AI REDGIO 5.0 AI Pipeline Designer - Revision history

Admino739mjm7: /* Features */

2024-09-03T15:18:53Z

Features

← Older revision		Revision as of 15:18, 3 September 2024
Line 30:		Line 30:
	\| Registration of externally trained compatible AI models \|\| Users that have trained ML/DL models outside the AI REDGIO 5.0 Platform yet in compatible libraries, can register them in the platform.		\| Registration of externally trained compatible AI models \|\| Users that have trained ML/DL models outside the AI REDGIO 5.0 Platform yet in compatible libraries, can register them in the platform.
	\|}		\|}
	~~<br>~~

	== Resources ==		== Resources ==

Admino739mjm7: /* Asset Description */

2024-09-03T15:18:45Z

Asset Description

← Older revision		Revision as of 15:18, 3 September 2024
Line 6:		Line 6:
	The AI Pipeline Designer platform allows users design, experiment with, train, execute and deploy AI models at the edge or on cloud resources. The AI Pipeline Designer platform is addressed towards different types of users (i.e. data scientists, technical users, business users) in terms of AI and ML execution, from executing simple data manipulation functions (e.g. filters and aggregations) and applying Artificial Intelligence models for the manufacturing domain analytics, to creating visualisations and reports to highlight insights extracted from datasets and from analytics processes, as well as exporting the outputs of these analyses through interfaces that can be consumed by other systems.		The AI Pipeline Designer platform allows users design, experiment with, train, execute and deploy AI models at the edge or on cloud resources. The AI Pipeline Designer platform is addressed towards different types of users (i.e. data scientists, technical users, business users) in terms of AI and ML execution, from executing simple data manipulation functions (e.g. filters and aggregations) and applying Artificial Intelligence models for the manufacturing domain analytics, to creating visualisations and reports to highlight insights extracted from datasets and from analytics processes, as well as exporting the outputs of these analyses through interfaces that can be consumed by other systems.
	</p>		</p>
	~~<br>~~


	== Features ==		== Features ==

Admino739mjm7 at 15:13, 3 September 2024

2024-09-03T15:13:30Z

← Older revision		Revision as of 15:13, 3 September 2024
Line 1:		Line 1:

	<strong>Facilitating the design, execution and deployment of AI pipelines for AI and ML applications</strong> ~~[[File:Ai pipeline designer 2.png\|thumb\|right\|Image 1: The AI REDGIO 5.0 AI Pipeline Designer - List of AI Pipelines & Execution Details]]~~		<strong>Facilitating the design, execution and deployment of AI pipelines for AI and ML applications</strong>

	== Asset Description ==		== Asset Description ==
Line 32:		Line 32:
	\| Registration of externally trained compatible AI models \|\| Users that have trained ML/DL models outside the AI REDGIO 5.0 Platform yet in compatible libraries, can register them in the platform.		\| Registration of externally trained compatible AI models \|\| Users that have trained ML/DL models outside the AI REDGIO 5.0 Platform yet in compatible libraries, can register them in the platform.
	\|}		\|}


	~~[[File:Ai pipeline designer 1.png\|center\|x300px\|Image Caption]]~~
	~~<div align="center" style="font-size:88%;line-height: 2em">''Image 2: AI Pipeline Designer - AI Pipeline Design Workbench''</div>~~
	<br>		<br>

Admino739mjm7: /* Relevant Categories */

2024-09-03T14:57:33Z

Relevant Categories

← Older revision		Revision as of 14:57, 3 September 2024
Line 45:		Line 45:

	== Relevant Categories==		== Relevant Categories==
	[[Category:~~Collaborative Intelligence / Human~~-in-~~the~~-~~Loop~~]][[Category:~~Software~~-~~as-a-Service~~]]		[[Category:Software-as-a-Service]][[Category:Generic Purpose]][[Category:Edge]][[Category:Cloud-based]]

Admino739mjm7 at 14:53, 3 September 2024

2024-09-03T14:53:05Z

← Older revision		Revision as of 14:53, 3 September 2024
Line 1:		Line 1:

	<strong>Facilitating the design, execution and deployment of AI pipelines for AI and ML applications</strong> [[File:Ai pipeline designer 2.png\|thumb\|right\|Image 1: The AI REDGIO 5.0 ~~Ai pipeline~~ Designer]]		<strong>Facilitating the design, execution and deployment of AI pipelines for AI and ML applications</strong> [[File:Ai pipeline designer 2.png\|thumb\|right\|Image 1: The AI REDGIO 5.0 AI Pipeline Designer - List of AI Pipelines & Execution Details]]

	== Asset Description ==		== Asset Description ==

Admino739mjm7: /* Features */

2024-09-03T14:52:44Z

Features

← Older revision		Revision as of 14:52, 3 September 2024
Line 35:		Line 35:

	[[File:Ai pipeline designer 1.png\|center\|x300px\|Image Caption]]		[[File:Ai pipeline designer 1.png\|center\|x300px\|Image Caption]]
	<div align="center" style="font-size:88%;line-height: 2em">''Image 2: AI Pipeline Designer - ~~List of~~ AI ~~Pipelines & Execution Details~~''</div>		<div align="center" style="font-size:88%;line-height: 2em">''Image 2: AI Pipeline Designer - AI Pipeline Design Workbench''</div>
	<br>		<br>

Admino739mjm7 at 14:52, 3 September 2024

2024-09-03T14:52:12Z

Show changes

Admino739mjm7: Created page with " Facilitating the design, execution and deployment of AI pipelines for AI and ML applications Image 1: The AI REDGIO 5.0 Ai pipeline Designer == Asset Description ==
The AI Pipeline Designer platform allows users design, experiment with, train, execute and deploy AI models at the edge or on cloud resources. The AI Pipeline Designer platform is addressed towards different typ..."

2024-09-03T13:17:37Z

Created page with " <strong>Facilitating the design, execution and deployment of AI pipelines for AI and ML applications</strong> thumb|right|Image 1: The AI REDGIO 5.0 Ai pipeline Designer == Asset Description == <p style="line-height: 1.5em"> The AI Pipeline Designer platform allows users design, experiment with, train, execute and deploy AI models at the edge or on cloud resources. The AI Pipeline Designer platform is addressed towards different typ..."

New page

<strong>Facilitating the design, execution and deployment of AI pipelines for AI and ML applications</strong> [[File:Ai pipeline designer 2.png|thumb|right|Image 1: The AI REDGIO 5.0 Ai pipeline Designer]]

== Asset Description ==
<p style="line-height: 1.5em">
The AI Pipeline Designer platform allows users design, experiment with, train, execute and deploy AI models at the edge or on cloud resources. The AI Pipeline Designer platform is addressed towards different types of users (i.e. data scientists, technical users, business users) in terms of AI and ML execution, from executing simple data manipulation functions (e.g. filters and aggregations) and applying Artificial Intelligence models for the manufacturing domain analytics, to creating visualisations and reports to highlight insights extracted from datasets and from analytics processes, as well as exporting the outputs of these analyses through interfaces that can be consumed by other systems.
</p>
<br>

== Features ==
{| class="wikitable" style="margin:auto; width:100%; line-height: 1.5em"
|-
! Feature !! Description
|-
| Integration of Advanced Technologies|| Integration of cutting-edge technologies, including AI-driven analytics, and Internet of Things (IoT)-enabled devices to create a synergistic ecosystem
|-
| Human-Machine Interaction|| Facilitating interaction between human operators and machines
|-
| Support of Real-time Operations|| The platform works in tandem with real-time data analytics on IoT data
|-
| Continuous Learning and Knowledge Management|| Storage in a knowledge-sharing and learning management repository
|-
| Collaborative Innovation || Fostering a culture of collaborative innovation, where human operators can interact with AI for improved processes and final product
|}

== User Journey ==
<p style="line-height: 1.5em">
Our platform adopts a practical problem-solving approach, data analysis, and process optimisation. This subsection offers a comprehensive usage walkthrough, explaining the interaction of the user with the Collaborative Intelligence Platform as a whole, and as a part of the user's MLOps operations and tools they already use.
</p>
[[File:Collaborative Intelligence Platfrom for Industry 5.0 User Journey.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 2: Collaborative Intelligence Platform for Industry 5.0 - User Journey''</div>
<br>

<p style="line-height: 1.5em">
# ''' Onboarding and User Authentication''': Ensure secure access to the system, safeguarding sensitive data and mitigating potential security breaches
# '''Integrating IoT and Smart Devices''': The integration of IoT and smart devices using a methodology for the interconnection of heterogeneous devices
# '''AI-driven Data Analytics and Insights''': The system can start working to detect patterns, anomalies, and trends within the data through ML algorithms
# '''Human-Machine Interaction and Augmentation''': Coordination of actions between human operators and machines, emphasising augmentation rather than substitution
# '''Real-Time Process Optimisation''': Automatically adjusting operational parameters in response to changing conditions and input provided by the user through the [[#Collaborative Intelligence Component]]
# '''Knowledge Base and Learning Management''': Exploring the knowledge repository containing organisational know-how, accumulated insights, and best practices
# '''Collaborative Innovation and Continuous Improvement''': The ultimate goal of Collaborative Intelligence: facilitating cross-functional collaboration and iterative enhancement through data analytics, human-machine collaboration, and organisational knowledge
</p>

<br>

== Collaborative Intelligence Platform Components==
<p style="line-height: 1.5em">
The AI REDGIO 5.0 Collaborative Intelligence Platform is comprised of 3 underlying components.
</p>
{| class="wikitable" style="margin:auto; width:100%; line-height: 1.5em"
|-
! Component!! Description!! Type!! Status
|-
| Collaborative Intelligence Component ||Allows the human operator to accept or reject a manufactured product. This component has an interface that can be configured around various metrics including, but not limited to, accuracy, interpretability, speed, etc ||User-facing ||Working Prototype
|-
| Pipeline Creation Component ||Easily create, configure, and deploy workflows in various work environments. This component acts as the backbone of a robust and efficient workflow management system, enabling organisations to optimise their processes and achieve higher productivity ||User-facing ||Work in Progress
|-
| Interfacing Component ||Connects the operator interface with hardware and housing pre-trained models ready for deployment in manufacturing environments. This component acts as the bridge between the digital and physical realms of experiments ||Backend ||Working Prototype
|}

=== Usage Walkthroughs===
How to use the AI REDGIO 5.0 Collaborative Intelligence Platform
==== Collaborative Intelligence Component====
<p style="line-height: 1.5em">
The human operator can assess the results of the AI/ML models and provide feedback with regards to 3 aspects (accuracy, energy-efficiency, latency) through the Collaborative Intelligence (C.I) component. The human operator's feedback is propagated back to the AI/ML analytics tool of the user to perform the relevant adjustments to the model configuration for results optimisation.
<br>
</p>

<strong>1. Entering the C.I Platform</strong>
* Navigate to the url https://github.com/AI-REDGIO-5-0/ci-componentt
* Enter your credentials ''(Note: User registration and account creation is managed by [https://www.scch.at/ Software Competence Centre Hagenberg - SCCH])

<strong>2. Data and ML/AI results inspection </strong>

After successfully entering the platform, you are led to the C.I Dashboard, where you can review your data and provide feedback.

* '''Upload Data''': First you need to provide the data that you want to inspect. ''(Note: Data are manually onboarded. In upcoming releases, the onboarding will happen through integration with the user's AI/ML tools and system.)''
<br>
[[File:00 - Upload Data.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 3: Upload data''</div>
<br>
* '''Select ML/AI evaluation axis''': ML/AI models can be evaluated based on various parameters. Click one of the available radio buttons (accuracy, energy-efficiency, latency) to instantiate the Table and the charts with the corresponding data and provide your feedback
* '''View the data''': The Table is loaded with the data from the integrated dataset (containing the ML/AI results). You can select in a dropdown the ML/AI output feature that you would like to inspect.
<br>
[[File:01 - Select feature.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 4: Selection of feature to be inspected''</div>
<br>

<strong>3. Human provides feedback to AI</strong>
* '''Provide feedback''': Check the value of the output feature per row and provide feedback on the ML/AI performance with regards to the evaluation axis you have selected. You provide feedback in the form of Yes (i.e., the ML/AI model for the specific output feature value was accurate/energy-efficient/fast) or No (i.e. the ML/AI model for the specific output feature value was not accurate/not energy-efficient/slow)
<br>
[[File:02 - Provision of feedback.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 5: Provision of feedback on the ML/AI model from the aspect of correctness''</div>
<br>
* '''Check the evaluation charts''': The bar charts that appear on the top of the C.I Dashboard provide an overview of the ML/AI evaluation by the human operator so far
* '''Export findings''': You can click on the option 'Print Non-OK Rows to a File' to export the rows marked as non-ok and further use them afterwards
<br>
[[File:03 - export non ok rows.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 6: Export non-ok rows''</div>
<br>

</p>

<strong>4. AI is adjusted according to Human Feedback</strong>
<div style="font-size:75%">''Note: Will be part of upcoming releases''</div>

=== Interplay with Input Analysis Tool===
<p style="line-height: 1.5em">
To further assist the data scientists and other users in making sense of their data, thus enhancing human-AI collaboration, the AI REDGIO 5.0 Collaborative Intelligence Platform works in tandem with the Input Analysis tool.
</p>
==== How it works====
<p style="line-height: 1.5em">
The human operator can explore through the [https://wiki.ai-redgio50.s5labs.eu/index.php?title=Data_Analysis_Dashboard Data Analysis Dashboard] their data and perform input analysis based on user-selected aspects and thresholds to find correlations between features.
<br>
</p>

<strong>1. Upload input data to C.I Dashboard and navigate to the Data Analysis Dashboard </strong>
* Go to the main Dashboard page of the C.I Dashboard
* Click 'Choose file' and select the input data file from the directory
* Click 'Upload csv' to onboard your input data
* Click the 'Input Analysis' hyperlink to be led to the relevant tool to inspect your data
<br>
[[File:Step 1.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 7: Link to Data Analysis Dashboard from C.I Dashboard''</div>
<br>

<strong>2. Perform input analysis</strong>

After clicking on the hyperlink, you are led to the Data Analysis Dashboard. Through the integration with the C.I Platform, the onboarded input data are available there in order to perform your exploration.

* Select Data Analysis aspects: Through the Data Analysis Dashboard you can select the analysis aspects, including the pairs of columns to compare, the missing value imputation method, the numeric conversion method. Additionally, you can set the thresholds for the analysis (Pearson correlation, Spearman rank correlation, Euclidean similarity)
<br>
[[File:Input analysis 001.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 8: Selection of data analysis aspects''</div>
<br>
* Review the analysis results: In the provided graphs you can see the visualised results of the analysis.
<br>
[[File:Input analysis 002.png|center|x300px|Image Caption]]
<div align="center" style="font-size:88%;line-height: 2em">''Image 9: Inspect analysis visualisations''</div>
<br>

== Resources ==
<ul>
<li>Code available in [https://github.com/AI-REDGIO-5-0/ci-component Github]</li>
<li>Created by [https://www.scch.at/ Software Competence Center Hagenberg - SCCH]</li>
<li>Contact ''jorge.martinez-gil@scch.at''</li>
</ul>

== Relevant Categories==
[[Category:Collaborative Intelligence / Human-in-the-Loop]][[Category:Software-as-a-Service]]