partena Ziekenfonds Busi...
 
AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...
AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...
AI & Ethics: The Belgian Industry Vision & Initiatives, presentation by Jelle...
 
Responsible AI: An Example AI Development Process with Focus on Risks and Con...
Responsible AI: An Example AI Development Process with Focus on Risks and Con...Responsible AI: An Example AI Development Process with Focus on Risks and Con...
Responsible AI: An Example AI Development Process with Focus on Risks and Con...
 
Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...
Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...
Fairness and Transparency: Algorithmic Explainability, some Legal and Ethical...
 
How obedient digital twins and intelligent beings contribute to ethics and ex...
How obedient digital twins and intelligent beings contribute to ethics and ex...How obedient digital twins and intelligent beings contribute to ethics and ex...
How obedient digital twins and intelligent beings contribute to ethics and ex...
 
He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...
He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...
He Said, She Said: Finding and Fixing Bias in NLP (Natural Language Processin...
 
Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...
Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...
Introduction to Bias in Machine Learning, presented by Matthias Feys, CTO @ M...
 
Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...
Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...
Business Case: Ozitem Groupe, where 80% of the company is working remotely. R...
 
Digital Workplace Case Study: How the Municipality of Duffel successfully swi...
Digital Workplace Case Study: How the Municipality of Duffel successfully swi...Digital Workplace Case Study: How the Municipality of Duffel successfully swi...
Digital Workplace Case Study: How the Municipality of Duffel successfully swi...
 
Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...
Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...
Unleashing the Full Potential of People, Teams and SOLVAY, presented by Bruce...
 
The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...
The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...
The Building Blocks of a Digital Workplace, presented by Sam Marshall at the ...
 
Engie's Digital Workplace and "Connecting the company" business case, present...
Engie's Digital Workplace and "Connecting the company" business case, present...Engie's Digital Workplace and "Connecting the company" business case, present...
Engie's Digital Workplace and "Connecting the company" business case, present...
 
Face your communication challenges when implementing a digital workplace, bas...
Face your communication challenges when implementing a digital workplace, bas...Face your communication challenges when implementing a digital workplace, bas...
Face your communication challenges when implementing a digital workplace, bas...
 
The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...
The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...
The first steps in Recticel's Digital Workplace program by Kenneth Meuleman (...
 
Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...
Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...
Presentation by Dave Geentjens at the "Successful Digital Workplace Adoption"...
 
Tim scottkoenverheyenpresentation
Tim scottkoenverheyenpresentationTim scottkoenverheyenpresentation
Tim scottkoenverheyenpresentation
 
Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...
Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...
Presentation by Ivan Schotsmans (DV Community) at the Data Vault Modelling an...
 
Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...
Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...
Presentation by Luc Delanglez (DataLumen) at the Data Vault Modelling and Dat...
 
Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...
Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...
Presentation by Erik van der Hoeven (Wisdom as a Service) at the Data Vault M...
 

Recently uploaded

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
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxnull - The Open Security Community
 
Hyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your Budget
Hyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your BudgetHyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your Budget
Hyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your BudgetEnjoy Anytime
 
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationBeyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationSafe Software
 
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
 
Key Features Of Token Development (1).pptx
Key  Features Of Token  Development (1).pptxKey  Features Of Token  Development (1).pptx
Key Features Of Token Development (1).pptxLBM Solutions
 
Benefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other FrameworksBenefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other FrameworksSoftradix Technologies
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitecturePixlogix Infotech
 
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
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraDeakin University
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesSinan KOZAK
 
IAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsIAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsEnterprise Knowledge
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Allon Mureinik
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...shyamraj55
 
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
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking MenDelhi Call girls
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreternaman860154
 
Azure Monitor & Application Insight to monitor Infrastructure & Application
Azure Monitor & Application Insight to monitor Infrastructure & ApplicationAzure Monitor & Application Insight to monitor Infrastructure & Application
Azure Monitor & Application Insight to monitor Infrastructure & ApplicationAndikSusilo4
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsMemoori
 
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
 

Recently uploaded (20)

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
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
 
Hyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your Budget
Hyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your BudgetHyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your Budget
Hyderabad Call Girls Khairatabad ✨ 7001305949 ✨ Cheap Price Your Budget
 
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry InnovationBeyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
Beyond Boundaries: Leveraging No-Code Solutions for Industry Innovation
 
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
 
Key Features Of Token Development (1).pptx
Key  Features Of Token  Development (1).pptxKey  Features Of Token  Development (1).pptx
Key Features Of Token Development (1).pptx
 
Benefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other FrameworksBenefits Of Flutter Compared To Other Frameworks
Benefits Of Flutter Compared To Other Frameworks
 
Understanding the Laravel MVC Architecture
Understanding the Laravel MVC ArchitectureUnderstanding the Laravel MVC Architecture
Understanding the Laravel MVC Architecture
 
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
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning era
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen Frames
 
IAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsIAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI Solutions
 
Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)Injustice - Developers Among Us (SciFiDevCon 2024)
Injustice - Developers Among Us (SciFiDevCon 2024)
 
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
Automating Business Process via MuleSoft Composer | Bangalore MuleSoft Meetup...
 
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
 
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
08448380779 Call Girls In Diplomatic Enclave Women Seeking Men
 
Presentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreterPresentation on how to chat with PDF using ChatGPT code interpreter
Presentation on how to chat with PDF using ChatGPT code interpreter
 
Azure Monitor & Application Insight to monitor Infrastructure & Application
Azure Monitor & Application Insight to monitor Infrastructure & ApplicationAzure Monitor & Application Insight to monitor Infrastructure & Application
Azure Monitor & Application Insight to monitor Infrastructure & Application
 
AI as an Interface for Commercial Buildings
AI as an Interface for Commercial BuildingsAI as an Interface for Commercial Buildings
AI as an Interface for Commercial Buildings
 
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
 

Implementing error-proof, business-critical Machine Learning, presentation by Deevid De Meyer, co-founder of Brainjar, presented at the Trustworthy and Ethical AI Conference on Feb 13th in Brussels

  • 1. Error-proof Machine Learning with explainability and human-centric design
  • 2. WE ARE AN END-TO-END ARTIFICIAL INTELLIGENCE SOFTWARE FACTORY Human-in-the-loop AI systems inevitably lose accuracy over time. Brainjar builds systems that are able to continually improve by working together with their human colleagues. Technology agnostic Because of our broad technology expertise, we always choose the best solution for problem. This can be AI API, pre-trained model or custom model and is always cloud agnostic. End-to-end Brainjar works end-to-end with the customer to build and integrate the AI solution. We start with a data analysis, followed by labeling, model training, front-end building and deployment.
  • 3. WE ARE PART OF THE RACCOONS GROUP Software prototyping AI end-to-end Chatbot framework Blockchain services Full stack Javascript Quantum computing Edge AI Digital production house AI Business Strategy
  • 5. NOT new (but has changed) Early AI (1950’s – 2000’s) • Deterministic, human-made algorithms • Centered around planning or abstract reasoning • Scaled horribly • Required maintenance of structured “knowledge systems” Modern AI (2000’s – Now) • “Learning” algorithms -> Machine Learning • Centered around human senses and prediction • Can scale with data • Can process unstructured data Source: Stanford news
  • 6. Probabilistic (Modern) AI makes mistakes! • ~95-99% accuracy is typical (depending on usecase) • Requires strategy for 5% (human review) • Algorithms can provide “confidence level”, allowing humans to only review edge cases Source: wikipedia
  • 8. NOT Self-Learning! Machine Learning learns from VERIFIED data • Many tasks require human verification for the system to learn & improve • Initial training can be done using historic data • Human feedback still required during production Source: Medium
  • 9. NOT Self-Learning! support question central support domain experts support question ML algorithm domain experts support question ML algorithm domain experts Example: Mail Forwarding (1) (2) (3)
  • 10. LESSON 2 Learn from your users
  • 11.
  • 12. Human-Centered Machine Learning Cooperation benefits all parties • AI eliminates tedium • Humans focus on edge cases • Two key principle builds trust with management • High quality data is produced • AI improves/adapts continuously Source: wikipedia
  • 13.
  • 16. CASE 1 ORDER ADMINSTRATION AT TVH TVH Group is a Belgian company active in the market of forklift trucks, aerial work platforms, agricultural tractors and industrial vehicles. The company has an order department that handles orders from customers via email. TVH has a large order administration department. O R D E R A D M I N I S T R AT I O N Customers can place orders by sending an email with their request to TVH. E M A I L O R D E R S The rental administrators handle these requests by reading the mails and registering the order in the ERP system. R E P E T I T I V E W O R K
  • 17. A R T I F I C I A L I N T E L L I G E N C E Determine if the email is a rental order or not & extracts all entities in the email Present an intuitive verification step for users with automatic retraining Structured order is placed in the ERP system C O N N E C T T O M A I L B O X SOLUTION Connects with existing mailbox & gets triggered when there is an incoming mail V E R I F I C A T I O N P L A C E O R D E R
  • 18. Customers of TVH send orders via email to the order administration. A typical example is shown below. CONNECT TO MAILBOX
  • 19. V E R I F I C A T I O N P L A C E O R D E RC O N N E C T T O M A I L B O X A R T I F I C I A L I N T E L L I G E N C E SOLUTION Determine if the email is a rental order or not & extracts all entities in the email Present an intuitive verification step for users with automatic retraining Structured order is placed in the ERP system Connects with existing mailbox & gets triggered when there is an incoming mail
  • 20. Automatically determines if the email is a rental order & extracts all entities in the email. ARTIFICIAL INTELLIGENCE
  • 21. P L A C E O R D E R V E R I F I C AT I O N SOLUTION Determine if the email is a rental order or not & extracts all entities in the email Present an intuitive verification step for users with automatic retraining Structured order is placed in the ERP system Connects with existing mailbox & gets triggered when there is an incoming mail C O N N E C T T O M A I L B O X A R T I F I C I A L I N T E L L I G E N C E
  • 22. P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g . VERIFICATION
  • 23. P L A C E O R D E R S t r u c t u r e d o r d e r i s p l a c e d i n t h e E R P s y s t e m P L A C E O R D E R SOLUTION Determine if the email is a rental order or not & extracts all entities in the email Present an intuitive verification step for users with automatic retraining Structured order is placed in the ERP system Connects with existing mailbox & gets triggered when there is an incoming mail C O N N E C T T O M A I L B O X A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N
  • 24. BUSINESS CASE The current order administration in each branch of TVH has 7 full time equivalents handling over 400 incoming orders via email per day. 5 MIN T I M E P E R E M A I L 1 MIN BEFORE: AFTER: 80% T I M E R E D U C T I O N
  • 25. THE KLASSIF.AI SOLUTION K l a s s i f . a i h a s i d e n t i f i e d f o u r k e y c o m p o n e n t s t h a t a l l o w o r g a n i s a t i o n s t o l e v e r a g e m a c h i n e l e a r n i n g t o i m p r o v e t h e i r d o c u m e n t p r o c e s s i n g w o r k f l o w s . C o m b i n i n g t h e s e c o m p o n e n t s a l l o w s K l a s s i f . a i t o p r o c e s s d o c u m e n t s a t c o m p u t e r s p e e d s w i t h h u m a n r e l i a b i l i t y . For your experts, it’s business as usual. Meanwhile, Klassif.ai is gathering training data to learn the ins and outs of your documents. The result is a training set that represents the real world. L A B E L I N G Just like any mature software project needs versioning and logging, so does a machine learning model! Klassif.ai provides a model history, along with statistics that show you the model’s performance on real, unseen data. G O V E R N A N C E Klassif.ai integrates seamlessly into your current document processing workflow, enriching your unstructured documents and connecting them to your structured systems. I N T E G R AT I O N Every document type requires its own approach. That’s why Klassif.ai classifies each document before extracting the required information. After all, an invoice needs to be processed differently to an employee contract. C L A S S I F I C AT I O N & E X T R A C T I O N
  • 26. KLASSIF.AI Platform features OCR USER MANAGEMENT AUTOMATIC RETRAINING SECURITY NOTIFICATIONS ADMIN PANEL MULTILINGUAL SUPPORT SCALABILITY MODULAR AI MODELS ON-PREMISE OR IN THE CLOUD
  • 27. PROJECT APPROACH K L A S S I F . A I We map and visualise your business processes 1 . I N TA K E Your domain experts generate data while doing their work as usual 4 . L A B E L We customise our platform based on your way of working 3 . C U S T O M I S E We integrate our services with your systems 2 . I N T E G R AT E We train a machine learning algorithm that meets your needs 5 . T R A I N M O D E L We ready the platform for your experts 6 . D E P L O Y P O C L E A D T I M E : 1 - 3 M O N T H S P R O D U C T I O N R E A D Y: 2 - 5 M O N T H S
  • 28. CASE 2 HEALTHCARE DATA EXTRACTION AT RIZIV RIZIV is the Belgian federal institute for health and disability insurance. The institute is, among others, responsible for handling files on medical accidents. Hospitals scan all case documents and email them to RIZIV. An administrator at RIZIV makes sure the case is complete. C A S E D O C U M E N T S A doctor from RIZIV goes over all case documents and decides on refund of medical costs. R E V I E W B Y D O C T O R An intern prints all case documents and sorts these on date before scanning them again. M A N U A L W O R K
  • 29. O C R A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N B Y D O C T O R Convert scanned document to computer readable text Extract all entities such as, doctor name, procedure date and procedure type from document Present an intuitive verification step for doctors with automatic retraining U P L O A D D O C U M E N T S SOLUTION Doctors can upload all case documents into application
  • 30. U P L O A D D O C U M E N T S O C R SOLUTION Convert scanned document to computer readable text Extract all entities such as, doctor name, procedure date and procedure type from document Present an intuitive verification step for doctors with automatic retraining Doctors can upload all case documents into application A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N B Y D O C T O R
  • 31. A R T I F I C I A L I N T E L L I G E N C EU P L O A D D O C U M E N T S O C R SOLUTION Convert scanned document to computer readable text Extract all entities such as, doctor name, procedure date and procedure type from document Present an intuitive verification step for doctors with automatic retraining Doctors can upload all case documents into application V E R I F I C A T I O N B Y D O C T O R
  • 32. P L A C E O R D E R S t r u c t u r e d o r d e r i s p l a c e d i n t h e E R P s y s t e m V E R I F I C AT I O N B Y D O C T O RU P L O A D D O C U M E N T S O C R SOLUTION Convert scanned document to computer readable text Extract all entities such as, doctor name, procedure date and procedure type from document Present an intuitive verification step for doctors with automatic retraining Doctors can upload all case documents into application A R T I F I C I A L I N T E L L I G E N C E
  • 33. Present an intuitive verification step for doctors with automatic retraining VERIFICATION BY DOCTOR
  • 34. CASE 3 DIGITALISING UNEMPLOYMENT BENEFITS AT HVW HVW, the Belgian relief fund for unemployment benefits, is a public institution for social security which pays unemployed citizens. Each month HVW pays more than 120.000 benefit recipients their replacement income. HVW has a large administration department who processes all unemployment benefits B E N E F I T A D M I N I S T R AT I O N The majority of unemployment forms are still submitted on paper PA P E R F O R M S Due to human mistakes, often error are introduced during the input step requiring extra steps to fix mistakes E R R O R P R O N E
  • 35. H A N D W R I T I N G R E C O G N I T I O N A R T I F I C I A L I N T E L L I G E N C E V E R I F I C A T I O N C o n v e r t h a n d w r i t t e n c o n t a c t i n f o r m a t i o n t o c o m p u t e r r e a d a b l e t e x t E x t r a c t c a l e n d a r i n f o r m a t i o n f r o m p h o t o P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g U P L O A D P H O T O U s e r s c a n u p l o a d a p h o t o o f t h e i r u n e m p l o y m e n t c a r d SOLUTION
  • 36. U s e r s c a n u p l o a d a p h o t o o f t h e i r u n e m p l o y m e n t c a r d UPLOAD PHOTO
  • 37. U P L O A D P H O T O U s e r s c a n u p l o a d a p h o t o o f t h e i r u n e m p l o y m e n t c a r d H A N D W R I T I N G R E C O G N I T I O N C o n v e r t h a n d w r i t t e n c o n t a c t i n f o r m a t i o n t o c o m p u t e r r e a d a b l e t e x t A R T I F I C I A L I N T E L L I G E N C E E x t r a c t c a l e n d a r i n f o r m a t i o n f r o m p h o t o V E R I F I C A T I O N P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g SOLUTION
  • 38. A R T I F I C I A L I N T E L L I G E N C E E x t r a c t c a l e n d a r i n f o r m a t i o n f r o m p h o t o U P L O A D P H O T O U s e r s c a n u p l o a d a p h o t o o f t h e i r u n e m p l o y m e n t c a r d H A N D W R I T I N G R E C O G N I T I O N C o n v e r t h a n d w r i t t e n c o n t a c t i n f o r m a t i o n t o c o m p u t e r r e a d a b l e t e x t V E R I F I C A T I O N P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g SOLUTION
  • 39. P L A C E O R D E R S t r u c t u r e d o r d e r i s p l a c e d i n t h e E R P s y s t e m V E R I F I C AT I O N P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g U P L O A D P H O T O U s e r s c a n u p l o a d a p h o t o o f t h e i r u n e m p l o y m e n t c a r d H A N D W R I T I N G R E C O G N I T I O N C o n v e r t h a n d w r i t t e n c o n t a c t i n f o r m a t i o n t o c o m p u t e r r e a d a b l e t e x t A R T I F I C I A L I N T E L L I G E N C E E x t r a c t c a l e n d a r i n f o r m a t i o n f r o m p h o t o SOLUTION
  • 40. P r e s e n t a n i n t u i t i v e v e r i f i c a t i o n s t e p f o r u s e r s w i t h a u t o m a t i c r e t r a i n i n g VERIFICATION
  • 41. CASE 4 AI PRESENTATION ANALIST In an international organisation a lot of effort is expended in keeping the personnel up-to-date using training videos. T R A I N I N G To help train their trainers, we built an AI system that provides feedback on their presentation style. A U T O M AT E F E E D B A C K Often trainers have extensive knowledge on their subject but are not natural born presenters. T R A I N T H E T R A I N E R
  • 42. A R C H I T E C T U R E A R T I F I C I A L I N T E L L I G E N C E U S E R I N T E R F A C E Our clever architecture only enables the expensive cloud components on demand. Audio and video are separated for further processing. Several AI components are combined to analyse the audio and video of the presentation. Present an intuitive user interface for presenters to view the feedback. SOLUTION U P L O A D V I D E O Users can upload their videos to our application for processing
  • 43. U s e r s c a n v i e w t h e f e e d b a c k a b o u t t h e i r p r e s e n t a t i o n USER INTERFACE
  • 44. U s e r s c a n v i e w t h e f e e d b a c k a b o u t t h e i r p r e s e n t a t i o n USER INTERFACE
  • 45. U s e r s c a n v i e w t h e f e e d b a c k a b o u t t h e i r p r e s e n t a t i o n USER INTERFACE
  • 46. And what about explainability? Explainability is mainly relevant for decision making AI (which we haven’t been asked to build yet) • Still early days • Be careful about feedback to external users! ➢ Micro loan approval experiment Source: wikipedia
  • 47.
  • 48. The choice among algorithm categories can partially be made based on the user persona's ability to intervene at different parts of a machine learning pipeline. If the user is allowed to modify the training data, then pre-processing can be used. If the user is allowed to change the learning algorithm, then in-processing can be used. If the user can only treat the learned model as a black box without any ability to modify the training data or learning algorithm, then only post-processing can be used. AIF360 recommends the earliest mediation category in the pipeline that the user has permission to apply because it gives the most flexibility and opportunity to correct bias as much as possible. If possible, all algorithms from all permissible categories should be tested because the ultimate performance depends on dataset characteristics: there is no one best algorithm independent of dataset
  • 49.
  • 50. NEED FOR A HUMAN IN THE LINK? LET’S GET STARTED! CONTACT US STAY IN TOUCH info@brainjar.ai +32 (0) 473 36 15 55 Gaston Geenslaan 11, B4 3001 Leuven https://brainjar.ai https://twitter.com/brainjarai https://nl.linkedin.com/company/brainjarbe