SlideShare a Scribd company logo

DU Series - Day 4.pptx

Document Understanding Series

1 of 28
Download to read offline
DU Series – Day 4:
Model Training in Document
Understanding
Lahiru Fernando
https://community.uipath.com/delhi-ncr/
Document Understanding Series
https://community.uipath.com/events/details/uipath-delhi-ncr-presents-document-understanding-series-2023/
UiPath Community MVP
The Most Valuable Professional (MVP) Award is the highest recognition
that we offer to our community members for their outstanding
contribution, innovation, and evangelism shown in
the larger automation community.
• Stand out as a leading contributor in the AI-Powered world!
• Envision the automation platform together!
• Become the next UiPath Community MVP: Accelerate Your Automation Impact
• Get recognized among the top contributors in the AI-Powered community!
APPLY
NOW!!!
UiPath Document
Understanding
Automation Technical Lead,
Accelirate
Process documents intelligently
UiPath Document
Understanding
Process documents intelligently
Day 4 DU Series
LAHIRU FERNANDO
UiPath AI ambassador APAC
4 X UiPath MVP
Director (Sri Lanka)/ RPA Lead (APAC)/
Practice Lead for Document Understanding
and AI
Boundaryless Group

Recommended

Introducing Compreno - Natural Language Processing Technology
Introducing Compreno - Natural Language Processing TechnologyIntroducing Compreno - Natural Language Processing Technology
Introducing Compreno - Natural Language Processing TechnologyABBYY
 
UiPath Document Understanding_Day 3.pptx
UiPath Document Understanding_Day 3.pptxUiPath Document Understanding_Day 3.pptx
UiPath Document Understanding_Day 3.pptxUiPathCommunity
 
How to classify documents automatically using NLP
How to classify documents automatically using NLPHow to classify documents automatically using NLP
How to classify documents automatically using NLPSkyl.ai
 
Dev Dives: Mastering AI-powered Document Understanding
Dev Dives: Mastering AI-powered Document UnderstandingDev Dives: Mastering AI-powered Document Understanding
Dev Dives: Mastering AI-powered Document UnderstandingUiPathCommunity
 
Intro of Key Features of SoftCAAT BI SQL Software
Intro of Key Features of SoftCAAT BI SQL SoftwareIntro of Key Features of SoftCAAT BI SQL Software
Intro of Key Features of SoftCAAT BI SQL Softwarerafeq
 
Intro of Key Features of SoftCAAT Pro software
Intro of Key Features of SoftCAAT Pro softwareIntro of Key Features of SoftCAAT Pro software
Intro of Key Features of SoftCAAT Pro softwarerafeq
 
Introduction to Microsoft Syntex
Introduction to Microsoft SyntexIntroduction to Microsoft Syntex
Introduction to Microsoft SyntexDrew Madelung
 

More Related Content

Similar to DU Series - Day 4.pptx

ChatGPT and not only: how can you use the power of Generative AI at scale
ChatGPT and not only: how can you use the power of Generative AI at scaleChatGPT and not only: how can you use the power of Generative AI at scale
ChatGPT and not only: how can you use the power of Generative AI at scaleMaxim Salnikov
 
Intelligent Automation in Accounting and Finance with IMA Queens College Stud...
Intelligent Automation in Accounting and Finance with IMA Queens College Stud...Intelligent Automation in Accounting and Finance with IMA Queens College Stud...
Intelligent Automation in Accounting and Finance with IMA Queens College Stud...Diana Gray, MBA
 
AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...
AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...
AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...DianaGray10
 
Getting started with with SharePoint Syntex
Getting started with with SharePoint SyntexGetting started with with SharePoint Syntex
Getting started with with SharePoint SyntexDrew Madelung
 
Building an effective sharepoint team
Building an effective sharepoint teamBuilding an effective sharepoint team
Building an effective sharepoint teamBaris Bruce Tuncertan
 
UiPath Document Understanding_Day 2.pptx
UiPath Document Understanding_Day 2.pptxUiPath Document Understanding_Day 2.pptx
UiPath Document Understanding_Day 2.pptxRohitRadhakrishnan8
 
Getting Knowledge Transfer Right Enterprise Wide Webinar
Getting Knowledge Transfer Right Enterprise Wide WebinarGetting Knowledge Transfer Right Enterprise Wide Webinar
Getting Knowledge Transfer Right Enterprise Wide WebinarConcept Searching, Inc
 
Azure Machine Learning
Azure Machine LearningAzure Machine Learning
Azure Machine LearningMostafa
 
Evolution of Online Delivery | Scott Youngblom
Evolution of Online Delivery | Scott YoungblomEvolution of Online Delivery | Scott Youngblom
Evolution of Online Delivery | Scott YoungblomLavaConConference
 
Climbing the Slippery Slope of SharePoint Migrations Webinar
Climbing the Slippery Slope of SharePoint Migrations WebinarClimbing the Slippery Slope of SharePoint Migrations Webinar
Climbing the Slippery Slope of SharePoint Migrations WebinarConcept Searching, Inc
 
WebXpress Business Intelligence Capability
WebXpress Business Intelligence CapabilityWebXpress Business Intelligence Capability
WebXpress Business Intelligence CapabilityWebXpress.IN
 
Using the power of OpenAI with your own data: what's possible and how to start?
Using the power of OpenAI with your own data: what's possible and how to start?Using the power of OpenAI with your own data: what's possible and how to start?
Using the power of OpenAI with your own data: what's possible and how to start?Maxim Salnikov
 
Success Factors for DITA Adoption with XMetaL: Best Practices and Fundamentals
Success Factors for DITA Adoption with XMetaL: Best Practices and FundamentalsSuccess Factors for DITA Adoption with XMetaL: Best Practices and Fundamentals
Success Factors for DITA Adoption with XMetaL: Best Practices and FundamentalsScott Abel
 
AI for Customer Service - How to Improve Contact Center Efficiency with Machi...
AI for Customer Service - How to Improve Contact Center Efficiency with Machi...AI for Customer Service - How to Improve Contact Center Efficiency with Machi...
AI for Customer Service - How to Improve Contact Center Efficiency with Machi...Skyl.ai
 
Intro of Key Features of Auto eCAAT Ent Software
Intro of Key Features of Auto eCAAT Ent SoftwareIntro of Key Features of Auto eCAAT Ent Software
Intro of Key Features of Auto eCAAT Ent Softwarerafeq
 
Intro of Key Features of Soft CAAT Ent Software
Intro of Key Features of Soft CAAT Ent SoftwareIntro of Key Features of Soft CAAT Ent Software
Intro of Key Features of Soft CAAT Ent Softwarerafeq
 

Similar to DU Series - Day 4.pptx (20)

ChatGPT and not only: how can you use the power of Generative AI at scale
ChatGPT and not only: how can you use the power of Generative AI at scaleChatGPT and not only: how can you use the power of Generative AI at scale
ChatGPT and not only: how can you use the power of Generative AI at scale
 
Intelligent Automation in Accounting and Finance with IMA Queens College Stud...
Intelligent Automation in Accounting and Finance with IMA Queens College Stud...Intelligent Automation in Accounting and Finance with IMA Queens College Stud...
Intelligent Automation in Accounting and Finance with IMA Queens College Stud...
 
AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...
AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...
AI and ML Series - Leveraging Generative AI and LLMs Using the UiPath Platfor...
 
Getting started with with SharePoint Syntex
Getting started with with SharePoint SyntexGetting started with with SharePoint Syntex
Getting started with with SharePoint Syntex
 
Jayanth_Resume
Jayanth_ResumeJayanth_Resume
Jayanth_Resume
 
Building an effective sharepoint team
Building an effective sharepoint teamBuilding an effective sharepoint team
Building an effective sharepoint team
 
UiPath Document Understanding_Day 2.pptx
UiPath Document Understanding_Day 2.pptxUiPath Document Understanding_Day 2.pptx
UiPath Document Understanding_Day 2.pptx
 
Getting Knowledge Transfer Right Enterprise Wide Webinar
Getting Knowledge Transfer Right Enterprise Wide WebinarGetting Knowledge Transfer Right Enterprise Wide Webinar
Getting Knowledge Transfer Right Enterprise Wide Webinar
 
Abdul ETL Resume
Abdul ETL ResumeAbdul ETL Resume
Abdul ETL Resume
 
Azure Machine Learning
Azure Machine LearningAzure Machine Learning
Azure Machine Learning
 
Lecture # 08 (developing business it solution)
Lecture # 08 (developing business it solution)Lecture # 08 (developing business it solution)
Lecture # 08 (developing business it solution)
 
Evolution of Online Delivery | Scott Youngblom
Evolution of Online Delivery | Scott YoungblomEvolution of Online Delivery | Scott Youngblom
Evolution of Online Delivery | Scott Youngblom
 
Climbing the Slippery Slope of SharePoint Migrations Webinar
Climbing the Slippery Slope of SharePoint Migrations WebinarClimbing the Slippery Slope of SharePoint Migrations Webinar
Climbing the Slippery Slope of SharePoint Migrations Webinar
 
SharePoint Fest Chicago Presentation
SharePoint Fest Chicago PresentationSharePoint Fest Chicago Presentation
SharePoint Fest Chicago Presentation
 
WebXpress Business Intelligence Capability
WebXpress Business Intelligence CapabilityWebXpress Business Intelligence Capability
WebXpress Business Intelligence Capability
 
Using the power of OpenAI with your own data: what's possible and how to start?
Using the power of OpenAI with your own data: what's possible and how to start?Using the power of OpenAI with your own data: what's possible and how to start?
Using the power of OpenAI with your own data: what's possible and how to start?
 
Success Factors for DITA Adoption with XMetaL: Best Practices and Fundamentals
Success Factors for DITA Adoption with XMetaL: Best Practices and FundamentalsSuccess Factors for DITA Adoption with XMetaL: Best Practices and Fundamentals
Success Factors for DITA Adoption with XMetaL: Best Practices and Fundamentals
 
AI for Customer Service - How to Improve Contact Center Efficiency with Machi...
AI for Customer Service - How to Improve Contact Center Efficiency with Machi...AI for Customer Service - How to Improve Contact Center Efficiency with Machi...
AI for Customer Service - How to Improve Contact Center Efficiency with Machi...
 
Intro of Key Features of Auto eCAAT Ent Software
Intro of Key Features of Auto eCAAT Ent SoftwareIntro of Key Features of Auto eCAAT Ent Software
Intro of Key Features of Auto eCAAT Ent Software
 
Intro of Key Features of Soft CAAT Ent Software
Intro of Key Features of Soft CAAT Ent SoftwareIntro of Key Features of Soft CAAT Ent Software
Intro of Key Features of Soft CAAT Ent Software
 

More from UiPathCommunity

Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...
Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...
Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...UiPathCommunity
 
Unleashing the force of AI-powered intelligent document processing
Unleashing the force of AI-powered intelligent document processingUnleashing the force of AI-powered intelligent document processing
Unleashing the force of AI-powered intelligent document processingUiPathCommunity
 
Communauté UiPath Suisse romande - Séance de janvier 2024
Communauté UiPath Suisse romande - Séance de janvier 2024Communauté UiPath Suisse romande - Séance de janvier 2024
Communauté UiPath Suisse romande - Séance de janvier 2024UiPathCommunity
 
Testautomatisierung: Heatmap für SAP und Community-Austausch
Testautomatisierung: Heatmap für SAP und Community-AustauschTestautomatisierung: Heatmap für SAP und Community-Austausch
Testautomatisierung: Heatmap für SAP und Community-AustauschUiPathCommunity
 
All about Process Management
All about Process ManagementAll about Process Management
All about Process ManagementUiPathCommunity
 
egacy-to-Windows Conversion: Your Migration Jump Start
egacy-to-Windows Conversion: Your Migration Jump Startegacy-to-Windows Conversion: Your Migration Jump Start
egacy-to-Windows Conversion: Your Migration Jump StartUiPathCommunity
 
Let's talk about Coded Automation
Let's talk about Coded AutomationLet's talk about Coded Automation
Let's talk about Coded AutomationUiPathCommunity
 
Masterclass on Building Disruptive Automation Solutions
Masterclass on Building Disruptive Automation Solutions Masterclass on Building Disruptive Automation Solutions
Masterclass on Building Disruptive Automation Solutions UiPathCommunity
 
RPA Business Analysis - Up and Running.pdf
RPA Business Analysis - Up and Running.pdfRPA Business Analysis - Up and Running.pdf
RPA Business Analysis - Up and Running.pdfUiPathCommunity
 
Women in Automation 2023- UiPath Studio Session 2.pdf
Women in Automation 2023- UiPath Studio Session 2.pdfWomen in Automation 2023- UiPath Studio Session 2.pdf
Women in Automation 2023- UiPath Studio Session 2.pdfUiPathCommunity
 

More from UiPathCommunity (11)

Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...
Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...
Dev Dives: Leverage APIs and Gen AI to power automations for RPA and software...
 
Unleashing the force of AI-powered intelligent document processing
Unleashing the force of AI-powered intelligent document processingUnleashing the force of AI-powered intelligent document processing
Unleashing the force of AI-powered intelligent document processing
 
Communauté UiPath Suisse romande - Séance de janvier 2024
Communauté UiPath Suisse romande - Séance de janvier 2024Communauté UiPath Suisse romande - Séance de janvier 2024
Communauté UiPath Suisse romande - Séance de janvier 2024
 
Azure CICD - Day1.pptx
Azure CICD - Day1.pptxAzure CICD - Day1.pptx
Azure CICD - Day1.pptx
 
Testautomatisierung: Heatmap für SAP und Community-Austausch
Testautomatisierung: Heatmap für SAP und Community-AustauschTestautomatisierung: Heatmap für SAP und Community-Austausch
Testautomatisierung: Heatmap für SAP und Community-Austausch
 
All about Process Management
All about Process ManagementAll about Process Management
All about Process Management
 
egacy-to-Windows Conversion: Your Migration Jump Start
egacy-to-Windows Conversion: Your Migration Jump Startegacy-to-Windows Conversion: Your Migration Jump Start
egacy-to-Windows Conversion: Your Migration Jump Start
 
Let's talk about Coded Automation
Let's talk about Coded AutomationLet's talk about Coded Automation
Let's talk about Coded Automation
 
Masterclass on Building Disruptive Automation Solutions
Masterclass on Building Disruptive Automation Solutions Masterclass on Building Disruptive Automation Solutions
Masterclass on Building Disruptive Automation Solutions
 
RPA Business Analysis - Up and Running.pdf
RPA Business Analysis - Up and Running.pdfRPA Business Analysis - Up and Running.pdf
RPA Business Analysis - Up and Running.pdf
 
Women in Automation 2023- UiPath Studio Session 2.pdf
Women in Automation 2023- UiPath Studio Session 2.pdfWomen in Automation 2023- UiPath Studio Session 2.pdf
Women in Automation 2023- UiPath Studio Session 2.pdf
 

Recently uploaded

UGBINTERNETBANKING FACILITY LAUNCHED.pptx
UGBINTERNETBANKING FACILITY LAUNCHED.pptxUGBINTERNETBANKING FACILITY LAUNCHED.pptx
UGBINTERNETBANKING FACILITY LAUNCHED.pptxRiteshsahu101
 
Augmented and Mixed Reality Solutions for Frontline Medical Professionals
Augmented and Mixed Reality Solutions for Frontline Medical ProfessionalsAugmented and Mixed Reality Solutions for Frontline Medical Professionals
Augmented and Mixed Reality Solutions for Frontline Medical Professionalsthirdeyegen65
 
history of tau gamma architect.1968.....
history of tau gamma architect.1968.....history of tau gamma architect.1968.....
history of tau gamma architect.1968.....josephiigo
 
Regulation is Coming - Trusted Media Summit 2023
Regulation is Coming - Trusted Media Summit 2023Regulation is Coming - Trusted Media Summit 2023
Regulation is Coming - Trusted Media Summit 2023Damar Juniarto
 
Obstructive jaundice is a medical condition characterized by the yellowing of...
Obstructive jaundice is a medical condition characterized by the yellowing of...Obstructive jaundice is a medical condition characterized by the yellowing of...
Obstructive jaundice is a medical condition characterized by the yellowing of...ssuser7b7f4e
 
Augmented and Mixed Reality Solutions for Aerospace & Defense
Augmented and Mixed Reality Solutions for Aerospace & DefenseAugmented and Mixed Reality Solutions for Aerospace & Defense
Augmented and Mixed Reality Solutions for Aerospace & Defensethirdeyegen65
 
APAN 57: APNIC Report at APAN 57, Bangkok, Thailand
APAN 57: APNIC Report at APAN 57, Bangkok, ThailandAPAN 57: APNIC Report at APAN 57, Bangkok, Thailand
APAN 57: APNIC Report at APAN 57, Bangkok, ThailandAPNIC
 
Modern Red Teaming - subverting mature defenses on a budget
Modern Red Teaming - subverting mature defenses on a budgetModern Red Teaming - subverting mature defenses on a budget
Modern Red Teaming - subverting mature defenses on a budgetmatt806068
 
AWS Overview of AWS Clarify, Feature Store, Hyper parameter Tuning
AWS Overview of AWS  Clarify, Feature Store, Hyper parameter TuningAWS Overview of AWS  Clarify, Feature Store, Hyper parameter Tuning
AWS Overview of AWS Clarify, Feature Store, Hyper parameter TuningVarun Garg
 
[Hackersuli]Privacy on the blockchain
[Hackersuli]Privacy on the blockchain[Hackersuli]Privacy on the blockchain
[Hackersuli]Privacy on the blockchainhackersuli
 
Red shadows ringing in Japan's Cyberspace
Red shadows ringing in Japan's CyberspaceRed shadows ringing in Japan's Cyberspace
Red shadows ringing in Japan's Cyberspacesttyk
 

Recently uploaded (13)

UGBINTERNETBANKING FACILITY LAUNCHED.pptx
UGBINTERNETBANKING FACILITY LAUNCHED.pptxUGBINTERNETBANKING FACILITY LAUNCHED.pptx
UGBINTERNETBANKING FACILITY LAUNCHED.pptx
 
Riesgos online
Riesgos onlineRiesgos online
Riesgos online
 
Augmented and Mixed Reality Solutions for Frontline Medical Professionals
Augmented and Mixed Reality Solutions for Frontline Medical ProfessionalsAugmented and Mixed Reality Solutions for Frontline Medical Professionals
Augmented and Mixed Reality Solutions for Frontline Medical Professionals
 
history of tau gamma architect.1968.....
history of tau gamma architect.1968.....history of tau gamma architect.1968.....
history of tau gamma architect.1968.....
 
Regulation is Coming - Trusted Media Summit 2023
Regulation is Coming - Trusted Media Summit 2023Regulation is Coming - Trusted Media Summit 2023
Regulation is Coming - Trusted Media Summit 2023
 
Obstructive jaundice is a medical condition characterized by the yellowing of...
Obstructive jaundice is a medical condition characterized by the yellowing of...Obstructive jaundice is a medical condition characterized by the yellowing of...
Obstructive jaundice is a medical condition characterized by the yellowing of...
 
Augmented and Mixed Reality Solutions for Aerospace & Defense
Augmented and Mixed Reality Solutions for Aerospace & DefenseAugmented and Mixed Reality Solutions for Aerospace & Defense
Augmented and Mixed Reality Solutions for Aerospace & Defense
 
APAN 57: APNIC Report at APAN 57, Bangkok, Thailand
APAN 57: APNIC Report at APAN 57, Bangkok, ThailandAPAN 57: APNIC Report at APAN 57, Bangkok, Thailand
APAN 57: APNIC Report at APAN 57, Bangkok, Thailand
 
Modern Red Teaming - subverting mature defenses on a budget
Modern Red Teaming - subverting mature defenses on a budgetModern Red Teaming - subverting mature defenses on a budget
Modern Red Teaming - subverting mature defenses on a budget
 
AWS Overview of AWS Clarify, Feature Store, Hyper parameter Tuning
AWS Overview of AWS  Clarify, Feature Store, Hyper parameter TuningAWS Overview of AWS  Clarify, Feature Store, Hyper parameter Tuning
AWS Overview of AWS Clarify, Feature Store, Hyper parameter Tuning
 
[Hackersuli]Privacy on the blockchain
[Hackersuli]Privacy on the blockchain[Hackersuli]Privacy on the blockchain
[Hackersuli]Privacy on the blockchain
 
Red shadows ringing in Japan's Cyberspace
Red shadows ringing in Japan's CyberspaceRed shadows ringing in Japan's Cyberspace
Red shadows ringing in Japan's Cyberspace
 
B1 Evaluation.docx
B1 Evaluation.docxB1 Evaluation.docx
B1 Evaluation.docx
 

DU Series - Day 4.pptx

  • 1. DU Series – Day 4: Model Training in Document Understanding Lahiru Fernando
  • 5. UiPath Community MVP The Most Valuable Professional (MVP) Award is the highest recognition that we offer to our community members for their outstanding contribution, innovation, and evangelism shown in the larger automation community. • Stand out as a leading contributor in the AI-Powered world! • Envision the automation platform together! • Become the next UiPath Community MVP: Accelerate Your Automation Impact • Get recognized among the top contributors in the AI-Powered community! APPLY NOW!!!
  • 6. UiPath Document Understanding Automation Technical Lead, Accelirate Process documents intelligently UiPath Document Understanding Process documents intelligently Day 4 DU Series LAHIRU FERNANDO UiPath AI ambassador APAC 4 X UiPath MVP Director (Sri Lanka)/ RPA Lead (APAC)/ Practice Lead for Document Understanding and AI Boundaryless Group
  • 7. Agenda • Introduction to AI Powered Automation • UiPath AI products • Introduction to AI Powered Document Understanding • Implementing IDP with RPA • Best Practices and Lessons Learned from Customer Deployments • Latest Product Updates
  • 8. What is AI Powered Automation • AI Powered Automation is about integrating of AI technologies with traditional automation processes, aiming to enhance decision- making, optimize operations, and improve efficiency. • It introduces cognitive capabilities to mimic human-like reasoning and understanding.
  • 9. Why AI in Automation • AI is the perfect complement to RPA, together providing more accurate and efficient automation. • RPA – automating repetitive rule-based activities • AI – Introduces intelligent tools to simplify processes • RPA + AI = Brings Intelligent Automation to automate complex processes
  • 10. Document Understanding and AI Document Understanding Artificial Intelligence (AI) Document Processing Robotic Process Automation (RPA) Document understanding is the ability to extract and interpret information and meaning from a wide range of documents. It emerges at the intersection of document processing, AI, and RPA. Get your documents processed intelligently Teach your robots to understand documents using AI- enhanced skills for data extraction and interpretation. Drag and drop these capabilities directly into your RPA workflows to combine AI and RPA
  • 11. AI in Document Understanding AI Center • 30+ Pre-trained Specialized AI models available out of the box • Models for Classification and Extraction • Bring your own model – custom or 3rd party • Retrain the models Integration Services • Connect with advanced Generative AI models for document processing through Connectors/ API Studio • Specialized Activities to Connect with Specialized AI models for Classification and Extraction • Specialized activities for Generative Classification and Extraction methods • Activities to support model fine-tuning Document Understanding • AI for fixed structure documents • Easy training of Classification models • Easy training of Extraction models
  • 12. Deciding Factors for AI in Document Understanding Document Layouts Document Content Language and Objects Document Classification Requirements Data Extraction Requirements • Structured/ semi-structured/ unstructured layouts • Dynamic structured content • Layout complexity • Content length • Content presentation • Content change frequency • Language requirements • Handwritten data processing • Object detection and extraction • Document type detection methods • Document splitting complexity • Extraction complexity • Extraction and transformation needs • Data representations and business requirements
  • 13. Deciding Factors for AI in Document Understanding – Generative vs Specialized Classification Extraction Specialized Generative Specialized Generative Semi-structured simple to complex layouts for a given document type Specialized models for a given type of document Unstructured data process requirements with complex cleansing and transformations Largely varying nature of the documents Complex extraction and validation rules Keyword groups are not capable to accurately predict the document types Largely varying content of documents making it difficult to predict the type Requirements to understand the context of document before splitting and classifying
  • 14. Implementing AI in Document Understanding Collecting documents for model training Collect all required documents (addressing various layouts and extraction needs) and perform initial model training in Document Manager. Consider using out-of-the-box models to perform initial training on your data where applicable Evaluating the performance of models Perform evaluation runs to test the model performance on your documents. Analyze results and fine-tune to optimize the model accuracy Fine-tune models with feedback loops HITL enables users to submit corrections where the model is not able to predict. Introduce feedback loops to capture user data to optimize the model performance by including them in the training pipelines. Analyze STP rates to optimize process flows Straight Through Processing rates are affected by model performance, OCR, and other reasons. Identify the root causes, categorize, and apply possible solutions to improve the STP rates.
  • 15. Collecting Documents for Model Training Identify the different layouts in documents and understand the complexity Collect documents from each layout for training Separate your dataset using 80 – 20 rule. 80% for training and 20% for evaluation Consider language, tabular data, handwritten data, splitting requirements when preparing the training and evaluation dataset. Prepare an unbiased dataset for optimal accuracy for all variations.
  • 16. Collecting Documents for Model Training Document Manager through AI Center Extractor Training Methods Document Understanding service (app) Classifier Training Methods Document Manager through AI Center Document Understanding service (app) UiPath Studio Activities (Update Classifier JSON)
  • 17. Collecting Documents for Classifier Training Identify what keywords/ keyword groups can be used for classification Understand the document splitting requirements Intelligent Keywords – Use only related documents for IKC training. Split if one file contains multiple documents Prepare an unbiased dataset for optimal accuracy for all variations. Maintain unique keyword groups across different document types for better accuracy Include sufficient keywords/ training documents for better accuracy
  • 18. Evaluating Model Performance Evaluation pipelines help analyze the model performance on predicted fields Add additional training documents to support overcome incorrect predictions for better accuracy
  • 19. Fine-tune Classifiers with Feedback Loops Enable feedback loops to push manually verified data into classifier JSON files Consider only manual verified data for fine- tuning Optimize keywords as you go to improve accuracy through Validation Screens
  • 20. Fine-tune Models with Feedback Loops Enable feedback loops to push manual verified data to AI Center Schedule Document Manager export, training pipeline and auto update skill to complete the loop Consider only manual verified data for fine- tuning Perform sanity checks to ensure correct manually verified data is submitted for training
  • 21. Quick Showcase on Classifier Training & Generative Extraction
  • 22. Analyze STP Rates to Optimize Process Flows Business value calculations & terminology Calculating the business value depends on the goals. Examples: • Increase document processing throughput per FTE • Decrease processing time • Improve data accuracy Solution value is typically measured by: Straight Through Processing (STP) rate & Average FTE handling time
  • 23. Implementing Business Validations Apply business validations to perform automatic data verification. Validations align with business rules and basic data checks. *Extracted from AI Summit 2023
  • 24. HITL Processing HITL – System Specific HITL – External Optical Character Recognition issues Field Detection Wrong or incomplete data External system issues Escalations based on extracted data
  • 25. Model Performance and Improvements Model Monitoring Causes Action Field accuracy checks Accuracy from specific document sources Low confidence score Inconsistent labeling Biased model training Evaluate confidence thresholds Review and correct data labeling Train model on different layouts and samples
  • 26. Latest Product Updates DocumentUnderstanding.Activities New Activity Pack Cross platform development ML Models Classify documents through pre- trained ML New Data Type: DocumentData Generative Classifier Generative Extractor Enhanced Data Labeling through: • Generative labeling • Specialized AI labeling • Generative + Specialized combined models
  • 28. Use Case Discussions and Lessons Learned Child Welfare – Foster Family Applications Child Welfare – Case Management Invoice Processing Legal Document Data Extraction Submitting applications for foster care licensing can be very time consuming because of the overwhelming amount of documentation required. Each application can contain over 80 document types. UiPath Document Understanding solution improved the processing time by 80% (reducing time from 6 hours to 10 mins per document) The AP team of a large organization process invoices from different vendors. They receive invoices from vendors based in 12 countries. Each country has around 150 vendors sending invoices. UiPath Document Understanding invoice processing solution reduced the processing time of all invoices by 80% and improved the accuracy by 90%. When case workers spend time on administrative tasks, they are not spending their time with children and families. We found a solution for one o the most arduous tasks filing case reports. The solution allowed case workers to save case notes locally. Then our virtual assistant takes are of posting them to the appropriate state and local systems. It saves case workers 15 minutes of admin time for every filing (saving over 20,000 hours of admin effort). An organization process legal documents on a regular basis to address various business requirements. These documents are largely unstructured. The users need to extract the content of the document exactly as it is mentioned. The Proof of concept was built using UiPath Document Understanding and Generative AI integration. Generative AI capabilities in UiPath was applied to process unstructured data.