Constructing the future of (digital) minds Home | Contact 
MIND|CONSTRUCT
Roadmap - Development & funding

MIND|CONSTRUCT is a big project with even bigger opportunities. To be able to develop something like we are doing, serious amounts of money are needed. We hope to get this money from investors (private and corporate) and by making use of several available subsidies and technology stimulation funds like the EU Horizon2020 fund.

For interested parties we have below listed an overview of the (current) planning and budgetting for the remainder of the project.

Interested in participating?

For inquiries regarding participation or other forms of funding you can directly contact:

  • Hans Peter Willems (all inquiries): +31(0)615-603-315
  • André de Koning (all inquiries): +31(0)655-111-375

You can also use our contact-form.

Planning, budgeting and realization
Phase 1: Development of concept and technical design

During this phase the underlying concept has been researched and developed. Much time and effort has been put into vindicating the concept against existing research. We also researched extensively any possibility or proof that our technology would NOT be possible, however we have not been able to find such information. On the contrary we have found large amounts of existing research that gives a solid foundation for our concept.

Based on the conceptual model we developed the technical design, that is the foundation for our first prototype. We have finalized the details of the technical design, in part based on the selection and availability of development tooling.

Meta
Planning/period2009/Q3 - 2014/Q4
Budget/Funding€ 300K
Actual status milestone_status_select_4
DeliverablesConceptual & technical design
ApplicationAGI/Strong-AI R&D-projects
Phase 2: Realization R&D-platform

To be able to implement, test and validate/benchmark the ASTRID Cognitive Architectiure, we need a 'workbench' that can support those development steps. We have build this platform with the use of the Model Driven Architecture (MDA) technology from OpenAdvantage (see our partner page for info). This platform now supports all our individual technologies that assemble into our final ASTRID-system.

During the remainder of the project, our platform will be expanded and tuned when and where needed. For this we have continued access to the MDA-technology for rapid prototyping and implementation.

Meta
Planning/period2015/Q1 - 2015/Q4
Budget/Funding€ 500K
Actual status milestone_status_select_4
DeliverablesFunctional R&D platform
ApplicationR&D support for the ASTRID system
Phase 3: Developing a Semantic Part-of-Speech tagger

To be able to fast-track the (future) training of the ASTRID-system, we need an efficient way for explaining the semantic properties of the language we use to describe the world. For this goal we developed a Part-ofSpeech (POS) tagger that can tag incomming sentences with semantic values unsupervised. Our tagger is based on 'Hidden Markov Models' and is trained with a 'sparse dataset' (as opposed to Big Data).

The results of our POS-tagger technology rival the best taggers currently in existance. Our tagger is benchmarked against the Xerox reference tagger.

Related news:

Meta
Planning/period2016/Q1 - 2016/Q4
Budget/Funding€ 250K
Actual status milestone_status_select_4
DeliverablesTrained & benchmarked Semantic POS-tagger
ApplicationASTRID-system training, Investor demos
Phase 4: Common Sense Knowledge framework

To be able to 'understand' the world, the ASTRID-system needs an internal representation of the world, known as 'Common Sense Knowledge'. This framework implements the ability to infer semantic relations from the training data and autonomously build and maintain a complex dataset that describes the world in intrinsic semantic detail. The ASTRID-system is now able to read documents to infer semantic relations and build an internal semantic reference of our reality.

Related news:

Meta
Planning/period2017/Q1 - 2017/Q4
Budget/Funding€ 200K
Actual status milestone_status_select_4
DeliverablesSemantic inference system, Semantic knowledge representation
ApplicationAutonomous semantic inference, Investor demos
Phase 5: Emotion system

The ASTRID-system implements 'Emotion based Cognition'. For this we need to implement an 'emotion layer' in the system that will be used by the Cognition Engine to be able to do emotion based inference.

At the current status, the emotion system is mostly implemented and is currently in active testing.

Meta
Planning/period2018/Q1 - 2018/Q2
Budget/Funding€ 100K - 200K
Actual status milestone_status_select_3
DeliverablesEmotions can be visualized in the system
ApplicationEmotion based Cognition, Investor demos
Phase 6: System training & behavioral research

Early on in the development, we will start with setting up training scenarios for the actual training of the system. Our AI is a supervised/autonomous learning system that uses real-world interaction to build its experience and perception of the world around it. We will employ additional specialists and (probably) academic students for this phase.

Meta
Planning/period2018/Q1 - 2018/Q4
Budget/Funding€ 200K - 300K
Actual status milestone_status_select_1
DeliverablesTraining sets, trained system
ApplicationWinnograd Schemas, Turing Test
Phase 7: Cognition Engine

The Cognition Engine will implement the capabilities to infer states and define actions, based on current available information and the internal (emotional) states of the ASTRID-system.

At the current status, the cognition system is largely engineered and ready for implementation.

Meta
Planning/period2018/Q3 - 2018/Q4
Budget/Funding€ 250K - 500K
Actual status milestone_status_select_1
DeliverablesASTRID-system fully operational
ApplicationHumanoid robotics, Healthcare, Support systems, etc.
©2010-2015 MIND|CONSTRUCT - All rights reserved @Google+