Notice: This Wiki is now read only and edits are no longer possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.
Difference between revisions of "AICloudEdgeOpenLabRequirements"
Line 99: | Line 99: | ||
*Capacity | *Capacity | ||
− | Castalia; testing/reproducibility area of contribution | + | **Castalia; testing/reproducibility area of contribution |
− | Huawei; | + | **Huawei; Infrastucture, servers, open source cloud software, open source software in AI/ML, edge |
*Data | *Data |
Revision as of 12:48, 20 January 2021
Contents
AI, Cloud & Edge OpenLab Meeting Minutes 15-01-2021
List of participants
[Christoph Peylo - Global Head - Bosch Centre for AI] [Bryan Che - Chief Strategy Officer Huawei] [Loretta Tioiela - Strategy Advisor Huawei, Member EC European AI Alliance] [Stefan Voget - Head of AI Infrastructure Laboratory - Continental] [Florian Kisly - Technology Partner Management - Continental] [Nathalie Hauk - Scientific expert - Fraunhofer-Institut FOKUS] [Tom Ritter - Deputy Director - Fraunhofer-Institut FOKUS] [Adam Gibson - CEO - Konduit] [Filippo Formica - Direzione Tecnica, Innovazione e Ricerca - Engineering] [Marco Alessi - Engineering R&D organizational unit - Engineering] [Michele Gabusi - Data Science and Analytics Organizational Unit - Engineering] [Carlos Cañado Moya - Head of Innovation & Software Development - Kairós Digital Solutions] [Simon Butler - Associate Senior Lecturer in Informatics - University of Skövde] [Alexander Karlsson - Senior Lecturer Computer Science - University of Skövde] [Sebastian Scholze - Engineering manager - Institut für angewandte Systemtechnik Bremen GmbH] [Andreas Riexinger - Automotive Expert - OpenADx] [Njec Bat - Project Manager - XLab] [Stratos Arampatzis - Director - Ortelio] [Boris Baldassari - Director/founder - Castalia Solutions] [Michael Plagge - Ecosystem Development Manager Eclipse Foundation - Eclipse Foundation] [Gael Blondelle - VP Ecosystem Development Eclipse Foundation, MD EF Europe - Eclipse Foundation] [Marc Vloemans - Head of AI Ecosystem Development - Eclipse Foundation]
General goals
- Intention to collaborate in an ecosystem. *The community effort will help towards coordination on topics eg trust, safety etc
- A central place to test, how to test, how to verify/qualify
- For now it's more informal initiative, but goal is to come up with production framework
- The OpenLab as a trust platform is an overall topic for everyone
- Work on horizontal applications for many sectors/verticals
- Share contributions
Potential Program themes
- Collaboration
- Common ecosystem / MLOPS, sharing technology
- Bring different initiative together, Federated Learning
- IoT Ecosystems
- Want to test with Hybrid cloud / Federated Services
- Create distributions / synchronous releases
- Streamlining processes for testing
- Trust
- Trust in AI / IOT, in AI decisions
- Need to see if we can have a collaboration or even better, implement the Digital Trust Forum, implement Digital Trust Forum in AI & IoT/in the OpenLab -> more trust in IoT devices, Trustworthy AI -> Trustworthy IoT
- Work with partners to make the ecosystem more trustfully
- Joint platform could be ideal for cooperation
- Addition to trust discussion, implementation to "explain" AI, interest to share this in the initiative > share contributions, share test methods
- Ethics
- Ethics of AI re hardware data collection
- Security & Safety & privacy
- IT security and safety is still an issue
- In automotive industry, definition of new safety strategy ongoing
- Automotive Safety strategy and need to increase robustness
- Privacy
- Integration
- Continuous integration/DevOps
- Good entry point for a couple of things, How to put together "everything"
- Projects
- Brainstorming:
- Infrastructure, sensor and wearables, health, Intelligence @ the Edge
- Deploy & Embed AI in Enterprise architectures
- Support for many languages
- Project to manage AI APIs (competing with Acumos)
- Integrate AI with workflows : Combining rules engine with machine learning. Falls back on human interaction when AI fails
- Cloud for I4.0
- Hybrid datastores
- Project: IOT Testware (Eclipse project)
- Application of research results
- Develop test suits/test cases
- Use Case : Deep Encode -> requires lots of compute power to adapt encoding on the fly thanks to AI
- Toolchain for autonomous driving to test first solutions
- Brainstorming:
- Technologies to be developed
- Re limitation of cloud platform: can't use custom libraries
- Requirements for applications, language support, support for continuous deployment
- Embedded hardware requirements, DevOps, test methods
- Rules engines with ML
- Data
- Test data collection
- Calibrated data sets needed/data reliability needed
- Other
- Community efforts -> Frameworks that can be used for business
- Verticals’ uses cases (Health/Hospital data-AI apps, project re Wearables/ with CEA)
Potential offers to bring to the OpenLab
- Capacity
- Castalia; testing/reproducibility area of contribution
- Huawei; Infrastucture, servers, open source cloud software, open source software in AI/ML, edge
- Data
- Non mentioned
- Other resources (eg use cases)
- Various attendees can bring use cases
- Miscellaneous items discussed
- Interest in MLOPS platform
- Custom libraries / OSS libraries -> Need to have continuous test / deployment
- https://smartclide.eu/ (Note, EF; OpenLab LoS signed with them)
- https://www.typhon-project.org/
- Main constraints: missing of data availability
- Develop cloud AI service for robots
- Model uncertainty with "black box" methods (AI) and probabilistic methods
- Combine Neural Network with statistical approaches
Next Steps/Action points
- Distribute meeting minutes via email w/ link to wiki
- Setup follow-up meeting to discuss the next steps
- Identify additional stakeholders for more detailed talks