ALMA: Human Centric Algebraic Machine Learning

CORDIS fornisce collegamenti ai risultati finali pubblici e alle pubblicazioni dei progetti ORIZZONTE.

I link ai risultati e alle pubblicazioni dei progetti del 7° PQ, così come i link ad alcuni tipi di risultati specifici come dataset e software, sono recuperati dinamicamente da .OpenAIRE .

Risultati finali

Open source experimentation tools - Initial

Tools and example code repositories will be made available to the general public.

AML-IP implementation - Initial

An AMPIP implementation prototype will be released as open source Also an implementation report which will describe challenges alternatives directions and results will be periodically updated The implementation delivered will contain the state of the different parts of AMLIP that belongs to tasks 61 62 and 63

Working Prototype and designs - Revised

We will develop a working prototype in conjunction with creative professionals that demonstrates how human users can successfully interact with and control the AML.

AML-IP implementation - Final

An AMP-IP implementation prototype will be released as open source. Also, an implementation report which will describe challenges, alternatives, directions and results will be periodically updated. The implementation delivered will contain the state of the different parts of AML-IP, that belongs to tasks 6.1, 6.2 and 6.3.

Intelligent Tools for supporting creative professionals - Initial

Prototype of the tools designed and tested with professional designers.

AML accelerator prototype implementation - Initial

Research and development on high performance AML engines.

AML accelerator prototype implementation - Final

Research and development on high performance AML engines.

Intelligent Tools for supporting creative professionals - Final

Prototype of the tools designed and tested with professional designers.

AML-DL Interpreter - Revised M36

A Python or C program to translate AML-DL into algebraic relations compatible with AML. Documentation, user manuals, and installation guides will be written as well. Released and revised in three different stages. Initial internal release for consortia at M12.

AML-DL Interpreter - Final

AML-IP implementation - Revised M36

AML-DL Interpreter - Initial

A Python or C program to translate AMLDL into algebraic relations compatible with AML Documentation user manuals and installation guides will be written as well Released and revised in three different stages Initial internal release for consortia at M12

AML-DL Consistency checker - Final

A C program that can read interpreted code from the AML-DL interpreter and determine its logical consistency.

AML-DL Debugger - Initial

A software tool to aid AML-DL developers with problems related to logical consistency and compatibility of AML-DL code, including documentation, user manuals, installation guides, etc.

Design of the benchmark tested for robotized ironing and garment folding - Final

Testbed prototype to analyse and compare the results of previous learning algorithms to the human centric AML based approaches.

AML-DL Consistency checker - Initial M9

A C program that can read interpreted code from the AMLDL interpreter and determine its logical consistency

Open source experimentation tools - Final

Tools and example code repositories will be made available to the general public.

AML-IP implementation - Revised M18

Working Prototype and designs - Initial

We will develop a working prototype in conjunction with creative professionals that demonstrates how human users can successfully interact with and control the AML

Working Prototype and designs - Final

We will develop a working prototype in conjunction with creative professionals that demonstrates how human users can successfully interact with and control the AML.

AML-DL Debugger - Final

A software tool to aid AML-DL developers with problems related to logical consistency and compatibility of AML-DL code, including documentation, user manuals, installation guides, etc.

Design of the benchmark tested for robotized ironing and garment folding - Initial

Testbed prototype to analyse and compare the results of previous learning algorithms to the human centric AML based approaches.

Overall Architecture Definition - Initial M9

Specifications for the AMLDL and AML Integrating Platform released and revised in four different stages

Overall Architecture Definition - Revised M36

Specifications for the AMLDL and AML Integrating Platform released and revised in four different stages

AML based world models - Final

We will demonstrate incrementally complex world models generated using AML and human input based T 54 55 and 56 code and evaluation report

AML complete Foundations

We will send for publication to a ML journal numerical results for many algebras interacting with humans using the results from Tasks 23 and 24

AML-DL Specification - Revised M36

A language specification document for a highlevel language interpretable to algebraic relations compatible with AML The document shall include some basic examplesReleased and revised in four different stages

Overall Architecture Definition - Revised M18

Specifications for the AMLDL and AML Integrating Platform released and revised in four different stages

Exploitation and Outlook Plan - M36

A document with the exploitation and outlook plan together with the revision of actions and priorities regarding communication dissemination and collaboration activities

Image Classification use case Results using AML & Human Interaction - Final

Report of the use case results.

Report on Collaboration - M18

Report describing main activities performed regarding collaboration

Human-AML Interaction - Final

We will provide a state-of-the-art review of the literature incorporating research from the cognitive and social sciences as well as research in information visualization.

Sensor based context and activity recognition methods based on AML - Revised

We will deliver incrementally complex generations of methods for using AML to perform sensor based activity recognition based on the work of T 5152 and 53 code and evaluation report

Evaluation - Initial

We will run a controlled laboratory experiment to determine the effectiveness of the different visualization and interaction techniques with a separate group of users. We will also develop a tutorial that describes our approach, for use by other AI researchers who want to adopt this form of human-computer partnership with shared agency.

AML-DL Specification - Initial M9

Embedding of ethics and cultural concerns into AML based world models - Final

We will report the results of our work on including using AML and the world models from D5.2 to teach Ai systems about culture and ethics (code and evaluation report results from T5.7 and 5.8)

AML Theorems

We will send to the arxiv a technical report including theorems showing which properties learning algebras need for generalisation corresponding to Task 21

AML-DL Specification - Final

A language specification document for a high-level language interpretable to algebraic relations compatible with AML. The document shall include some basic examples.Released and revised in four different stages.

Embedding of ethics and cultural concerns into AML based world models - Initial

We will report the results of our work on including using AML and the world models from D52 to teach Ai systems about culture and ethics code and evaluation report results from T57 and 58

Report on Collaboration - M36

Report describing main activities performed regarding collaboration

Human-AML Interaction - Initial

We will provide a stateoftheart review of the literature incorporating research from the cognitive and social sciences as well as research in information visualization

Report on Communication and Dissemination - M18

Report describing main activities performed regarding communication and dissemination

Exploitation and Outlook Plan - M18

A document with the exploitation and outlook plan together with the revision of actions and priorities regarding communication dissemination and collaboration activities

Evaluation - Final

Analysis of algorithmic characteristics and hardware architecture templates - Initial

Research and development about the algorithmic characteristics and hardware architectures

AML complete Foundations - Revised

We will send for publication to a ML journal numerical results for many algebras interacting with humans using the results from Tasks 2.3 and 2.4.

AML based world models - Initial

We will demonstrate incrementally complex world models generated using AML and human input based T 54 55 and 56 code and evaluation report

AML-DL Specification - Revised M18

Image Classification use case Results using AML & Human Interaction - Initial

Report of the use case results

Overall Architecture Definition - Final

Specifications for the AML-DL and AML Integrating Platform, released and revised in four different stages.

Analysis of algorithmic characteristics and hardware architecture templates - Final

Research and development about the algorithmic characteristics and hardware architectures

Sensor based context and activity recognition methods based on AML - Initial

We will deliver incrementally complex generations of methods for using AML to perform sensor based activity recognition based on the work of T 5152 and 53 code and evaluation report

Interaction Paradigm Methodology - Initial

We will extend current participatory design and generative design methodologies to incorporate the role of an intelligent system in the interaction loop where agency is revealed and shared by the human user and the AML system We will also provide new visualization methods that help novice users understand the state of the algorithm in terms they can understand

Interaction Paradigm Methodology - Revised

Interaction Paradigm Methodology - Final

We will extend current participatory design and generative design methodologies to incorporate the role of an intelligent system in the interaction loop, where agency is revealed and shared by the human user and the AML system. We will also provide new visualization methods that help novice users understand the state of the algorithm in terms they can understand.

Sensor based context and activity recognition methods based on AML - Final

We will deliver incrementally complex generations of methods for using AML to perform sensor based activity recognition based on the work of T 5152 and 53 code and evaluation report

AML based world models - Revised

We will demonstrate incrementally complex world models generated using AML and human input based T 54 55 and 56 code and evaluation report

AML Theorems II

We will send for publication to a ML journal our theorems showing which properties learning algebras need to have for generalisation extending Deliverable 21a and including Theorems we can derive also from Task 22 The goal is to have the publication at M30 but we will submit the latest revision of the publication at M36 to align the deliverable with the periodic report

Exploitation and Outlook Plan - Final

A document with the exploitation and outlook plan, together with the revision of actions and priorities regarding communication, dissemination and collaboration activities.

Human-AML Interaction - Revised

We will provide a stateoftheart review of the literature incorporating research from the cognitive and social sciences as well as research in information visualization

Report on Collaboration - Final

Report describing main activities performed regarding collaboration.

Report on Communication and Dissemination - M36

Report describing main activities performed regarding communication and dissemination

Report on Communication and Dissemination - Final

Report describing main activities performed regarding communication and dissemination.

Data Management Plan - Initial

Data Management Plan specifying what data will be open detailing what data the project will generate whether and how it will be exploited or made accessible for verification and reuse and how it will be curated and preservedALMA project will produce data from experiments benchmarking of the AML engine on different scenarios WP3 along with data relative to the different use cases WP7 Also the performance of the different system tools could produce some reportable data WP6 It will also be updated to reflect the most recent state of the project as the project progresses

Data Management Plan - Revised

Data Management Plan specifying what data will be open, detailing what data the project will generate, whether and how it will be exploited or made accessible for verification and re-use, and how it will be curated and preserved.ALMA project will produce data from experiments & benchmarking of the AML engine on different scenarios (WP3), along with data relative to the different use cases (WP7). Also the performance of the different system tools could produce some reportable data (WP6) It will also be updated to reflect the most recent state of the project as the project progresses.

Pubblicazioni

Transferring human emotions to robot motions using Neural Policy Style Transfer

Autori: Raul Fernandez-Fernandez; Bartek Łukawski; Juan G. Victores; Claudio Pacchierotti
Pubblicato in: Cognitive Systems Research, Numero 30, 2023, ISSN 1389-0417
Editore: Elsevier BV
DOI: 10.1016/j.cogsys.2023.05.010

Practical implications from distinguishing between Pearl blankets and Friston blankets

Autori: Stephen Fox
Pubblicato in: Behavioral and Brain Sciences, Numero 45, E194., 2022, ISSN 2076-328X
Editore: Behavioral and Brain Sciences

Human–Artificial Intelligence Systems: How Human Survival First Principles Influence Machine Learning World Models

Autori: Stephen Fox
Pubblicato in: Systems, Numero 10(6), 260., 2022, ISSN 2079-8954
Editore: MDPI
DOI: 10.3390/systems10060260

Representing Human Ethical Requirements in Hybrid Machine Learning Models: Technical Opportunities and Fundamental Challenges

Autori: Stephen Fox, Vitor Fortes Rey
Pubblicato in: Machine Learning and Knowledge Extraction, Numero 6, 2024, Pagina/e 580-592, ISSN 2504-4990
Editore: MDPI
DOI: 10.3390/make6010027

Psychomotor Predictive Processing

Autori: Stephen Fox
Pubblicato in: Entropy, Numero 23(7), 806, 2021, Pagina/e 1-19, ISSN 2504-3900
Editore: MDPI
DOI: 10.3390/e23070806

Safety of Human–Artificial Intelligence Systems: Applying Safety Science to Analyze Loopholes in Interactions between Human Organizations, Artificial Intelligence, and Individual People

Autori: Stephen Fox, Juan G. Victores
Pubblicato in: Informatics, Numero 11, 2025, Pagina/e 36, ISSN 2227-9709
Editore: MDPI
DOI: 10.3390/informatics11020036

Infinite Atomized Semilattices

Autori: Fernando Martin Maroto, Antonio Ricciardo, David Méndez and Gonzalo G. de Polavieja
Pubblicato in: arXiv (on peer review), 2023, ISSN 2331-8422
Editore: arXiv
DOI: 10.48550/arxiv.2311.01389

Deep Robot Sketching: An application of Deep Q-Learning Networks for human-like sketching,

Autori: Raul Fernandez-Fernandez, Juan G. Victores, Carlos Balaguer
Pubblicato in: Cognitive Systems Research, Numero Volume 81, 2023, Pagina/e Pages 57-63, ISSN 1389-0417
Editore: Elsevier BV
DOI: 10.1016/j.cogsys.2023.05.004

Better than Trees: Applying Semilattices to Balance the Accuracy and Complexity of Machine Learning Models

Autori: Stephen Fox, Antonio Ricciardo
Pubblicato in: Machine Learning and Knowledge Extraction, Numero 7, 2025, Pagina/e 5, ISSN 2504-4990
Editore: MDPI
DOI: 10.3390/make7010005

A Cognitive Load Theory (CLT) Analysis of Machine Learning Explainability, Transparency, Interpretability, and Shared Interpretability

Autori: Stephen Fox, Vitor Fortes Rey
Pubblicato in: Machine Learning and Knowledge Extraction, Numero 6, 2024, Pagina/e 1494-1509, ISSN 2504-4990
Editore: MDPI
DOI: 10.3390/make6030071

Behavioral Ethics Ecologies of Human-Artificial Intelligence Systems

Autori: Stephen Fox
Pubblicato in: Behavioral Sciences, Numero 12(4), 103, 2022, Pagina/e 1-18, ISSN 2076-328X
Editore: MDPI
DOI: 10.3390/bs12040103

Finite Atomized semilattices

Autori: Fernando MartinMaroto, Gonzalo G. de Polavieja.
Pubblicato in: arXiv, 2021, ISSN 2331-8422
Editore: arXiv
DOI: 10.48550/arxiv.2102.08050

Semantic Embeddings in Semilattices

Autori: Fernando Martin-Maroto, Gonzalo G. de Polavieja
Pubblicato in: arXiv, 2022, Pagina/e 58, ISSN 2331-8422
Editore: arXiv
DOI: 10.48550/arxiv.2205.12618

algebraic machine learning: learning by the extraction of stable components in the freest model

Autori: Fernando Martin Maroto, Gonzalo G. de Polavieja.
Pubblicato in: arXiv, 2022, ISSN 2331-8422
Editore: ArXiv

Extending Piecewise Bezier Fitting Methods ́to SO(3) and SE(3) for Robot Paths

Autori: Montesino Valle, Ignacio Victores, Juan G. Balaguer, Carlos Jardón Huete, Alberto
Pubblicato in: FW12: Geometric Representations: The Roles of Screw Theory, Lie algebra, & Geometric Algebra, Numero Volume 1, 2023
Editore: IEEE International Conference on Robotics and Automation 2023

Impedance based Gym environment for the IIWA Collaborative robot towards human robot interaction

Autori: Montesino Valle, Ignacio; Łukawski, Bartek; Victores, Juan G.; Jardón Huete, Alberto; Balaguer, Carlos
Pubblicato in: reponame:RUC. Repositorio da Universidade da Coruña, Numero 2, 2022, Pagina/e 762-769, ISBN 978-84-9749-841-8
Editore: Universidade da Coruña. Servizo de Publicacións
DOI: 10.17979/spudc.9788497498418.0762

Combining Human-centred Explainability and Explainable AI.

Autori: Koch, J., & Fortes Rey, V.
Pubblicato in: 2022
Editore: ACM CHI Workshop Human-Centered Perspectives in Explainable AI

Don't freeze: Finetune encoders for better Self-Supervised HAR

Autori: Vitor Fortes Rey, Dominique Nshimyimana, and Paul Lukowicz
Pubblicato in: Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing (UbiComp/ISWC '23 Adjunct), 2023, ISBN 979-8-4007-0200-6
Editore: ACM
DOI: 10.1145/3594739.3610790

Manipulación de objetos dirigida a la asistencia de personas con movilidad reducida

Autori: """Naranjo-Campos, Francisco J. Matías-Martínez, Ainhoa de Victores, Juan G. Álvarez, Nicolás Alcaide, Almudena Balaguer, Carlos"""
Pubblicato in: XLIII Jornadas de Automática: libro de actas, Numero Volume 1, 2022, ISBN 978-84-9749-841-8
Editore: Universidade da Coruña. Servizo de Publicacións
DOI: 10.17979/spudc.9788497498418.0798

An inverse kinematics problem solver based on screw theory for manipulator arms

Autori: Łukawski, Bartek; Montesino Valle, Ignacio; Victores, Juan G.; Jardón Huete, Alberto; Balaguer, Carlos
Pubblicato in: XLIII Jornadas de Automática: libro de actas, Numero 1, 2022, Pagina/e 864-869, ISBN 978-84-9749-841-8
Editore: Universidade da Coruña. Servizo de Publicacións
DOI: 10.17979/spudc.9788497498418.0864

A generic controller for teleoperation on robotic manipulators using low-cost devices

Autori: Bartek Łukawski, Juan G. Victores, Carlos Balaguer
Pubblicato in: XLIV Jornadas de Automática: libro de actas: Universidad de Zaragoza, Escuela de Ingeniería y Arquitectura, 6, 7 y 8 de septiembre de 2023, Zaragoza, 2023, Pagina/e 785-788
Editore: Servizo de Publicacións. Universidade da Coruña
DOI: 10.17979/spudc.9788497498609.785

È in corso la ricerca di dati su OpenAIRE...

Risultati finali

Pubblicazioni

Condividi questa pagina Condividi questa pagina sui social network

Scarica Scarica il contenuto della pagina